Mercator Ocean Journal 57 by Copernicus Marine_Service

#57 SPECIAL ISSUE

COPERNICUS MARINE SERVICE 2015-2021 ACHIEVEMENTS

Implemented by

THE COPERNICUS MARINE SERVICE FROM 2015 TO 2021: SIX YEARS OF ACHIEVEMENTS BY Le Traon, P.Y. , Abadie, V. , Ali, A. , Aouf, L. , Artioli, Y.5, Ascione, I.6, Autret, E.4, Aydogdu, A.7, Aznar, R.8, Bahurel, P.1, Barrera, E.9, Bastide, L.1, Bazin, D.1, Behrens, A.10, Bentamy, A.4, Bertino, L.11, Böhm, E.12, Bonofiglio, L.13, Bourdallé Badie, R.1, Brando, V.E.12, Brandt-Kreiner, M.14, Bricaud, C.1, Bruciaferri, D.6, Buongiorno Nardelli, B.12, Buus-Hinkler, J.14, Cailleau, S.1, Calton, B.5, Carrasco, A.2, Carval, T.4, Cesarini, C.12, Chabot, G.1, Charles, E.15, Chau, T.16, Ciliberti, S.7, Cipollone, A.7, Claustre, H.18, Clementi, E.7, Colella, S.12, Colombo, F.13, Coppini, G.7, Cossarini, G.17, Crosnier, L.1, D’Alimonte, D.19, Dabrowski, T.20, De Alfonso, M.21, de Nucé, A.1, Delamarche, A.1, Derval, C.1, Di Cicco, A.12, Dibarboure, G.22, Dinessen, F.2, Dodet, G.4, Drevillon, M.1, Drillet, Y.1, Drudi, M.7, Durand, E.1, Escudier, R.7, Etienne, H.15, Fabardines, M.1, Faugère, Y.15, Fleming, A.23, Forneris, V.12, Garcia-Hermosa, I.1, Sotillo, M.G21, Garnesson, P.24, Garric, G.1, Gasciarino, G.1, Gehlen, M.16, Giesen, R.25, Giordan, C.1, Girard-Ardhuin, F.4, Golbeck, I.26, Good, S.A.6, Gourrion, J.27, Gregoire, M.28, Greiner, E.15, Guinehut, S.15, Gutknecht, E.1, Harris, C.6, Hernandez, F.29, Hieronymi, M.10, Høyer, J.14, Huess, V.14, Husson, R.15, Jandt-Scheelke, S.26, Jansen, E.7, Kärnä, T.30, Karvonen J.30, Kay, S.6, King, R.R.6, Korres, G.31, Krasemann, H.10, Labrousse, C.1, Lagemaa, P.32, Lamouroux, J.1, Law Chune, S.1, Lecci, R.7, Legros, V.1, Lellouche, J.M.1, Levier, B.1, Li, X.26, Lien, V.S.33, Lima, L.7, Lindenthal, A.26, Linders, J.34, Lorente, P.8, Lorkowski, I.26, Mader, J.35, Maksymczuk, J.6, Maljutenko, I.32, Mangin, A.24, Marinova, V.36, Masina, S.7, Matreata, M.37, McConnell, N.6, Melet, A.1, Melsom, A.2, Messal, F.1, Morrow, R.38, Mouche, A.4, Mulet, S.15, Netting, J.5, Nord, A.34, Novellino, A.13, Obaton, D.1, Palazov, A.39, Pascual, A.40, Payet, J.M.3, Peneva, E.41, Pequignet, A.C.6, Perivolis, L.31, Perruche, C.1, Pfeil, B.42, Piollé, J.F.4, Pisano, A.12, Pistoia, J.7, Polton, J.A.43, Pouliquen, S.4, Pujol, M.I.15, Quade, G.1, Quéau, A.1, Ravdas, M.31, Reffray, G.1, Régnier, C.1, Renshaw, R.6, Reppucci, A.1, Rotlan, P.44, Ruggiero, G.1, Saldo, R.45, Salon, S.17, Samson, G.1, Santoleri, R.12, Saulter, A.6, Saux-Picart, E.3, Sauzède, R.18, Schwichtenberg, F.26, She, J.14, Siddorn, J.6, Skakala, J.5, Staneva, J.10, Stelzer, K.46, Stoffelen, A.25, Sykes, P.6, Szczypta, C.47, Szekely, T.27, Tamm, S.26, Tan, T.A.48, Tarot, S.4, Teruzzi, A.17, Thomas-Courcoux, C.1, Tintore, J.44, Titaud, O.15, Tonani, M.6, Tronci, M.1, Tuomi, L.30, Van der Zande, D.49, Vandenbulcke, L.28, Verbrugge, N.15, Volpe, G.12, von Schuckmann, K.1, Wakelin, S.L.43, Wedhe, H.33, Zacharioudaki, A.31, Zuo, H.50 1

Mercator Ocean international, France. 2 MET No, Norway. 3 MeteoFrance, France. 4 IFREMER, France. 5 PML, United Kingdom. 6 MetOffice, United Kingdom. 7 CMCC, Italy. 8 Nologin Consulting. 9 AEMET, Spain. 10 Hereon, Germany. 11 NERSC, Norway. 12 CNR ISMAR, Italy. 13 ETT S.p.A., Italy. 14 DMI, Denmark. 15 CLS, France. 16 LSCE, France. 17 OGS, Italy. 1

LOV-IMEV, France. AEQUORA, Portugal. 20 Marine Institute, Ireland. 21 Puertos del Estado, Spain. 22 CNES, France. 23 BAS, United Kingdom. 24 ACRI, France. 25 KNMI, The Netherlands. 26 BSH, Germany. 27 OceanScope, France. 28 University of Liege, Belgium. 29 IRD, France. 30 FMI, Finland. 31 HCMR, Greece. 32 TalTech, Estonia. 33 IMR, Norway. 34 SHMI, Sweden. 35 AZTI, Spain.

IOBAS, Bulgaria. NIHWM, Romania. 38 LEGOS, France. 39 Bulgarian Academy of Science, Bulgaria. 40 IMEDEA, Spain. 41 Sofia University, Bulgaria. 42 UiB, Norway. 43 NOC, United Kingdom. 44 SOCIB, Spain. 45 DTU, Denmark. 46 Brockmann Consult, Germany. 47 CELAD, France. 48 SCALIAN, France. 49 RBINS, Belgium. 50 ECMWF.

Please cite as: Le Traon, P.Y., Abadie, V., Ali, A., Aouf, L., Ardhuin, F., Artioli, Y., Ascione, I., Autret, E., Aydogdu, A., Aznar, R., Bahurel, P., Barrera, E., Bastide, L., Bazin, D., Behrens, A., Bentamy, A., Bertino, L., Böhm, E., Bonofiglio, L., Bourdallé Badie, R., Bourdallé-Badie, R., Brando, V.E., Brandt-Kreiner, M., Bricaud, C., Bruciaferri, D., Buongiorno Nardelli, B., Buus-Hinkler, J., Cailleau, S., Calton, B., Carrasco, A., Carval, T., Cesarini, C., Chabot, G., Charles, E., Chau, T., Ciliberti, S., Cipollone, A., Claustre, H., Clementi, E., Colella, S., Colombo, F., Coppini, G., Cossarini, G., Crosnier, L., D’Alimonte, D., Dabrowski, T., De Alfonso, M., de Nucé, A., Delamarche, A., Derval, C., Di Cicco, A., Dibarboure, G., Dinessen, F., Dodet, G., Drevillon, M., Drillet, Y., Drudi, M., Durand, E., Escudier, R., Etienne, H., Fabardines, M., Faugère, Y., Fleming, A., Forneris, V., Garcia-Hermosa, I., Sotillo, M.G, Garnesson, P., Garric, G., Gasciarino, G., Gehlen, M., Giesen, R., Giordan, C., Golbeck, I., Good, S.A., Gourrion, J., Gregoire, M., Greiner, E., Guinehut, S., Gutknecht, E., Harris, C., Hernandez, F., Hieronymi, M., Høyer, J., Huess, V., Husson, R., Jandt-Scheelke, S., Jansen, E., Kärnä, T., Karvonen, J., Kay, S., King, R.R., Korres, G., Krasemann, H., Labrousse, C., Lagemaa, P., Lamouroux, J., Law Chune, S., Lecci, R., Legros, V., Lellouche, J.M., Levier, B., Li, X., Lien, V.S., Lima, L., Lindenthal, A., Linders, J., Lorente, P., Lorkowski, I., Mader, J., Maksymczuk, J., Maljutenko, I., Mangin, A., Marinova, V., Masina, S., Matreata, M., McConnell, N., Melet, A., Melsom, A., Messal, F., Morrow, R., Mouche, A., Mulet, S., Netting, J., Nord, A., Novellino, A., Obaton, D., Palazov, A., Pascual, A., Payet, J.M., Peneva, E., Pequignet, A.C., Perivolis, L., Perruche, C., Pfeil, B., Piollé, J.F., Pisano, A., Pistoia, J., Polton, J.A., Pouliquen, S., Pujol, M.I., Quade, G., Quéau, A., Ravdas, M., Reffray, G., Régnier, C., Renshaw, R., Reppucci, A., Rotlan, P., Ruggiero, G., Saldo, R., Salon, S., Samson, G., Santoleri, R., Saulter, A., Saux-Picart, E., Sauzède, R., Schwichtenberg, F., She, J., Siddorn, J., Skakala, J., Staneva, J., Stelzer, K., Stoffelen, A., Sykes, P., Szczypta, C., Szekely, T., Tamm, S., Tan, T.A., Tarot, S., Teruzzi, A., Thomas-Courcoux, C., Tintore, J., Titaud, O., Tonani, M., Tronci, M., Tuomi, L., Van der Zande, D., Vandenbulcke, L., Verbrugge, N., Volpe, G., von Schuckmann, K., Wakelin, S.L., Wedhe, H., Zacharioudaki, A., Zuo, H., .2021. The Copernicus Marine Service from 2015 to 2021: six years of achievements Special Issue Mercator Océan Journal #57. https://doi.org/10.48670/moi-cafr-n813

#57

TABLE OF CONTENTS P.5

P.108

INTRODUCTION

PRODUCTION CENTERS - MONITORING AND FORECASTING CENTERS

P.12

GLOBAL HIGH RESOLUTION......................................................... 108 GLOBAL COUPLED.......................................................................... 118 ARCTIC..............................................................................................124 BALTIC.............................................................................................. 130 BLACK SEA...................................................................................... 137 IBERIA BISCAY IRELAND............................................................... 147 MEDITERRANEAN.......................................................................... 155 NORTH WEST EUROPEAN SHELF................................................ 164

PRODUCTION COORDINATION ACTIVITIES SERVICE EVOLUTION.................................................................. 12 PRODUCT QUALITY..................................................................... 19 UPSTREAM OBSERVATION INFRASTRUCTURE...................... 27 MULTI-YEAR PRODUCTS............................................................ 32 OCEAN REPORTING.................................................................... 37

P.173

P.44

INFORMATION SYSTEM & DATA MANAGEMENT CENTRAL INFORMATION SYSTEM.............................................. 173 DISSEMINATION UNIT................................................................... 180

PRODUCTION CENTERS THEMATIC ASSEMBLY CENTERS SEA LEVEL.................................................................................... 44 IN-SITU......................................................................................... 53 OCEAN COLOUR........................................................................... 62 SEA SURFACE TEMPERATURE.................................................. 73 SEA ICE......................................................................................... 82 WAVE............................................................................................. 90 WIND............................................................................................. 95 MULTI-OBS ................................................................................ 100

P.187 USER SERVICES AND INTERACTIONS SERVICE MONITORING.................................................................. 187 TRAINING AND USER SUPPORT.................................................. 192 USER ENGAGEMENT/MARKETS................................................. 200 USER UPTAKE................................................................................ 207 COMMUNICATION AND OCEAN LITERACY................................ 215

THE COPERNICUS MARINE SERVICE FROM 2015 TO 2021 : OVERVIEW AND INTRODUCTION TO THE MOI JOURNAL SPECIAL ISSUE LE TRAON, P.Y., BAHUREL, P., CROSNIER, L., DELAMARCHE, A., DREVILLON, M.,

DRILLET, Y., DURAND E., FABARDINES, M., GARCIA-HERMOSA, I., MELET, A., REPPUCCI, A., SOTILLO, M.G, VON SCHUCKMANN, K.

This introduction paper provides an overview of the Copernicus Marine Service implementation, architecture and organization as well as an overall assessment of achievements during the 2015-2021 period. The components of the Copernicus Marine Service are: - Thematic Assembly Centers (TACs), - Monitoring and Forecasting Centers (MFCs), - Central Information System (CIS) and Dissemination Unit (DU), - User Service.

six years of operations, the Copernicus Marine Service is recognized internationally as one of the most advanced service capacity in ocean monitoring and forecasting. So far, it has convinced more than thirty thousand expert services and users worldwide. The Copernicus Marine Service provides regular and systematic reference information on the physical (blue) and biogeochemical (green) ocean and sea-ice (white) state for the European regional seas and the global ocean (Figure 1). This capacity encompasses the description of the current situation (analysis), the prediction of the situation 10 days ahead (forecast), and the provision of consistent retrospective data records (reprocessing for in situ and satellite measurements, and reanalysis data). The development of the Copernicus Marine Service has required collaboration and innovation across research and technology, in observations, modelling, assimilation, reporting, and product and service delivery. This European Union service is unique in the world for: - its coverage and comprehensiveness, - its balance between state-of-the-art science and operational commitments, - the consistency of its portfolio where satellite observations, in situ observations, and 3D model simulations are proposed in a coherent way to describe the ocean.

These components, along with cross-cutting coordination activities (dedicated to scientific evolution, product quality, multi-year processing, ocean reporting and user uptake) are detailed in the different papers of this special issue.

INTRODUCTION The Copernicus Marine Service is one of the six pillar services of the European Union’s Copernicus programme. Mercator Ocean International (MOi) was entrusted at the end of 2014 by the European Union to implement the operational phase of the service from 2015 to 2021. After

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: The European Union Copernicus Marine Service.

The Copernicus Marine Service mission includes: - providing forecasts and outlooks on marine conditions to ensure security and safety for those operating at sea and for warnings of and/or rapid responses to extreme or hazardous events, - providing detailed descriptions of the ocean state to better understand and predict evolutions of weather and climate from regional to global scales, - generating information and services to support sustainable development of the ocean and coastal regions and their food and energy resources, - monitoring and reporting on past and present marine environmental conditions, in particular, the ocean response to climate change and other stressors and providing services that support a clean, healthy and resilient ocean, - connecting users with marine experts, through tailored information and reports, workshops and training sessions, operational and customized service support including the provision of scientific quality information.

1. ORGANISATION AND ARCHITECTURE OF THE COPERNICUS MARINE SERVICE 1.1 Organisation Mercator Ocean International (MOi) set up an organisation to manage operations and regular evolutions of the Copernicus Marine Service (Figure 2) system, its main sub systems (TACs, MFCs, CIS, DU) and their interfaces. The organization also includes essential activities related to user service, user requirement gathering and analysis and user engagement (Delamarche et al., this issue; Giordan et al., this issue; Abadie et al., this issue).

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 2: Copernicus Marine Service Organisation.

MOi has managed system and service evolution through a formal development process (specification, design, acceptance, entry into service) with a review process at the end of each development phase. This has led to regular releases of both system and catalogue (25 catalogue releases were issued from 2015 to 2021).

The main objective was to ensure a consistent approach and cross-fertilisation between different sub-systems (TACs and MFCs). An ad hoc coordination group was set up to organize the interfaces between the TACs and MFCs and the central DU. The Copernicus Marine Service is highly dependent on the timely availability of comprehensive satellite and in-situ observations (Le Traon et al., 2019). Change and transformation in the service are inextricably linked to the supply of upstream data and the proper specification of requirements. To manage this core ependency, MOi established coordination mechanisms with the upstream satellite and in-situ coordination bodies (ESA and Eumetsat for space, EEA, EuroGOOS and EMODnet for in situ) (Reppucci and Le Traon, this issue).

Innovation and technology are critical for maintaining a user responsive and state-of-the-art Marine Service. Innovation includes all parts of the Copernicus Marine Service, from data processing and quality control, through modelling and data assimilation and service layers. Two innovation programmes have been set up to manage mid term (Tier 2) R&D service evolution and interfaces with H2020 R&D programmes (Melet and Le Traon, this issue) and to strenghten interfaces with downstream services, applications and users (Durand et al., this issue).

Interfaces with other Copernicus Services have also been managed to ensure their consistency and complementarity. A key objective is to address marine information needs from other services (e.g., climate, security, emergency) and to develop cross-service offers for thematic areas (e.g., development of the Copernicus Coastal roadmap).

MOi also coordinated cross-cutting activities related to product quality (Sotillo et al., this issue), multi-year products (Drevillon et al., this issue), ocean reporting (von Schuckmann et al., this issue) and biogeochemistry data assimilation.

#57

Two high-level panels were set up to advise MOi on the implementation of the Copernicus Marine Service: a Scientific and Technical Advisory Committee (STAC) and a Champion User Advisory Group (CUAG). The STAC has advised MOi on science and technology issues related to the Copernicus Marine Service delivery and evolution. The STAC contributed to the definition and regular updates of the Copernicus Marine R&D Service Evolution strategy and the management of the Service Evolution R&D programme (Melet and Le Traon, this issue). The CUAG has advised MOi on user requirements and service evolution solutions implemented to meet user needs (Delamarche et al., this issue).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

situ & satellite data to elaborate high-level products). These production centres gather observation data from in situ networks (e.g., GOOS, EuroGOOS) and from the Copernicus satellite component (through ESA and EUMETSAT). TACs generate validated data sets directly useable for assimilation in models (MFCs) and derive high-level products (i.e., gridded multi-sensor products) directly useable for downstream applications. - Seven MFCs, distributed geographically (Global Ocean, Arctic Ocean, Baltic Sea, North Atlantic North West European Shelf, North Atlantic Iberia-Biscay-Ireland area, Mediterranean Sea and Black Sea), that generate model-based products on the ocean physical state (including waves and sea ice) and biogeochemical characteristics, including near real-time analyses, forecasts, hindcasts and reanalyses. - A CIS, encompassing the management and organization of information (Gasciarino and Tan, this issue) and products and a cloud based Dissemination Unit (DU) (Forneris et al., this issue). A single catalogue (global and European coverage) is offered to users. The CIS enables searching, viewing, downloading products and system monitoring. A manned service desk provides a network of technical and marine experts to support users.

1.2 Architecture The backbone of the Copernicus Marine Service relies on a distributed architecture of production centres for observations (Thematic Assembly Centres – TACs), modelling/assimilation (Monitoring and Forecasting Centres – MFCs), a Central Information System (CIS) and a cloud Dissemination Unit (DU) (Figure 3); it includes: - Eight TACs with six satellite TACs organized by ocean variables (sea level, ocean colour, sea-surface temperature, sea-ice, winds and waves), one for in situ observations and a last one, multi-observation TAC (uses data driven techniques to merge different in

Figure 3: Copernicus Marine Service Architecture.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. PRODUCTS

3. USERS

Copernicus Marine Service products are based on stateof-the-art data processing and advanced modelling and data assimilation techniques. Product uncertainties are assessed through rigorous internationally recognized quality assessment methods (Sotillo et al., this issue). The Copernicus Marine Service currently provides about 200 different products for observations and model outputs (see catalogue) covering ocean physics (e.g., temperature, salinity, sea level, currents, waves), seaice (e.g., concentration, thickness, drift, icebergs) and biogeochemistry (chl-a, oxygen, pH, nutrients). Modelling and data assimilation products have a resolution of 1/12° for the global scale and from 1/24° to 1/72° for regional applications. All disseminated Copernicus Marine Service products are fully documented with a Product User Manuel (PUM) and a Quality Information Document (QUID).

Copernicus Marine is a user-driven and policy-driven service: users are explicitly and transparently involved in the service delivery definition (Delamarche et al., this issue). It responds to public and private user needs and policies related to all marine and maritime sectors: maritime safety, coastal environment monitoring, trade and marine navigation, fishery, aquaculture, marine renewable energy, marine conservation and biodiversity, ocean health, climate and climate adaptation, recreation, education, science and innovation (Figure 4). The Marine Service directly contributes to European Union marine and maritime-related policies and priorities such as the Integrated Maritime Policy, the Common Fishery Policies and the Marine Strategy Framework Directive. The Copernicus Marine Service provides a core and generic service targeting downstream service providers (intermediate users) and serving a wide range of users and applications. Details of Copernicus Marine Service benefit areas and a series of use cases (more than 200) are available on the Copernicus Marine Service website (see Abadie et al., this issue).

The Copernicus Marine Service produces Ocean Monitoring Indicators (OMIs) and publishes an annual Ocean State Report (OSR) (von Schuckmann et. al., this issue) and its summary for the scientific community, as well as for policy and decision-makers. The OMIs and OSRs rely on the unique capability and expertise that the Copernicus Marine Service gathers in Europe to analyse and interpret changes in the marine environment and develop sciencebased assessments of the state and health of the ocean.

The Copernicus Marine Service service desk regularly monitors the number and types of users, statistics on downloaded products and user satisfaction (Delamarche et al, this issue; Giordan et al., this issue). More than 30,000 users are now registered on the service and a steady increase has been observed over the past couple of years.

Figure 4: Copernicus Marine, application sectors and policies.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

4. MAIN ACHIEVEMENTS

5. POST 2021 PROSPECTS

The first operational phase 2015-2021 of the Copernicus Marine Service has successfully implemented a unique European Union ocean monitoring and forecasting service.

Over the past 6 years, the Copernicus Marine Service has demonstrated its capabilities to run a state-of-the-art EU operational marine service responsive to user needs and scientific/technological advances. The service relies on a strong network of skilled European ocean information producers and a unique pool of EU scientific experts to assess the state of the ocean. These are major assets for the start of a new phase of the Copernicus Marine Service.

Major advances have been achieved during the period 2015-2021. The offer for the blue, green and white ocean has been regularly improved (see TACs and MFCs papers in this issue) with: - new products and marine parameters (surface currents, waves, pH, CO2, icebergs), - higher resolution and representation of more dynamic processes, - improved product quality and product quality assessment, - more satellite data (Sentinels) used as upstream inputs and improved algorithms, - longer time series of reprocessed in situ and satellite data and ocean reanalyses, - ocean monitoring indicators and ocean state reports and new visualisation tools.

The need for a more responsible and sustainable management of the ocean, relying on comprehensive ocean observing, monitoring, forecasting and assessment activities, is the main driver for Copernicus Marine Service and its evolution. This demand has never been so prominent in international (e.g., UN SDGs and SDG 14, UN Decade of Ocean Science) and European (e.g., Green Deal) political agendas. MOi has developed an ambitious plan for the next 7 years (2021-2027). Designed in close interaction with the European Commission and member states and with the advice of its two committees (STAC and CUAG), it allows a step-by-step implementation depending on budget, user needs and priorities and feasibility/maturity (Figure 5).

The uptake of Sentinel-1 (S-1) (sea-ice coverage, ocean waves), Sentinel-3 (S-3) (altimetry and surface currents, sea-surface temperature, ocean colour) data and Sentinel-2 (turbidity, ocean colour) has, in particular, greatly improved Copernicus Marine Service offer.

The plan addresses main users and policy needs identified by the European Commission and MOi during Copernicus 1 (SWD, 2019). It identifies three levels of implementation for the evolution of the Copernicus Marine Service product and service portfolio over the period 2021-2027: baseline (continuity of service with incremental evolutions), enhanced continuity (major product improvements) and new services.

MFCs and TACs have been very robust since the start of the Copernicus Marine Service and have delivered an operational service. The same holds for the CIS & DU components even though the transition from a distributed DU to a cloud-centralized DU has been challenging due to the short transition period. The Copernicus Marine Service Ocean State Reports, its summary for policy makers and Ocean Monitoring Indicators (von Schuckmann et al., this issue) have provided a unique ocean monitoring dashboard for policy and decision makers as well as for the general public. They are now part of the EU ocean state assessment landscape and have federated a unique pool of EU scientific experts to assess the state of the ocean based on the Copernicus Marine Service ocean monitoring products.

Baseline will be implemented from the start of Copernicus 2 to ensure the continuity of present service and maintain a consistent blue, white and green offer. It includes incremental evolutions to improve product quality, integrate future Sentinel missions and new in-situ observations and leverage new capabilities of digital services through the WEkEO DIAS platform. User interaction and user engagement will be strengthened by developing dedicated sectorial offers per applications and policies. The objective is also to reinforce the Copernicus programmme consistency by producing marine data for other Copernicus services and developing sectorial approaches with the development of Copernicus Thematic Hubs, MOi being positioned to lead future Coastal and Arctic Thematic Hubs.

There have been major achievements of user uptake and user engagement over the past 5 years (Abadie et al., this issue; Quade et al., this issue; Durand et al., this issue). The user uptake has been progressing regularly reaching 30,000 subscribers at the end of Copernicus 1. User satisfaction has stayed high all along the 2015-2021 period illustrating the high level of service. Our service monitoring, user interaction and user engagement and training activities also improved a lot (Giordan et al., this issue).

The enhanced continuity and new services streams will build from present and future H2020 and Horizon Europe R&D projects and will be developed depending on budget and priorities. Improved digital services, ensemble forecasts, higher resolution, step change in Arctic monitoring, air/sea CO2 fluxes, 20th century reanalyses are

#57

proposed under the enhanced continuity scenario. Coastal ocean monitoring, marine biology, climate marine projection (coastal, ecosystem) are proposed under the new services scenario. A major priority is to offer new services for the coastal ocean through a co-design and co-development approach between the EU Copernicus Marine Service and coastal marine services operated by member states.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Challenging issues necessary to establish a comprehensive monitoring of the global ocean require international cooperation. The Copernicus Marine Service has set up important partnerships with GOOS and IOC, OceanPredict, GEO and GEO Blue Planet. The UN Decade of Ocean Science will be a unique opportunity to develop further the required international cooperation.

Figure 5: Evolution of Copernicus Marine Service in Copernicus 2.

REFERENCES Mercator Ocean (2021). CMEMS Marine Service Portofolio. Mercator Ocean (2021). CMEMS User Requirement Document. CMEMS (2017). CMEMS Requirements for the Evolution of the Copernicus Satellite Component. Mercator Ocean and CMEMS Partners.

Le Traon P.Y. et al., (2019). From Observation to Information and Users: The Copernicus Marine Service Perspective. Front. Mar. Sci. 6:234. doi: 10.3389/ fmars.2019.00234.

of the in-situ observing system. Mercator Ocean, CMEMS Partners and EuroGOOS. CMEMS STAC (2021). Copernicus Marine Environment Monitoring Service (CMEMS) Service Evolution Strategy: R&D priorities. Version 5.

Mercator Ocean (2016). CMEMS High Level Service Evolution Strategy. Document prepared with the support of the CMEMS Science and Technology Advisory Committee (STAC).

CMEMS (2021). CMEMS Requirements for the Evolution

Roadmap for the evolution of Copernicus marine and land services to better serve coastal user, December 2018. SWD (2019). Commission Staff Working Document, Expression of user needs for the Copernicus programme, 394 final, 25 October 2019.

THE COPERNICUS MARINE SERVICE EVOLUTION R&D PROGRAMME MELET, A., LE TRAON, P.Y. Mercator Ocean International

consolidate and to expand the catalogue (e.g., addition of new products) and to prepare the next generation of operational systems. R&D Service Evolution activities therefore correspond to a line of activity complementing the main operational activities. Associated research developments and activities are then transferred into operations in Copernicus Marine Service production centres.

ABSTRACT The Copernicus Marine Service is continuously evolving to better serve user and policy needs. To stay a world leading service and remain at the state-of-the-art, scientific and technical developments are required across different time scales. This paper presents the Copernicus Marine Service Evolution R&D programme principles and strategy, the associated roadmap and main achievements during the first implementation phase of Copernicus.

2. SERVICE EVOLUTION PRINCIPLES AND STRATEGY

1. INTRODUCTION

2.1 Principles and Drivers Three main principles are guiding the Copernicus Marine Service Evolution (SE). First, service evolutions are driven by user and policy needs (e.g., SWD 2019, URD 2019). Users from public and private marine-related sectors are explicitly and transparently involved in the service’s delivery definition and its required evolutions (Figure 1). The collection of user feedback and needs is described in Delamarche et al., this issue. In this bottom-up approach, user and policy needs are translated into requirements corresponding to achievable technical or scientific objectives for the Copernicus Marine Service.

After six years of operations, the Copernicus Marine Service is internationally recognized as one of the most advanced service capacities in ocean monitoring and forecasting, and has convinced more than thirty thousand expert-services and users worldwide (Le Traon et al., this issue). A strength of the Copernicus Marine Service is its dynamic nature. Its offer is continuously evolving to ensure that distributed products remain state-of-the-art and meet a wider range of existing and emerging user and policy needs (URD 2019 and MSP 2019). Over 2015-2020, a total of 85 new products, including the distribution of new requested parameters such as wind waves, iceberg density or micronekton and ocean acidity were added to the Copernicus Marine Service data catalogue.

Secondly, this user and policy push for service evolution is complemented by a science pull (Figure 1). In this topdown approach, scientific and technological advances relevant for the Copernicus Marine Service are analysed and R&D objectives are developed (see Section 2.2) so that the Service remains at the state-of-the-art.

This article focuses on the R&D aspects of the ‘Service Evolution’ of the Copernicus Marine Service. The main objective of the Service Evolution R&D programme is to improve its scientific content. In that regard, leading-edge R&D activities are required to consolidate scientific tools and methods used in production centres for delivering the best possible data on the ocean state. Better data to support new developments to

The third principle guiding the Service evolution deals with the delineation with downstream activities. The Copernicus Marine is a core service focusing on activities that are best performed at pan-European scale, supporting expert value-adding downstream services.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: The Copernicus Marine Service is user and policy driven. User feedbacks are translated into achievable scientific or technical evolutions for the Service.

2.2 Strategy

Eight strategic thrusts were identified: - upstream observations, - data and products, - information and services, - science and technology, - service architecture and operations, - private and public downstream stakeholders and users, - capability and performance.

The Service Evolution Strategy is maintained on the basis of direct feedback from users, scientific and technical gaps analysis of emerging and existing user requirements, and the potential to improve the Core Service elements. The Copernicus Marine Service evolution strategy is defined by Mercator Ocean International (MOi), with support from a high-level Scientific and Technical Advisory Committee (STAC). The STAC is composed of 13 independent and internationally recognized scientific and technical experts on various thematics of interest for the Copernicus Marine Service. STAC members work impartially and strict rules are applied to avoid any conflict of interest.

The second reference document details the “science and technology” evolution overall strategy and objectives. The Service Evolution: R&D priorities document (STAC 2018), aka the Copernicus Marine Service R&D roadmap, notably identifies required developments per R&D thematic areas that are spanning the different activities of the Service, gathered over five overarching themes: - ocean circulation, ocean-wave and ocean-ice coupling, - biogeochemistry and ecosystems, - seamless interactions with coastal systems, - ocean-Atmosphere coupling and climate, - cross-cutting developments on observation, assimilation and product quality improvements.

Two reference documents present the Service Evolution strategy; both were prepared with support from the STAC. The High-Level Service Evolution Strategy (Mercator Ocean, 2016) presents strategic thrusts/goals for an evolution of the full scope of the Service Evolution aligned with stakeholders and users expectations, together with their corresponding lines of action and associated benefits.

The Service R&D roadmap is updated on an annual basis and constitutes a key reference document for R&D activities performed externally to the production centres (e.g., Tier 2 and Tier 3 activities), presented in the next section.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

3.2 Tier 2: Mid-term R&D activities

3. SERVICE EVOLUTION ROADMAP

Mid-term R&D objectives (2-year cycle) are mainly addressed through Copernicus Marine Service calls for tenders. The goal of these calls for tenders is to ensure the scientific evolution of the Service and to improve the operational service within 3-4 years from the start of the projects (Figure 2). The STAC assists MOi in preparing service evolution calls for tenders and in the corresponding evaluation by external reviewers, selection and monitoring processes. Two calls for tenders were published over 2014-2021 for a total budget of 5 M€. Following the first open call for tenders, a batch of 12 R&D projects were funded for 2 years (April 2016-March 2018). The second call for tenders led to the funding of 18 2-year R&D projects (April 2018-March 2020).

To ensure a continuous scientific and technical evolution of the Copernicus Marine Service over time, R&D activities are broken down into different categories (Tier 1, Tier 2, Tier 3), with different associated time scales and actors (Figure 2).

3.1 Tier 3: Long-term R&D activities Long-term (>2 years and up to 10 years) evolution needs are identified in the Copernicus Marine service evolution R&D strategy based on an analysis by MOi and the STAC of R&D needs. Monitoring and Forecasting Centre (MFC) and Thematic Assembly Centre (TAC) 3-year and 6-year development plans are reviewed by MOi and the STAC and also contribute to the definition of the long-term service evolution strategy.

To ensure the research to operation transition, the Copernicus Marine Service funded Service Evolution R&D projects are closely monitored by MOi and the STAC. Coordination activities gathering the production centres and R&D project teams are organized twice during the lifetime of a project: at mid-term and at the end. After completion of projects, the uptake by production centres is also monitored by MOi especially during the formal development process for the evolution of production centres’ systems and products (e.g., Design and Specification Reviews of the production centres, see Le Traon et al., this issue).

Unlike short and mid-term R&D activities, long-term Service Evolution R&D activities are promoted in the framework of external projects, such as Horizon 2020 and other European and national R&D programmes (Figure 2). Of particular importance for the long-term evolution of Copernicus Marine Service are the EU H2020 Space Programme with Copernicus Evolution calls for tenders. MOi, as entrusted entity by the European Commission (EC) to implement the Copernicus Marine Service, provided the EC with guidance on the content of the call so that they address long-term R&D needs. MOi monitored several projects funded through H2020, some funded through the H2020 Copernicus Evolution programme (e.g., CEASELESS, IMMERSE, SEAMLESS, FORCOAST), other funded from different H2020 programmes (e.g., KEPLER, EUROSEA). MOi also organizes interfaces between these projects and the Copernicus Marine Service coordination teams, to assess the impact on operational production centres.

3.3 Tier 1: Short-term R&D activities Short-term R&D objectives (timescale of several months to 1 year) are carried out by the Copernicus Marine Service production centres (Figure 2). Tier 1 activities address issues requiring fast responses for rapid implementation within the Copernicus Marine Service. They can integrate developments from Tier 2 or Tier 3 activities that have reached maturity and have proven benefits for uptake into operational systems.

Long-term R&D activities are as crucial as short and midterm activities for the sustainable evolution of the Service.

Evolutions of systems and products are managed by MOi through a formal development process. Evolutions proposed by production centres (TACs, MFCs) are validated by MOi through a specification and design process, and the quality-control of new systems is performed to ensure non-regressions in terms of quality compared to former distributed versions (Sotillo et al., this issue).

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 2: The three tiers of R&D activities in the Service Evolution programme and their corresponding timescales, funding frameworks and players.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

4.2 Tier-2

4. ACHIEVEMENTS

Service Evolution R&D projects led to a large uptake in the Copernicus Marine Service (which is still on-going), spread across the different overarching themes of the service (Figure 3). A synthesis of the outcomes of Service Evolution R&D projects and of potential uptake in the service can be found online for the 1st batch and 2nd batch of projects.

4.1 Tier-3 Several Tier-3 R&D projects were identified as especially relevant for the Copernicus Marine Service long-term evolutions, as listed in the Service Evolution: R&D priorities document (v4).

Service Evolution R&D projects led to the addition of new products in the Copernicus Marine Service portfolio. In particular, during the 2018-2021 period and following the first batch of projects, observations from European highfrequency radars have been progressively distributed to provide information on ocean surface currents over coastal areas with high spatial and temporal resolution (Figure 4). Estimates of phytoplankton functional types derived from satellite ocean colour observations were added in the Copernicus Marine Service portfolio and are assimilated in the North-West Shelf (NWS) reanalysis (Figure 4). The overall goal was to improve the capability of operational systems to simulate the biogeochemical ocean state and, therefore, to enhance the capability of the Service to monitor the ocean health. To better address marine resources monitoring and management, information on micronekton has been added to the portfolio. Micronekton is a key ecosystem component at mid-trophic level to understand and model the habitats and population dynamics of most large marine species. Following the 2nd batch of projects, new products are and will be progressively added to the portfolio, such as information on the largest waves and on diurnal sea surface warming for the blue ocean, phytoplankton functional type spectral absorption for the green ocean.

Three H2020 projects have been funded as H2020 Copernicus Marine Service Evolution: -C EASELESS (Copernicus Evolution and Applications with Sentinels Enhancements and Land Effluents for Shores and Seas) [2016-2019], targeting the coastal ocean and aiming at developing proof-of-concept coastal extensions of Copernicus Marine Service products, - IMMERSE [2019-2021], targeting the blue ocean and aiming at preparing next generation numerical ocean models used by the Copernicus Marine Service, -S EAMLESS [2021-2023], targeting the green ocean and aiming at improving the current European capability to simulate and predict the state of marine ecosystems. These projects have been particularly monitored (invitation to kick-off and annual meetings, analysis of deliverables and analysis or contribution to the roadmap of developments integration into the Service). The uptake from these projects is being analysed.

Figure 3: Pie charts of countries hosting the Principal Investigator of Copernicus Marine Service Evolution R&D projects for (a) the 1st batch of 12 projects and (c) the 2nd batch of 18 projects. Pie charts showing the repartition of projects in terms of overarching R&D thematics for (b) the 1st batch of 12 projects and (d) the 2nd batch of 18 Copernicus Marine Service Evolution R&D projects.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Examples of Copernicus Marine Service Evolution R&D activities that have been transferred into operations.

Finally, several R&D projects paved the way for future key evolutions for potential new service lines envisioned for the Copernicus Service Marine. Such evolutions include: - a better coastal zone state monitoring and forecasting (e.g., Sentinel satellite derived nearshore bathymetry, shoreline position, and turbidity, improved interfacing of regional products with coastal products), - land-estuary-ocean continuum for consolidated river discharges of dissolved and particulate matter), - development of probabilistic forecasts and improved characterization of uncertainties associated to model products, - preparation of the next generation of Arctic sea-ice forecasting system, - development of regional ocean climate change projections for the 21st century, - use of machine learning techniques to generate highlevel observation-based products, - development of indicators for marine species (e.g., habitats).

Service Evolution R&D projects also contributed to upgrades in systems underpinning Copernicus Marine Service products to generate the best possible ocean information and to prepare the next generation of operational systems. Among various developments, there are: -e nhanced representations of coupling effects between ocean-wave-sea-ice-atmosphere components, - a more complete representation of dynamical processes in ocean and wave models, - upgraded data assimilation capabilities (including to prepare ensemble data assimilation), - enhanced capabilities in regional ocean uncertainty quantification, - development of bio-optical models and assimilation of optical data, - improved modelling of tides in global models of the general ocean circulation. Several projects also focused on enhanced quality assessment procedures of Copernicus Marine Service products (Sotillo et al., this issue).

Another type of achievement from Tier 2 Service Evolution R&D projects is the widening of the Copernicus Marine Service scientific community they provoked, with an involvement of various EU countries (Figure 3). For instance, the 2nd batch of 18 projects were conducted by scientific teams involving more than 26 public and private institutes across Europe, thereby expanding the Copernicus Marine scientific community and developing new and fruitful partnerships with the research community. For 68% of the 1st batch of projects, the Principal Investigator’s institute was not an already existing partner of the Service.

Overall, projects have also provided a better scientific understanding on ocean dynamics, which provides insights on processes for which an enhanced representation in systems could allow a better representation and monitoring of the ocean state. The corresponding gradual scientific and technical improvements of Copernicus Marine Service integrated systems have contributed to deliver ocean forecasts and reanalyses of increased accuracy for a better marine environment monitoring.

#57

4.3 Tier-1

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

external initiatives relevant for the Service, development of upstream data, etc. A special issue on Copernicus Marine Service scientific advances during its first phase of implementation (2015-2018) was also published in Ocean Science. 24 scientific articles were gathered in this special issue, covering various topics related to the different MFCs and TACs and to improve scientific ocean knowledge.

Achievements from production centres, including scientific and technical improvements of their products over 2015-2021, are described in the production centres individual papers in this Special Issue. A synthesis of R&D achievements by production centres during 2015-2017 can also be found in Le Traon et al. (2017) and Le Traon et al. (2019). An illustration of the transfer from research conducted in Tier-2 R&D projects to the Copernicus Marine Service operations is provided here for two projects (Figure 4). The first example project, INCREASE, laid the necessary foundations to prepare the integration of existing European high-frequency radar (HFR) operational systems into the Copernicus Marine Service portfolio. This technical project provided quality-controlled real-time HFR observations, set the basis for the management of historical data and enabled an HFR operational node to acquire data from partners, quality control and distribute them operationally. In the wake of the project, Copernicus Marine Service In Situ Thematic Assembly Centre (TAC) progressively ingested and distributed HFR data (Figure 4), in collaboration with EMODnet and EuroGOOS.

5. CONCLUSIONS

To remain a state-of-the-art, world leading service addressing more user and policy needs, the Copernicus Marine Service has organized R&D Service Evolution activities. The Service Evolution R&D programme tackles short-, mid- and long-term R&D needs through different actors and frameworks, from producers, R&D open calls for tenders, to external national and European projects (e.g., H2020). These activities also expand the Copernicus Marine Service scientific community and link it with production centres in order to foster transfers from research to operations.

The second example project, TOSCA, developed the scientific and technical capability of the Copernicus Marine Service to produce a new satellite ocean-colour product for Plankton Functional Types (PFTs, i.e., categories of Chlorophyll) and assimilate them into a marine ecosystem model to enhance the monitoring and simulation of biogeochemical indicators of the health of European shelf-sea (Figure 4). In the wake of the project, satellitederived PFTs were operationally produced and distributed by Ocean Colour TAC, and the ocean biogeochemical reanalysis covering the EU NWS now assimilates these data instead of total Chlorophyll to refine estimates of the biogeochemical ocean state.

Over the first implementation phase of the Copernicus Marine Service, the R&D Service Evolution programme led to major and continuous improvements of the Service, with a large uptake. This includes the addition of new products, a better scientific understanding of ocean dynamics, the upgrade or preparation of the next generation of systems operated in the Copernicus Marine Service, and the development of enhanced or new service lines. The R&D Service Evolution programme and its short- and mid-term activities covered the various thematics addressed by the Service, from the blue, green, white ocean to models and observations, quality assessment of the products, etc. The uptake of R&D developments in the Service will continue over the coming years.

4.4 Communications Short-, mid- and long-term R&D outcomes from Copernicus Marine Service have been presented and disseminated to a wider scientific audience in numerous conferences and publications in international peer-reviewed journals. For instance, more than 43 articles (resp. 47 articles) were published in international peer-reviewed journals from the 1st batch (resp. 2nd batch) of Tier 2 R&D Service Evolution projects.

As the Copernicus Marine Service enters a new phase over 2021-2027, Service Evolution R&D (Tier 2 and especially Tier 3) will contribute to pave the way for future key evolutions for potential enhanced or new service lines envisioned for the Service. These include a coastal ocean core monitoring, step-change monitoring of the marine biology going up to high-trophic levels (exploited and protected species), stepchange monitoring of the Arctic Ocean, climate projections over the 21st century dedicated to the marine environment (incl. coastal zones and ecosystems), but also 20th century ocean reanalyses, ensemble forecasts, higher-resolution products and improved CO2 air-sea fluxes monitoring.

Since 2017, a Copernicus Marine Service session has been organized at the European Geophysical Union (EGU) General Assembly. The session includes contributions from other lines of activities than the R&D Service Evolution, such as the User Uptake programme (Durand et al., this issue),

COPERNICUS MARINE SERVICE PRODUCT QUALITY ASSESSMENT SOTILLO, M.G.1,2, GARCIA-HERMOSA, I.1, DRÉVILLON, M.1,

RÉGNIER, C.1, SZCZYPTA, C.3, HERNANDEZ, F.4, MELET, A.1 , LE TRAON, P.Y.1 Mercator Ocean International, Toulouse, France - 2Puertos del Estado, Madrid, Spain - 3CELAD, Toulouse, France - 4IRD, Toulouse, France

specific PQ documentation for each product available in the catalogue. The PQ cross-cutting activity is under continuous improvement, and this paper indeed illustrates how PQ processes, firstly implemented during MyOcean Projects, have been standardized and reinforced during the Copernicus 1 service period (2015-2021). Finally, a brief overview of main guidelines to enhance PQ activities along the future Copernicus 2 service phase is provided.

INTRODUCTION Useful information on scientific quality of the Copernicus Marine Service (CMEMS) products (such as error levels for ocean observations, or reliability of delivered forecast/ analysis/reanalysis model products) has to be provided to end-users in a consistent and effective way. A wide range of operational oceanography products, such as remote sensed or in-situ observations, and modelled outputs, are delivered by the Copernicus Marine Service for ocean physical, biogeochemical and sea-ice variables. It is, therefore, a challenge to establish and to implement required homogeneous product quality (PQ) procedures. The Copernicus Marine Service relies on PQ metrics and validation procedures inherited from MERSEA and MyOcean projects (Maksymczuk, 2016), and follows operational oceanography’s best practices, well established within the GODAE Oceanview/OceanPredict international community (Hernandez et al., 2015; 2018).

1. PRODUCT QUALITY: STRATEGY, PROCESSES AND ORGANIZATION The Copernicus Marine Service relies on a complex set of observing and modelling systems. Also, the generation and evolution of the products’ portfolio is characterized by an increasingly complex data (and software) management process. Apart from ensuring generation and delivery of its product portfolio, the Copernicus Marine Service evaluates with quantitative metrics the scientific quality of its products and is responsible of informing end-users about relevant PQ information. Thus, scientific PQ documentation is issued for each product, and it is delivered alongside products at their release.

The Copernicus Marine Service product quality assessment strongly relies on the global ocean coverage provided by Sentinel and other satellite observations. Marine in situ observations are our main sources of information on the ocean and its “ground-truth”. However, measuring at sea remains a troublesome technical challenge, and, despite a general source quantity increase during the last decade, in situ observations are still sparse in the global ocean. Satellite and in situ observations are fully integrated in the Copernicus Marine Service system and in consequence, the number of valid observations and its evolution in time are primary key performance indicators and quality metrics. Modelled products provide Near-Real-Time (NRT) forecasts and analyses as well as Multi-Year (MY) reanalysis products.

To achieve its PQ objectives, CMEMS has outlined a PQ strategy and has established a product quality assurance loop, across the different service elements and common for all producers. This PQ assurance loop is similar to generally adopted by other operational ocean and meteorological/ climatic services in terms of PQ issues (shown in Figure 1). User requirements and feedback (applied to all steps) come first but they are not shown in the diagram as they are gathered and guide the production systems evolution. On one end there are the research and development activities that support the implementation of new products within CMEMS including: - those from production centres (PCs) and inputs from Service Evolution projects (see Melet and Le Traon, this issue); - other research programmes (i.e., H2020, Horizon Europe).

This article provides a synthesis of the Copernicus Marine Service PQ activities, the current organization and main user oriented PQ outcomes. PQ information is disseminated to end-users (e.g., PQ metrics delivered through the web portal, updated in NRT) through highlights along with the

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: CMEMS PQ Assurance Loop. Schematic view of the stages of the operational oceanography PQ assurance loop (in green), enabling the production of consistent PQ information (exchanged through PQ documents and websites) at the various stages of the product life cycle, and including the iteration (/collection of feedback) from the Marine Service users through the Service Desk. For the complete list of the CMEMS documents (internal and end-user oriented) related to the PQ process (and their acronyms) see Table 1.

During Copernicus 1, PQ activities were based on each producer’s expertise and conducted as per the loop shown in Figure 1. The different organizational elements (PCs or Production Centres), which produce observational products (the Thematic Assembly Centres; TACs) or modelled forecasts, analyses and reanalyses (the Monitoring and Forecasting Centres; MFCs), are responsible for monitoring the scientific quality of their products, generating the required product quality information, delivering regular updates of a selection of metrics and issuing the requested CMEMS PQ reference documentation. As shown in Figure 2, producers’ PQ activities contain both the pre-operational qualification phase and the operational validation. Part of this PQ information, generated by MFCs and TACs, is used to generate end-user-oriented PQ documents (i.e., QUIDs). PQ metrics, generated by producers during the operational validation, are managed by producers themselves. In some cases, this information is provided online via the producers’ internal websites. A number of metrics, agreed among all producers, are delivered in a homogeneous and standardized way, they are centrally monitored and later disseminated to end-users through the Copernicus Marine Service web portal (by means of the new PQ-Dashboard).

At this stage, relevant scientific quality assessments are carried out (some of it is gathered in peer reviewed publications) ensuring that production systems maintain stateof-the-art performance and are based on reliable / cuttingedge science. Another pillar of PQ assurance process is documentation. The complete list of Copernicus Marine Service documents (reference, internal and end-user oriented) related to the PQ process is provided in Table 1. Reference documents describe: PQ activities coordination, documentation requirements, lay out existing guidelines, and define the strategy. Internal documentation is furnished by producers for management purposes. All PQ activities carried out by producers in the pre-operational qualification phase (for new systems scheduled for entry into service) are detailed by producers in the (CMEMS internal document) Scientific Qualification Plan (ScQP). The same applies for the product’s operational validation, procedures are detailed in the (CMEMS internal document) Scientific Validation Plan (ScVP). End user documents include the product quality information document (QUID) and the synthesis quality overview (SQO), more information in Section 3.

#57

A strong coordination process is established to ensure this cross-cutting PQ activity, involving all MFC/TAC production centres (18) and more than 60 PQ experts. Two organizational elements are in place to ease coordination: - t he Product Quality Coordination team, managed by Mercator Ocean International, - t he Product Quality Working Group (PQWG), gathering up to 60 thematic PQ experts, assigned by production units to ensure regional expertise on blue, green and white observational and modelled products.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- design, implementation and operational maintenance of the PQ-Dashboard, - collection of metrics provided by the different organizational elements (plus implementation of any required operational processes needed to manage them), - facilitation of PQWG meetings and Copernicus Marine Service PQ representation in different forums, - definition of Copernicus Marine Service PQ strategy and reporting activities. The PQWG is a unique forum, enabling the exchange of methods and expertise between product quality experts, and across common variables and metrics. It is a key organizational element to ensure the needed (bottom-up and topdown) exchanges about PQ within the Marine Service.

The PQ coordination takes around 3 person-years in order to ensure: - management and evolution of the product quality information (including PQ documentation evolution and review process),

Figure 2: Schematic view of the PQ cross-cutting organization. Main organizational elements, functions and processes ruling the PQ info production, dissemination and evolution within CMEMS.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

indicators. Now, this last functionality is available only for selected variables (i.e., sea surface temperature, Chlorophyll, and sea ice concentration), but will soon be extended to a wider set. Furthermore, this user-friendly platform improves the interactive display of PQ metrics (including monthly updates of estimated accuracy numbers, EAN), and enables the visualization of direct comparisons between any Copernicus Marine Service model product and in-situ observation timeseries at fixed mooring locations (the socalled ‘Class 2’ metrics, in GODAE terminology).

2. PRODUCT QUALITY EVOLUTION ALONG COPERNICUS-1: MAJOR ACHIEVEMENTS 2.1 The PQ-Dashboard: A new end-user gateway to PQ information The new version of a Copernicus Marine Service PQ central website called “Product Quality Dashboard” (PQD) was developed in 2020 and launched in January 2021. The new feature of this PQ-Dashboard concerns ocean monitoring and forecasting capacity summaries. They can now be displayed for the global ocean and the regional European seas, and for the three families of parameters: -B lue Ocean (physical), -G reen Ocean (biogeochemical), - White Ocean (sea ice).

The PQ-Dashboard is intended for new and experienced Copernicus Marine Service users, as it includes a virtual tour, detailed explanations for each graphic and a dedicated section gathering “expert” validation metrics (with an extended number of parameters). A search zone lets the user directly display the scientific quality information related to a given product using its catalogue identifier (ex: GLO_ANALYSIS_ FORECAST_001_024). The new Class 2 metrics, showing timeseries of observations at fixed buoys together with their model counterpart are displayed by a click on the map as shown in Figure 4.

These summaries (Figure 3) include recent information on the number of satellite and in-situ observations available, and a first attempt of user-friendly forecast accuracy

Figure 3: The PQ Dashboard: Snapshot of the Monitoring and Forecasting Capacity Summary for the global ocean sea-surface temperature (as provided by the PQ-Dashboard in February 2021).

#57

2.2 Enhancement of PQ metrics and standardization of their operational production

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- to produce user friendly quality scores and synthetic overviews of product quality. - to provide the number of valid observation availability, more information on quality flags and quality checks, as well as why some model products may be less validated (usually due to lack of observational data to be used as reference). - to provide higher-resolution error estimates (and error estimates as an extra product dataset).

During the first phase of the Copernicus Marine Service (2015-2018), the product quality cross-cutting coordination ensured that metrics, individually defined and exploited by producers since MyOcean Project, were identified using a common naming system. This was necessary for individual verification procedures and constituted also a major step toward homogenization and standardization of PQ procedures. Likewise, basic sets of metrics were defined as baseline, especially for modelled product validation. Following the GODAE standards, four classes of metrics are defined for model products in order to check: 1) S ynoptic and large-scale behaviour of modelled fields (Class 1: eyeball validation), 2) M odel timeseries and vertical sections at higher frequencies (Class 2: statistics at specific mooring locations and sections), 3) I ntegrated quantities against previous estimates in literature (Class 3: applied to transport, heat content, etc.), 4) Last but not least, evaluation of model performance against contemporaneous collocated observations (Class 4: skill scores, residual, pointwise and spatially averaged model-observation metrics).

While it was possible to start addressing the first two points before 2021, the last one is linked to the continuous improvement of the catalogue, and progress toward this objective shall be made during Copernicus 2 (see Section 4). Product quality experts have implemented new metrics in their respective validation procedures and validated new variables all along 2015-2021 period, including wave variables (new in the catalogue at the end of phase 1) and biogeochemical variables. Quality improvement of products themselves was also quantified (see Copernicus Marine Service element description articles, this issue). Meanwhile, the number of metrics monitored centrally almost tripled in 3 years (as illustrated in Figure 4). The significant increase between 2020 and 2021 is due to: - the addition of Class 2 metrics in monthly deliveries of producers for the central validation website (example of Class 2 metric in the PQ-Dashboard shown in the lower panel of Figure 4), and - the homogenization effort done as part of the new website development.

During the second phase of the Copernicus Marine Service (2018-2021), validation documents and PQ processes in place were reviewed, and their evolution was scheduled in order to meet some of the main users’ feedback which were:

Figure 4: Temporal evolution in the number of PQ monitoring metrics generated/delivered for physical ocean (blue), biogeochemical (green), wave (orange) and sea-ice (cyan) products, managed through the CMEMS central cal/val website (upper panel). Snapshot of the PQ-Dashboard interactive display of Estimated Accuracy Numbers (EAN) for each product on predefined areas, together with visualization of model and insitu observed timeseries at mooring locations (lower panel).

#57

2.3 Evolution of user-oriented PQ documentation achieved: toward more fit-for-purpose documents

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

download throughout Copernicus 1. In 2016, 40 QUIDs were issued (with 1850 downloads in 2017), whereas these numbers increased up to 60 QUIDs issued in 2020 (with 3154 downloads in the same year).

User-oriented (external) documents are: QUIDs and the recently added SQOs. QUIDs are reports that contain a succinct description of the scientific steps taken, by production unit experts in the subject, to validate product datasets. It reports pre-operational qualification results and validation results. Reference values for product quality metrics (EANs are also provided). Updates of these reports are released with product updates and they are available on the catalogue as part of products’ documentation. Once the product is operational, websites complement the “static” PQ information available in QUIDs by displaying recent updates of the EANs, either in near real time (mostly through the producer elements, PC websites) or in slightly delayed mode through the (monthly updated) PQDashboard.

2.4 Enhancement of PQ cross-cutting activities PQ activities have an important cross-cutting component and evolutions of these activities depend on coordinating elements at various levels within the service. The observed enhancement in PQ assessment was possible thanks to the commitment of all MFC/TACs and specifically, their PQ experts (through the PQWG, a consolidated discussion group, developed in the last 3 years, with periodic meetings). The combined action of the PQ Coordination Team and the PQWG was key to facilitate required exchanges. Such as: discussion on objectives, identification of solutions and agreements on standards; and to ease the timely implementation of processes across the different Production Units. Coordinating this synergy, had a significant role in the homogeneous evolution of PQ processes across the complex system of systems that is CMEMS. Indeed, the seamless coordination between the different organizational PQ elements is a significant success achieved during Copernicus 1.

The resulting evolution of the user-oriented PQ documentation throughout 2015-2021 was an achievement. The aim was to meet the service requirements and users’ needs by improving, standardizing and homogenizing contents delivered by producers. PQ documentation during Copernicus 1 reached a higher level of maturity in quality control procedures: e.g., application of standard templates and inclusion of tabulated resulting EANs; becoming more fit for purpose; and the setting-up of a document review process associated to releases (internally by MOi). Another notable achievement was the preparation of a new lighter Synthesis Quality Overview document for each catalogue product by 2021. The SQOs provide users with a more synthetic evaluation of product quality to assess its fitness for purpose.

This collaborative framework fostered cross-fertilization processes and knowledge transfer between groups, favouring links between the operational and the scientific community. This permitted feedback from external research projects that, in time, may provide interesting applications to CMEMS PQ processes. Likewise, CMEMS favours sharing relevant scientific quality information, presenting it in peer-reviewed publications (see the Special Issue in Ocean Science (2019) that includes 19 papers).

Finally, note the significant increase in the number of produced external user-oriented PQ documents (delivered by producers and managed by the Service) and their

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

CMEMS Referential PQ Documents PQ Strategic Plan (PQ-SP)

Reference document for all CMEMS contributors. Provides overarching priorities that governs PQ policy in CMEMS.

PQ Management Plan (PQ-MP)

Top-level document (referential for all contributors) describing all CMEMS PQ processes that provides guidance to the Production Centres concerning the PQ activities (including cross-cutting ones). CMEMS Internal PQ Documents

Scientific Qualification Plan (ScQP)

Scientific Validation Plan (ScVP)

Sci. MultiYear Prod. Validation Plan (ScMYVP)

Reflects PQ qualification procedures performed by Production Centres during the Pre-Operational Qualification Phase, describing the full set of tools, methodologies, and data used to: perform tests, verify non regression hypothesis and assess reliability/performance of proposed “N+1” future systems. Describes the PQ routine verification procedure performed by each Production Centre for all their NRT products using its current operational systems. Document specific for MY products (i.e., reanalyses or reprocessed), describing metrics dedicated to: • the identification of main biases and suspicious trends in the time-series of MYPs, RAN or REP products. • the characterization of the overall “ocean representativity” (in terms of main scales and ocean features variability) depicted by the MY products. End-User Documents (available for each catalogue product):

Quality Information Document (QuID)

Document dedicated to Users. Describes the quality of a specific product. It is a «static and comprehensive» document, informing on the scientific information provided by producers about how a product is created and evaluated and its accuracy level.

Synthesis Quality Overview (SQO)

Homogenized short summary of PQ, general users oriented; (New document; SQOs prepared for all products at the end of Copernicus 1; to be published in Copernicus 2).

Table 1: List of reference, internal and user-oriented documents related to the CMEMS PQ process. Status of the PQ documentation (at the end of Copernicus.

scientific quality experts, enhancing timeliness of quality information with the objective of achieving PQ responses closer to production/delivery times (in NRT terms for observing and forecasting products and on a monthly basis for the MY products), - extend the currently available PQ processes to generate more metrics; defining more pertinent skill scores for forecasts; and implementing new approaches to validate the new very high-resolution products and biogeochemical model products (the validation of the latest is currently quite limited due to a lack of reference observational products, especially on a NRT basis), - the promotion/viewing of external quality metrics. Actions to increase the number of metrics shall come along a prior analysis and discussion about the design of new metrics to integrate and meet specific users’ views and needs, - streamlining of PQ documentation (specially the review process) focusing on fit-for-purpose and useroriented. New SQO documents (already delivered internally for products in 2020 and 2021) will be available to users. Furthermore, QUIDs will evolve; as they are eminently scientific, very technical and,

3. THE PRODUCT QUALITY ROADMAP: FORESEEN CHALLENGES FOR COPERNICUS 2 The evolution of the Copernicus Marine Service PQ processes will continue in the framework of Copernicus 2 with the main objective to keep up-to-date the PQprocesses currently in place. Hence, required upgrades will be performed to: -e xtend PQ info available (both for producers and endusers), and - i nclude any new observational data source or evolved model product in PQ processes. A roadmap for post-2021 scenario that includes, among others, the following objectives and working lines: - increase the current offer of PQ information for users. To this aim, there will be a continuous evolution of the new PQ-Dashboard, increasing the number of metrics shown and improving the interface to meet additional user needs (feedback from users on the PQDashboard will be collected to for future evolutions), - strengthen operational links between producers and

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- reinforce cross-cutting collaborations within the Service (promoting discussion forums and task groups on specific issues, as well as closer relationships between TACs and MFCs), with other Copernicus services and with other services and organizations (i.e., EUMETSAT, NOAA, etc.) in a global framework.

in general, expert-user oriented. Differences between products’ documents (e.g., in the extent) will also be further homogenized, - provide highly synthesized and simple information on products’ quality (applying the synthesis approach) to be useful for high-level users (policy makers/ managers/ beginner users), searching for ways to add scientific quality information in a simple way with product maturity matrices,

ACKNOWLEDGEMENTS The authors want to acknowledge and thank the invaluable input and enriching exchanges from all PQ working group experts that participated through the Copernicus 1 phase. ARC-MFC: Arne Melsom, Çağlar Yumruktepe, Patrik Bohlinger. BAL-MFC: Anja Lindenthal, Lena Spruch, Priidik Lagemaa, Svetlana Verjovkina, Jandt-Scheelke, Laura Tuomi, Ina Lorkowski. BS-MFC: Eric Jansen, Simon Leonardo Lima, Veselka Marinova, Luc Vandenbulcke, Stefania Ciliberti, Elisaveta Peneva, Marilaure Grégoire, Arno Behrens, Joanna Staneva, Ali Aydogdu. GLO-CPL: Jan Maksymczuk. GLO-MFC: Coralie Perruche, Lotfi Aouf. IBI-MFC: Pablo Lorente, Elodie Gutknecht, Bruno Levier, Cristina Toledano, Tomasz Dabrowski, Joe Mcgovern. MED-MFC: Vladyslav Lyubartsev, Anna Teruzzi, Gianpiero Cossarini, Emanuela Clementi, Romain Escudier, Anna Zacharioudaki. INS-TAC: Henning Wehde, Marta de Alfonso, Jerome Gourrion, Inmaculada Ruiz. MOBTAC: Nathalie Verbrugge. NWS-MFC: Christine Pequignet, Susan Kay, Inga Golbeck, Andrew Saulter, Wibke Düsterhöft-Wriggers. OC-TAC: Antoine Mangin, Gianluca Volpe, Philippe Garnesson. SI-TAC: Cecilie D Wettre, Anton Korsov, Mohamed Babiker. SL-TAC : Guillaume Taburet, Marie Isabelle Pujol. SST-TAC: Emmanuelle Autret, WAV-TAC: Elodie Charles, Romain Husson. WIND-TAC: Abderrahim Bentamy, Gerd-Jan van Zadelhoff, Rianne Giesen, Maria Belmonte Rivas.

BIBLIOGRAPHY: Hernandez F, E Blockley, G B. Brassington, F Davidson, P Divakaran, M Drévillon, S Ishizaki, M.G Sotillo, P J. Hogan, P Lagemaa, B Levier, M Martin, A Mehra, C Mooers, N Ferry, A Ryan, C Regnier, A Sellar, G C. Smith, S Sofianos, T Spindler, G Volpe, J Wilkin, E D. Zaron & A Zhang (2015) Recent progress in performance evaluations and near real-time assessment of operational ocean products, Journal of Operational Oceanography, 8:sup2, s221-s238, DOI:10.1080/175587 6X.2015.1050282

Maksymczuk J., F. Hernandez, A. Sellar, K. Baetens, M. Drevillon, R. Mahdon, B. Levier, C. Regnier, A. Ryan, (2016), Product Quality Achievements Within MyOcean, Mercator Ocean Journal #54

Hernandez, F., Smith, G., Baetens, K., Cossarini, G., Garcia-Hermosa, I., Drevillon, M., ... & von Schuckman, K. (2018). Measuring performances, skill and accuracy in operational oceanography: New challenges and approaches. New Frontiers in Operational Oceanography, 759-795.

Le Traon, P. Y., Reppucci, A., Alvarez Fanjul, E., Aouf, L., Behrens, A., Belmonte, M., ... & Zacharioudaki, A. (2019). From observation to information and users: the Copernicus Marine Service perspective. Frontiers in Marine Science, 6, 234

Melet and Le Traon (this issue) Ryan, A. G., Regnier, C., Divakaran, P., Spindler, T., Mehra, A., Smith, G. C., ... & Liu, Y. (2015). GODAE OceanView Class 4 forecast verification framework: global ocean inter-comparison. Journal of Operational Oceanography, 8(sup1), s98-s111.

J. M. Huthnance, P.-Y. Le Traon, A. Melet, M. Tonani, E. Stanev, M. Grégoire, and A. Pascual. Ocean Science. Special issue. The Copernicus Marine Environment Monitoring Service (CMEMS): scientific advances. (24 papers) Editor(s):

THE COPERNICUS MARINE SERVICE UPSTREAM OBSERVATION INFRASTRUCTURE REPPUCCI, A., LE TRAON, P.Y. Mercator Ocean International

Change and transformation in the Copernicus Marine Service is inextricably linked to data supply and the proper specification of requirements. The core dependency from upstream data must be managed strategically to ensure a steady data flow, and that new requirements are regularly included in observing plans of upstream data providers (Mercator Ocean, 2016; Le Traon et al., 2019).

INTRODUCTION The Copernicus Marine Service (CMEMS) is highly dependent on the timely availability of comprehensive satellite and in-situ observations (Le Traon et al., 2019). In situ and satellite observations are used, through dataassimilation, to constrain the analysis and forecasting systems, and validate their outputs, and to provide highlevel products to downstream users.

To manage this core dependency, Mercator Ocean international (MOi) has established coordination mechanisms with upstream satellite and in-situ coordination bodies (ESA and Eumetsat for satellite, EEA, EuroGOOS and EMODnet for in situ).

Satellite l

n ea oc olor c

S1A Sentinel 1A S2A Sentinel 2A S3A Sentinel 3A S1B Sentinel 1B

20 21

S2A

waves 20

S1A

S3B

S1B S2B S3A

wa ve

introduction of n

201

ce fa re r u u a s at se mper ce ai te se

2 0 18

20 17

l ea

Figure 1: Sentinels data integration in CMEMS.

sea ice sea level

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

In situ data are of paramount importance for the Copernicus Marine Service because they provide information about the ocean interior which cannot be observed from space. In situ observations also can locally sample high-frequency and high-resolution ocean processes, in particular in coastal zones, that are essential for model and satellite validation activities. This is why a dedicated centre, the InSitu Thematic Assembly Center (INS-TAC), gathers observation data from in situ networks (e.g., the Global Ocean Observing System (GOOS), and the European Global Ocean Observing System (EuroGOOS)). Then, the INS-TAC distributes high-level quality-controlled datasets in nearreal-time (NRT) and reprocessed (REP) mode to Copernicus Marine Service production centres and down-stream users. Nowadays, the INS-TAC is one of the major aggregators of in-situ observation at global level (see Figure 2). It has a structure developed coherently and with a strong synergy with other components of the pan European data management landscape such as EMODnet and as SeaDataNet.

1. SATELLITE AND IN-SITU OBSERVATIONS USED BY THE COPERNICUS MARINE SERVICE As a Copernicus service, the Copernicus Marine Service makes an extensive use of satellite data from the Copernicus satellite component, through the European Space Agency (ESA) and the European organisation for the exploitation of Meteorological Satellites (EUMETSAT). During the first phase of the service, all Sentinel satellites carrying an instrument able to observe the ocean have been integrated as soon as data were made available by space agencies (see Figure 1). In addition, all contributing missions made available to Copernicus services thanks to international agreements, by the European Commission and member states, have been used to complement the Sentinels improving spatial and temporal coverage.

Figure 2: In-Situ TAC Dashboard.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- verify that observation information is optimally used in the analysis step and improve the assimilation components, - quantify the impact of the present observation network on ocean analyses and forecasts, - demonstrate the value of an observation network for operational purposes, and - help define and test new mission concepts.

2. UPSTREAM SATELLITE COORDINATION AND REQUIREMENTS The Copernicus Marine Service critically depends on the near-real-time availability of high-resolution satellite data. As delegated body by the European Commission, MOi is in charge of the coordination of satellite data required for the Copernicus Marine Service. This is implemented through regular interactions with space agencies to contribute to: - t he definition of upstream data Requirements, - t he Coordination of Acquisition plans for Sentinels and Contributing mission, - the Evolution of Sentinels (Next Generation) contributing to the Copernicus Expert Groups.

Analyses (see MFCs and TACs papers in this issue) have shown that the outstanding development of the Copernicus Sentinel Constellations has had a major impact on the Copernicus Marine Service. The Service is primarily driven by the needs identified by its users, which are regularly gathered by MOi and Service’s partners through workshops (regional and thematic), training sessions, questionnaires and regular user interactions with the Service Desk. These needs are then used to define future satellites observation requirements of the Service, considering the complementary nature of satellite and in situ observations and the role of modelling and data assimilation.

Regular meetings with ESA and EUMETSAT are also held to exchange feedback on upstream data quality, to plan the introduction of new satellite missions, or to improve data flow. MOi is also invited to contribute to the definition of new satellite missions attending the dedicated Mission Advisory Group (MAG) meetings. Figure 3 gives an overview of the coordination activities carried out for the Copernicus Marine Service.

Satellite requirements for the Copernicus Marine Service have been first defined in the Global Monitoring for Environment and Security (GMES) Marine Core Service (MCS) implementation group report (Ryder, 2007) and then regularly updated (e.g., Copernicus Marine Service, 2017) to take into account the evolution of the Copernicus satellite component and emerging user needs.

In parallel, dedicated discussions take place with Copernicus Marine Service production centres to provide regular updates about the impact that Copernicus Satellites has on their systems to:

Figure 3: Coordination of Satellite data.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

quality assurance, standardization) for physical, chemistry and biological in-situ data and avoid duplication of work and developments. Figure 4 gives an overview of the insitu coordination activities carried out for the Copernicus Marine Service.

3. IN-SITU UPSTREAM COORDINATION AND REQUIREMENTS

Operational aspects of the Service’s in-situ component are implemented by the INS-TAC which is responsible for the implementation of technical interfaces with upstream data providers, and day to day operations.

The Copernicus in-situ component is organized and implemented by member states. The European Commission delegated to the European Environment Agency (EEA) and the Entrusted Entity (Mercator Ocean International) the high-level coordination aspects, and political agreements at member state and EU levels.

Copernicus Marine Service requirements for the in-situ observing system have been previously gathered in the framework of the GISC (Global Monitoring for the Environment and Security/Copernicus In-Situ Coordination) project led by the European Environmental Agency. As the Copernicus Marine Service evolves, some of these requirements have been met, while others need to be updated to the new commitments and capabilities of the service. As for Satellite observations, new in-situ needs (regularly gathered by MOi and the Service’s partners) must go through the added-value chain of the service to be translated into observation requirements. The latter are then presented and discussed with the main European institution dealing with in-situ data coordination, such as EEA, EuroGOOS, Euro-Argo, to plan their implementation.

A long-standing partnership with EuroGOOS (an association of national governmental agencies and organizations that form the UNESCO-IOC GOOS Regional Alliance for Europe) has been established for definition of in-situ requirements and the identification of potential improvements for in-situ observing systems relevant to the Service. At European level, MOi contributes to the development of the European Ocean Observing System (EOOS) a framework to better coordinate Europe’s ocean observing capacity, and collaborates with DG MARE EMODnet initiative to improve data assembly tasks (ingestion of new data sets,

Figure 4: Coordination of in-situ data.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

toward the service is expected to increase in the coming years thanks to the availability of new Satellite missions and new in-situ platforms. MOi will maintain high-level coordination with ESA and Eumetsat to contribute to the definition of future Sentinels and to prepare their integration into the Service. Exchanges at European level with EEA (insitu cross cutting coordination), EuroGOOS and EOOS, and at international level with GOOS, will be strengthened. Links with projects aiming at improving ocean observation systems will be kept and new links will be established.

4. PROSPECTS The Copernicus Marine Service is fed by the Copernicus Space component with Sentinel data, contributing and thirdparty missions and by in-situ observations from European and international networks, under the supervision of the Copernicus in-situ component; this flow of upstream data

REFERENCES: CMEMS (2017). “CMEMS requirements for the evolution of the Copernicus Satellite Component Mercator Ocean and CMEMS partners”, February 21, 2017. CMEMS (2021). “Copernicus Marine Service requirements for the evolution of the Copernicus In Situ Component”, March 2021

Mercator Ocean (2016). CMEMS High Level Service Evolution Strategy, September, 2016. https://marine.copernicus. eu/sites/default/files/ CMEMS-High-Level-ServiceEvolution-Strategy-FVSeptember-20-2016.pdf

Le Traon et al. (2019). From observation to information and users: The Copernicus Marine Service Perspective. Frontiers in Marine Sciencevol 6, pp 234. https://doi.org/10.3389/ fmars.2019.00234

Ryder (2007). “GMES Fast Track Marine Core Service Strategic Implementation Plan”. Final draft. By Dr. Peter Ryder for the Marine Core Service Implementation Group. Marine Core Service implementation group report.

DEVELOPING COPERNICUS MARINE MULTI-YEAR PRODUCTS IN SUPPORT OF OCEAN STATE REPORTING ACTIVITIES DRÉVILLON, M., VON SCHUCKMANN, K., LE TRAON, P.Y.,

REPPUCCI, A., DERVAL, C., DELAMARCHE, A., DRILLET, Y., BOURDALLÉ-BADIE, R., BRICAUD, C., MELET, A. Mercator Ocean International

ocean state reporting and marine environment monitoring based on these multi-year products is achieved through: - the annual release of the Copernicus Marine Service Ocean State Report (see von Schuckmann et al, this issue), - the development of operational Ocean Monitoring Indicators (OMIs), and related error bars (e.g., von Schuckmann et al., this issue).

INTRODUCTION The Copernicus Marine Environment Monitoring Service (hereafter referred to as Marine Service) provides open, free, regular and systematic reference information on the physical state, variability and dynamics of the ocean, sea ice and marine ecosystems, for the European regional seas and the global ocean. This capacity includes the provision of multi-year products (MYP), which consist of high-quality retrospective data records for the recent decades, obtained from the reprocessing of observational data and from model reanalyses (Le Traon et al., 2019). Additionally,

These developments must rely on continuous, homogeneous and high-quality ocean estimates that: - go up to real time, - ensure high resolution coverage of the European regional seas and the global ocean, - can serve as a reference for operational users.

Figure 1: Number of multiyear products (reprocessed observations and reanalyses) on the CMEMS catalogue in 2016 (orange) and in 2020 (blue), for the ocean physical variables, biogeochemical variables, and waves.

#57

Production Centres (Monitoring and Forecasting Centres [MFCs] and Thematic Assembly Centres for ocean observations [TACs]) achieve operational production of reanalyses (RAN) and reprocessed observational products (REP L3/L4 and delayed-time quality controlled in situ) using delayed time quality-controlled input observations and boundary conditions (e.g., reanalysed atmospheric forcing). Each centre possesses the expertise, and can make specific scientific choices, in order to provide stateof-the-art RAN or REP for each region (global and regional seas) and for all variables.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

1. TOWARD MORE CONSISTENT MULTI-YEAR PRODUCTS As illustrated in Figure 2, the main upstream dependency for the Marine Service MYPs is the regular uptake of observations of past, state-of-the-art reprocessed satellite L2 products as well as collections of in situ databases. All RAN products, and many REP, need reanalysed atmospheric fields (ERAinterim, then ERA5), are progressively making use of varying river inputs, and atmospheric deposition. During the period 20152021, high level coordination was set up with observation agencies (ESA, EUMETSAT, NASA, NOAA, etc.), with EMODNET, as well as with other Copernicus services (Climate Change, Land, Emergency, Atmosphere) for the provision of reprocessed upstream data. This requires a dedicated effort of integrated operations to ensure a fully consistent, long term approach, across the different service elements, organize their interoperability, their inter-dependencies, and joint operations.

As illustrated in Figure 1, the number of MYPs proposed to users has increased during the period 2015-2021, and many new products describing a larger number of Essential Ocean Variables are now available on the Marine Service catalogue, such as surface pH and CO2 reconstructions, ocean currents deduced from altimetry, or the first wave model reanalyses. The Marine Service ensures the collection of “best quality” input data and maximal use of multiple observation systems. In the longer term, the Service aims at a fully consistent approach across global and regional reanalyses, organizing their inter-operability, their inter-dependencies, and joint operations closer to real time (a few months only) with a systematic yearly update. Therefore, specific cross-cutting coordination actions were led (see also).

The first steps toward this integration were: - to promote the use of Marine Service catalogue products by the Marine Service producers as input, - to identify how and when the TACs and MFCs multiyear production schedules could be phased. Regarding the first objective, new products were designed and implemented in order to be used as reanalysis inputs: the REP of OSTIA SST being fully consistent with the real time SST product, and the easyCORA dataset, designed by INS TAC using the MFCs inputs.

This article describes the Marine Service achievements during the Copernicus 1 period 2015-2021, including improvements for consistency (section 1), continuity (section 2) and quality assurance (section 3) of Marine Service MYPs.

Figure 2: Schematic of the time coverage over past decades depending on the type of observations (coloured arrows). Main categories of multi-year products (REP/RAN) making use of these different types of observations (white arrows).

#57

Secondly, a first integration schedule was drawn, showing the importance of relying on a common atmospheric forcing produced operationally (ERA5 from Copernicus Climate Change service, available since 2019). This schedule also highlighted three datasets that were limiting the extension in time of “best estimate” MYPs up to 3 months before present: -E RA5 (best estimate stops 3 months before present), - sea level anomalies (orbital corrections available 3-4 months before present), - in situ (delayed time quality control induces a 6-month delay).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. CONTINUITY AND TIMELINESS: TOWARD INTERIM PRODUCTION, UP TO ONE MONTH BEFORE PRESENT In order to provide reliable trends together with consistent and recent information on the ocean state, MYPs have to be continuous in time and ideally reach one month before present. Starting 3 months before present, ERA5T atmospheric fields continue the ERA5 reanalysis by using the same atmospheric reanalysis system and leveraging near real-time upstream data. In consequence, the ERA5/ ERA5T series allows the provision of a continuous atmospheric forcing over the past decades up to one month before present, which in turn can be used by the Marine Service MYPs. If not homogeneous in terms of quality, ERA5/ERA5T is homogeneous in terms of technical characteristics (grid, variables, frequency, main settings).

In consequence, the current implementation of regular and timely deliveries of in situ CORA products, together with reprocessed sea level anomaly time extensions in January and June each year, were organized with the technical coordination. In parallel, automatic “time extension mode” is progressively implemented for most REP. When this is in place, regular updates of reanalyses can be scheduled every 6 months, for the best estimate MYPs to reach 1 year before real time.

As shown in Figure 3, one objective of the cross-cutting MYP activity was to collaborate with the Marine Service producers to propose such continuous MYPs, making an optimal use of the ERA5/ERA5T suite. Depending on upstream dependencies, production centres will be implementing scenario (A) or (B) between 2021 and 2024.

Figure 3: Schematic of the organization of the Marine Service multi-year interim production.

#57

Scenario (A) is the final objective where the reanalysis or reprocessing has a “best estimate” time extension production mode (making use of best upstream data and producing regular time extensions) while the “interim production” making use of near real-time upstream data fills the gap between the end of the best estimate timeseries (more than 3 months before present) and the present. Most TACs can implement the scenario (A), except TACs relying on altimetry (Sea Level and Wave) for which interim production is still under development.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

observations. Quality information documents (QUIDs) are also available for OMIs, including intercomparison whenever possible (see Sotillo, M.G et al, this issue). In collaboration with the Copernicus Climate Change Service, and following international standards (Global Climate Observing System), the appropriate quality of the products for climate studies is regularly assessed, including for ocean model reanalyses. As reviewed by Storto (2019a), ocean reanalyses have the capacity to capture ocean variability and trends, and are used as oceanic initial conditions by seasonal forecasting systems. However, biases and errors appear in areas where observations are sparse. In order to provide information on areas where the signals derived from ocean reanalyses are robust and reliable (Von Schuckmann et al., 2018), the Global Reanalysis Ensemble Product GREP was developed from four global ocean reanalyses using NEMO but differing on their model parameterizations and data assimilation systems (Storto et al., 2019b). Error bars were also directly derived from the standard deviation between the four members as shown in Figure 4.

Scenario (B) is a first step, providing an “interim mode time extension” of the MYP, initialized at the end of the best estimate MYP at its entry into service, and will be implemented by most MFCs.

3. QUALITY ASSURANCE FRAMEWORK: DEVELOPING ROBUST INFORMATION One of the major objectives of the Copernicus Marine Service is to deliver useful scientific quality information for each product, including model reanalyses and reprocessed

Figure 4: Volume transport (units SV) from the multi-product approach averaged over the period 1993-2014 and the 2017 (both red). Estimates of Lumpkin and Speer (2007) have been added for comparison (blue). Uncertainty ranges are derived from the ensemble standard deviation. Arrows indicate the direction of the mean flow through the sections. See https://marine.copernicus.eu/access-data/oceanmonitoring-indicators/mean-volume-transport-across-section.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

cutting coordination will focus on the use of river inputs, and the use of Copernicus Marine Service products (Ocean data assimilation in wave reanalyses and biogeochemistry reanalyses, assimilation of easyCORA in situ observations) by Service producers.

4. POST 2021 OBJECTIVES

The first objective for the continuation of Copernicus Marine Service is to maintain and improve the integration of the multiyear products, by ensuring the interim production for all variables which are the basis for ocean monitoring and ocean state reporting activities (see also von Schuckmann et al., this issue).

This work on past timeseries is a prerequisite to the development of ocean climate projections. Collaboration across Copernicus Services will be further developed, in order to improve the consistency of the Essential Ocean Variables critical for ocean state monitoring and reporting in polar and/or coastal areas.

Among the many improvements that will be planned in each production centre, the scientific and technical cross-

ACKNOWLEDGEMENTS

These achievements rely on the work of the Copernicus Marine Service Multi-Year product experts including (nonexhaustive list): Signe Aaboe, Emmanuelle Autret, Lars Axell, Roland Aznar, Jonathan Baker, Bruno Buongiorno Nardelli, Andrea Cipollone, Gianpiero Cossarini, Eric de Boisseson, Antoine Delepoulle, Romain Escudier, Helene Etienne, Claudia Fratianni, Gilles Garric, Rianne Giesen, Simon Good, Jerome Gourrion, Marie-Laure Gregoire, Eric Greiner, Elodie Gutknecht, Doroteacino Iovino, Laura Jackson, Anette Jonsson, Thomas Lavergne, Stephane Law Chune, Jean-François Legeais, Patrick Lehodey, Bruno Levier, Vidar Lien, Leonardo Lima, Ye Liu, Simona Masina, Sandrine Mulet, Adam Nord, Coralie Perruche, Andrea Pisano, Isabelle Pujol, Roshin Raj, Richard Renshaw, Joanna Staneva, Ad Stoffelen, Andrea Storto, Tanguy Szekely, Guillaume Taburet, Jonathan Tinker, Anton Verhoef, Richard Wood, Jiping Xie, Chunxue Yang, Hao Zuo.

REFERENCES: Bojinski, S., Verstraete, M., Peterson, T. C., Richter, C., Simmons, A., & Zemp, M. (2014). The concept of essential climate variables in support of climate research, applications, and policy. Bulletin of the American Meteorological Society, 95(9), 1431-1443. Storto A., Alvera-Azcárate A., Balmaseda M. A., Barth A., Chevallier M., Counillon F., Domingues C. M., Drevillon M., Drillet Y., Forget G., Garric G., Haines K., Hernandez F., Iovino D., Jackson L. C., Lellouche J.M., Masina S., Mayer M., Oke P. R., Penny S. G., Peterson K. A., Yang C., Zuo H. (2019)a Ocean Reanalyses: Recent Advances and Unsolved Challenges. Front. Mar. Sci., https://www. frontiersin.org/article/10.3389/ fmars.2019.00418

Le Traon P.Y., Reppucci A., Alvarez Fanjul E., Aouf L., Behrens A., Belmonte M., Bentamy A., Bertino L., Brando V.E., Kreiner M. B., Benkiran M., Carval T., Ciliberti S.A., Claustre H., Clementi E., Coppini G., Cossarini G., De Alfonso Alonso-Muñoyerro M., Delamarche A., Dibarboure G., Dinessen F., Drevillon M., Drillet Y., Faugere Y., Fernández V., Fleming A., Garcia-Hermosa M. I, Sotillo M.G., Garric G., Gasparin F., Giordan C., Gehlen M., Gregoire M. L., Guinehut S., Hamon M., Harris C., Hernandez F., Hinkler J.B., Hoyer J., Karvonen J., Kay S., King R., Lavergne T., LemieuxDudon B., Lima L., Mao C., Martin M.J., Masina S., Melet A., Buongiorno Nardelli B., Nolan G., Pascual A., Pistoia J., Palazov A., Piolle J.-F., Pujol M. I., Pequignet A. C., Peneva

E., Pérez Gómez B., Petit de la Villeon L., Pinardi N., Pisano A., Pouliquen S., Reid R., Remy E., Santoleri R., Siddorn J., She J., Staneva J., Stoffelen A., Tonani M., Vandenbulcke L., von Schuckmann K., Volpe G., Wettre C., Zacharioudaki A. (2019) From Observation to Information and Users: The Copernicus Marine Service Perspective. Front. Mar. Sci., https://doi.org/10.3389/ fmars.2019.00234

A. Storto, S. Masina, S. Simoncelli, D. Iovino, A. Cipollone, M. Drévillon, Y. Drillet, K. von Schuckman, L. Parent, G. Garric, E. Greiner, C. Desportes, H. Zuo, M. A. Balmaseda, K. A. Peterson (2019)b The added value of the multi-system spread information for ocean heat content and steric sea level investigations in the CMEMS GREP ensemble reanalysis product Climate Dynamics 53: 287. https://doi.org/10.1007/ s00382-018-4585-5 von Schuckmann, Karina, et al., 2018: The CMEMS Ocean State Report, issue 2, Journal of Operational Oceanography, 11:sup1, S1-S142, https:// doi.org/10.1080/175587 6X.2018.1489208

THE COPERNICUS MARINE SERVICE OCEAN REPORTING FRAMEWORK VON SCHUCKMANN, K., LE TRAON, P.Y., DREVILLON, M., DERVAL, C., QUADE, G., CROSNIER, L. Mercator Ocean International

organizations in charge of environmental and climate monitoring, policy and decision-makers. Its additional aim is to increase general public awareness about the status of, and changes in, the marine environment (von Schuckmann et al., 2016).

INTRODUCTION In 2015, Mercator Ocean International (MOi) and the Copernicus Marine Service (CMEMS) launched the ocean reporting activity, which consists of three principal tools: - t he peer-reviewed annual Ocean State Report (OSR), - t he Ocean Monitoring Indicator (OMI), - the annual summary for policy makers based on information from the OSR and OMIs.

Section 2 provides insight into the Copernicus Marine Service OSR functioning and major outcomes, Section 3 focusses on the Service OMI framework and Section 4 describes the Service summary for policy makers. Section 5 displays an outlook on future plans for the Copernicus Marine Service ocean monitoring activity during the second phase of Copernicus Marine Service.

The Copernicus Marine Service ocean reporting activity contributes to European and international agencies or

Figure 1: Overview on major components of the Copernicus Marine Service ocean reporting framework merging scientific knowledge and data products. The ocean reporting framework consists of three principal tools: the peer-reviewed annual Ocean State Report (OSR), the Ocean Monitoring Indicator (OMI) framework, and an annual summary for policy makers based on information from the OSR and OMIs.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The key function of the OSR is to bring together insightful studies and updates (including indicators) in the form of a peer-reviewed publications compiled into an annual report. The OSR is organized into 4 principal chapters (Figure 2): - chapter 1: state-of-the-art background information on Copernicus Marine Service activities focused on ocean reporting, synthesis of report outcomes, and solicited topical foci, - chapter 2: state-of-the-art science developments representing an advance in understanding of an important problem, - chapter 3: science-based demonstrators at the interface between science and society with immediate, far-reaching implications, - chapter 4: science-based reporting on changes in the ocean close to real time.

1. THE COPERNICUS MARINE SERVICE OCEAN STATE REPORT The development of the annual Copernicus Marine Service Ocean State Report is one of the priority tasks given by the EU Delegation Agreement for the Copernicus Marine Service implementation (CMEMS, 2014): «The preparation and release of annual Ocean State Reports, with clear schedule and content for an annual analysis describing the state of the global ocean and the European regional seas, in particular for supporting the Member States in their assessment obligations». The OSR activity was launched in 2015 with the preparation of the first OSR (von Schuckmann et al., 2016). Currently, the Ocean State Report activity is in its 6th cycle of successful reporting of the state, variability and changes in the marine environment.

The development of each report per cycle is organized through several steps which are shown in Figure 3. Each cycle of the OSR underlies a specific roadmap established to support the draft developments for chapters, as well as to support the author teams.

Figure 2: Overview on the topical organization of the Copernicus Marine Service Ocean State Report into 4 principal chapters (see text for more details). Scientific topics cover the blue, white and green ocean, and range from information close to real time up to longer term changes.

#57

period is organized within several steps in support of the author teams, - MS5 - OSR first submission to the journal for peerreview in July, - MS6 - OSR peer review process in collaboration with the Journal of Operational Oceanography.

The overall development is supported by the OSR chair through regular remote calls, and individual exchanges. Each cycle for the draft development spans nearly one year, and is organized within several milestones (MS). These include: - MS1 - Establishment of call of contribution guideline: The OSR chair develops each year in OctoberNovember a guideline for the call of contribution, which includes amongst others the criteria for contributions for each OSR chapter, - MS2 - OSR call of contribution: The call of contribution runs each year from December to January (year+1), which usually ends during the last week of January, - MS3 - Launch of OSR cycle: Mid-February, the outcomes of the call of contribution are communicated to the author teams, and the draft development is launched, - MS4 - OSR draft development: The period February to July is dedicated to the section development. This

Copernicus Marine Monitoring

Marine

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Albeit the huge challenges faced with a regular reporting process passing peer-review such as the Copernicus Marine Service Ocean State Report, or the Bulletin of the American meteorological Society (BAMS) statement of the climate (e.g., Blunden et al., 2020), the science community and the community working at the science-policy interface have strongly recognized these efforts. One particular reason is that due to the peerreview, this regular reporting can be accomplished in, for example, assessment processes, which act at the science policy-interface (e.g., IPCC).

Ocean

State

Report

Organisation: Example 4th issue Call of contribution Q4 2018

Launch draft Q1 2019

First full draft Q2 2019

Final draft Q2 2019

internal review

Revision Q2 2020

Revision Q1 2020

Launch draft Q1 2020

science & comm.

Draft develop.

Draft structure Q1 2019

Final edits

internal review

Submission July 2019

Accepted Q2 2020

Production & proof

Publication Aug. 2020

Publication June 2020 SUMMARY FOR POLICY MAKERS

Figure 3: Overview on the CMEMS OSR process as shown here for the development of the 4th cycle of the OSR. See text for more details. The lower panel shows the roadmap of the summary for policy makers (see Section 4).

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

distributed within a set of 11 families, which contain today a total of about 80 OMIs, either available as timeseries, maps of regional trends or maps of regional anomalies (Figure 4). The latter does not include a dissemination of numerical values, whereas timeseries and maps of trends are disseminated as numerical values, together with documentation on scientific value, quality and product information. All elements are freely available through the Copernicus Marine Service portal.

2. THE COPERNICUS MARINE SERVICE OCEAN MONITORING INDICATOR FRAMEWORK An Ocean Monitoring Indicator (OMI) framework has been implemented in Copernicus Marine Service which relies on scientific developments principally provided in the Ocean State Report activity (Figure 1). A large set of OMIs is

Figure 4: Organization of Copernicus Marine Service OMI framework into 11 OMI families during phase 1. Today, a total of about 80 indicators are available, distributed amongst these families. Three types of indicators are available, i.e., timeseries, maps of regional trends, and maps of anomalies.

3. Indicator quality evaluation and control: The indicator then needs to undergo rigorous quality checks, as well as to provide a reliable uncertainty framework, which relies on data processing, method evaluation, and the underlying data used. Also, this step needs a dual expertise of technique and science. This step requires to consider also flexibility to advancements in research and development activities, including the monitoring of potential biases detected in the upstream products. 4. Technical implementation: The technical implementation demands the design of an indicatorspecific tool, in which all needed information are included. A roadmap for the technical implementation and the choice of the tool is needed, and should be established by specific technical experts accordingly to this proposal. The major task is to establish first

The development, implementation and dissemination of ocean indicators is a complex activity, which demands a wide range of expertise, capabilities and technical solutions. Figure 5 provides a schematic overview on the chain of an indicator development, and the overall process encompasses 6 essential steps: 1. Co-construction for indicator development: Coconstruction needs to be put in place in order to define the theoretical baseline for a potential indicator, ideally through close collaboration with a scientific expert and a stakeholder. 2. Technical indicator development: In a next step, the indicator must be technically developed, including the choice and use of one (or several) relevant data products and the technical tool for the indicator processing. This step would need to be done by a technical expert in collaboration with a scientific expert.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

the establishment of scientific added-value, including a regular update of application context, thus tightly linked to developments linked to step (1) and (3). Moreover, this step also demands the contribution of communication experts, including graphic designers, science writers and experts for the development of visualization tools on the web. In addition, a regular bulletin to disseminate information close to real time would be an asset.

the tool, but also demands a regular review, and update in-line with new indicator developments (see also step (1) and (2)). 5. Operational production: The operational production demands the establishment of the regular implementation into the indicator tool (see implementation phase (4)), and should be linked to activities on step (3). This step is the key element for assuring a regular update to meet the indicator criterium (6) (see section 2). In addition, the operational production is tightly linked to the release of products used to evaluate the indicators. Any changes in this operational chain will have direct implications on this step (5). 6. Added value & communication: An important step for the recognition and dissemination of the indicators is

All these steps are needed to satisfy the indicator criteria 1-6 as introduced in Section 3.1. The criterion 7 is included to provide a tool to limit and frame the indicator activity, supporting the need of a roadmap establishment for this activity.

Figure 5: Representation of the complexity of ocean indicator development: Co-construction of indicator concepts through scientific development and development of relevance with stakeholders are needed to establish a theoretical baseline for an indicator concept, which then needs to go through technical development, including quality control and uncertainty evaluations. Once an indicator is then established, a rigorous technical implementation is needed, including the establishment and provision of numerical values and documentation. In a next step, the indicator will run through operational production to follow the timeliness criteria. Finally, outcomes need regular scientific assessment and interpretation, and communication tools assure a provision of ocean information to a wide audience.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

March-April. This review is organized through a large set of co-working sessions during remote calls, - once the storyboard is drafted, a science writing expert reworks the text, and is then revised by the OSR chair and the internal communication expert through several remote exchanges, and co-working sessions (April-May). Occasionally, few additional experts are solicited for a point-wise review of the text to assure that the science content is appropriate, - in parallel, co-working sessions are organised for the graphical development, the design outlook and the development of schematics. For some cases, additional science experts are solicited for pointwise revision of new schematics to assure science credibility, - the final phase takes place during end of May to early June to prepare the final draft of the OSR summary document, which is usually anticipated for publication in early June.

3. THE COPERNICUS MARINE SERVICE OCEAN REPORTING SUMMARY FOR POLICY MAKERS Each year, a summary for policy makers of the OSR (cycle – 1) is coordinated by the OSR chair, and relies on a coconstruction approach with scientists, communication and graphic designers. These documents are freely available on the Copernicus Marine Service portal. The launch of this work runs in parallel with the draft development of the new upcoming OSR cycle (Figure 2). The OSR summary organisation is shown in the lower part of Figure 2, and includes the following steps: - during Q1 of the OSR cycle-1 year (February to March), the storyline of the OSR summary is developed in collaboration with Mercator Ocean international and a team of communication experts, - the OSR summary storyboard is then revised and refined by a science writer and concept expert from an external company collaborating with the OSR chair, and the internal communication expert during

The Summary is aiming to reach a wide audience from policy and decision makers, environmental agencies and institutions to the wider public. The topical content is preliminary guided by the evaluations of the OSR, and includes regular updates from several OMIs (Figure 6).

OCEAN STATE REPORT (4) SUMMARY

MAJOR IMPACTS OF CLIMATE CHANGE

The ocean is undergoing sweeping, severe, and unavoidable changes, with major impacts on marine ecosystems and humanity. Rising seas threaten coastal and low-lying areas, increased ocean

OCEAN ACIDIFICATION: THE OCEAN IS BECOMING MORE ACIDIC

acidification threatens marine organisms and ecosystems, and sea ice is retreating. Hundreds of millions of people live alongside the ocean, and over three billion

Yearly mean surface sea water pH reported on total scale Units: pH reported on total scale

KEY FIGURES

Trend from 1985-2018

~30%

8.11

Atmospheric Carbon Dioxide

8.10

20-30%

8.09 pHT [-]

people depend on marine biodiversity for their livelihood. Consequently, these changes are forcing people across the globe to fundamentally alter how they coexist with the ocean.

8.08

The ocean is more acidic than in preindustrial times

of excess CO2 has been absorbed by the ocean since preindustrial times

8.07

WMO GLOBAL CLIMATE INDICATORS

8.06

Absorbed by the ocean

The pH of contemporary surface ocean waters is already 0.1 pH units lower than in pre-industrial times. As pH is logarithmic, this 0.1 pH unit change is equivalent to a nearly 30% increase in ocean acidity since pre-industrial times.

Global Mean Sea Level Units: Centimetres/year

2010

Ocean and Water

2015

UN SUSTAINABLE DEVELOPMENT GOALS

2008

2012

2016

Ocean Thermal Expansion

1.2

The rate which sea level rise is accelerating ±0.073 MM/YEAR each year

5 30°N 0

0°

3.3

Rate of sea level rise per year since 1993 in mm per year

±0.4 MM/YEAR

30°S

no TP-A drift correction, trend = 3.35 mm yr-1

-5

60°S

Sea level rise in centimetres (cm) for the period 1993-2018.The curves show different corrections for an instrument anomaly on the TOPEX-A altimeter (TP-A). The black, red and green curves show TP-A drift corrections applied based on work by Ablain, Watson/Dieng, and Beckley, respectively. The shaded red area shows a 90% confidence interval for each measurement, and the blue curve represents no TP-A correction. Source: Modified after WCRP Sea Level Budget Closure Group (2018), appears in the Copernicus Marine Ocean State Report 4. Warming Melting Atmosphere Continental Ice

Sea Level Rise

KEY FIGURES

Trend from 1993-2018 10

TP-A drift corrected (Beckley), trend = 3.12 mm yr-1

Since 1993, global mean sea level has risen at a rate of 3.3 ± 0.4 mm per year. New calculations in the fourth Ocean State Report reveal that sea level rise is accelerating, with this rate increasing by 0.12 ± 0.073 mm/year each year. An instrument drift correction to the global mean sea level time series is also discussed in the fourth Ocean State Report.

2005

60°N

TP-A drift corrected (Watson/Dieng), trend = 3.13 mm yr-1

2004

2000

Regional Mean Sea Level Trends scale Units: Millimetres/year

Trend from 1993-2018

2000

1995

Annual global mean surface seawater pH derived from the Ocean Monitoring Indicator “Surface Ocean pH”, showing an overall trend for decreasing pH and increasing acidification. Source: Copernicus Marine Ocean State Report 4.

TP-A drift corrected (Ablain), trend = 3.08 ± 0.39 mm yr-1

1996

1990

Ocean Acidification

-2 1992

1985

Carbonic Acid

6 Global Mean Sea Level [cm]

SEA LEVEL RISE CONTINUES, BUT AT AN ACCELERATED PACE

Water

Ocean acidification threatens marine ecosystems and impacts many biological processes. Changing ocean chemistry is particularly dangerous to calcifying organisms like shellfish and coral.

Carbonate

mm/yr

The ocean is a major sink of anthropogenic excess CO2. This absorption of carbon mitigates the effects of global warming, but it also results in a major threat to marine life — ocean acidification.

-10 60°E

120°E

180°

120°W

60°W

0°

Regional trends in sea level change from 1993-2018 in millimetres per year (mm yr-1) showing that sea level is rising for the vast majority of the global ocean. Source: Copernicus Marine Ocean Monitoring Indicator.

WMO GLOBAL CLIMATE INDICATORS Ocean and Water

UN SUSTAINABLE DEVELOPMENT GOALS

Ocean Ocean Warming Warming

Knowledge of sea level rise is fundamental, as it allows us to better characterise the consequences of rising seas for coastal populations and low-lying areas.

Figure 6: : Example for the CMEMS ocean reporting summary for policy makers as part of the OSR4 cycle.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- improvements for the dialogue and collaboration between the OSR chairs, the OSR Review Chairs and the Journal.

4. FUTURE PLANS FOR THE CMEMS OCEAN MONITORING FRAMEWORK

Another critical area for progress has been identified, i.e., to improve the OSR (and its summary for policy makers) visibility, credibility and value. Three areas have been identified for this task, i.e.,: - increase contribution, visibility and credibility within the science community, - improve knowledge transfer and OSR visibility at the science-policy interface, - increase knowledge transfer and visibility at the science – wider audience interface.

The Copernicus Marine Service ocean reporting activity has obtained huge success and experienced continuous momentum since its implementation in 2015. For example, statistics for the OSR reveal a significant increase of recognition from nearly 14,000 views for OSR1 to nearly 23,000 views for OSR4. The contribution of European experts has grown from about 100 authors up to more than 150 experts. Also, the OMI framework has grown since its implementation, and more OMIs have been implemented into the framework. Moreover, a lot of experience has been gained over the past years, which provide a unique opportunity to further evolve the ocean reporting framework, and to identify critical areas of advancements.

Also, advancements are foreseen for the CMEMS OMI framework. One critical area of development is an improved implementation of the OMI criteria as discussed in section 3. Moreover, major revisions of the current OMI framework are planned, predominantly aiming to increase the topical foci, to strengthen the consideration of indicator criteria, and to reinforce the balance between technical challenges and user needs.

For the OSR, areas of advancements are discussed in collaboration with the Copernicus Marine Service Scientific and Technical Advisory Team (STAC) which include: - the reinforcement of the editorial and review boards to strengthen the overall process. the improvement of OSR criteria and guidelines for a more accurate support to the author teams,

REFERENCES: Blunden, J., & Arndt, D. S. (2020) State of the Climate in 2019. Bulletin of the American Meteorological Society, 101(8), S1–S429. https://doi.org/10.1175/2020BA MSStateoftheClimate.1

Von Schuckmann, K. P.-Y. Le Traon, E. Alvarez-Fanjul, L. Axell, M. Balmaseda, L.A. Breivik, R.J. W. Brewin, C. Bricaud, M. Drevillon, Y. Drillet, C. Dubois, O. Embury, H. Etienne, M. García Sotillo, G. Garric, F. Gasparin, E. Gutknecht, S. Guinehut, F. Hernandez, M. Juza, B. Karlson, G. Korres, J.-F. Legeais, B. Levier, V.S. Lien, R. Morrow, G. Notarstefano, L. Parent, Á. Pascual, B. Pérez-Gómez,

C. Perruche, N. Pinardi, A. Pisano, P.-M. Poulain, I.M. Pujol, R.P. Raj, U. Raudsepp, H. Roquet, A. Samuelsen, S. Sathyendranath, J. She, S. Simoncelli, C. Solidoro, J. Tinker, J. Tintoré, L. Viktorsson, M. Ablain, E. Almroth-Rosell, A. Bonaduce, E. Clementi, G. Cossarini, Q. Dagneaux, C. Desportes, S. Dye, C. Fratianni, S. Good, E. Greiner, J. Gourrion, M. Hamon, J. Holt, P. Hyder, J. Kennedy, F. Manzano-Muñoz,

A. Melet, B. Meyssignac, S. Mulet, B. Buongiorno Nardelli, E. O’Dea, E. Olason, A. Paulmier, I. Pérez-González, R. Reid, M.-F. Racault, D.E. Raitsos, A. Ramos, P. Sykes, T. Szekely & N. Verbrugge (2016) The Copernicus Marine Environment Monitoring Service Ocean State Report, Journal of Operational Oceanography, 9:sup2, s235-s320, DOI: 10.1080/ 1755876X.2016.1273446

SEA LEVEL THEMATIC ASSEMBLY CENTER

PUJOL, M.I1., FAUGÈRE, Y.1, PASCUAL, A.², DIBARBOURE, G.3, MORROW, R.4 AND SL-TAC TEAM CLS, France, ²IMEDEA, Spain, 3CNES, France, 4LEGOS, France

between producers and users was enabled by the copernicus Marine Service’s (CMEMS) unique structure including observations and models. Two full reprocessing were also implemented (2018, 2021) leading to major improvements. The precise multi-missions crosscalibration provided accurate Ocean Monitoring Indicator (OMI) such as the Global Mean Sea Level for climate application. The error budget of products has also been refined and promoted through four peer-reviewed publications (Pujol et al., 2016, Taburet et al 2019, Ballarotta et al 2019, Sánchez-Román et al., 2020) and Copernicus Marine Service Quality Information Documentation (QUID).

OVERVIEW During the 6 years of Copernicus phase 1, the Sea Level Thematic Assembly Center (SL-TAC) has been completely transformed from a demonstration to a fully operational system. This transformation has been carried out according three main axes. First, the consolidation of the production robustness has been a major priority. Data from four new missions have been assimilated in 6 years: Jason-3, reference mission for climate scales (2016), Sentinel-3A (2017) and Sentinel-3B (since 2019) to complete coverage of mesocales. The Copernicus constellation is now operational and fully ingested in the SL-TAC system. To complete this coverage, the Chinese Collaborative mission HaiYang-2B was integrated in 2020, still increasing the sampling quality. From 4 missions in 2015, the near-real time (NRT) system is now ingesting operationally 6 missions in 2021, ensuring the system robustness. The team has been reinforced and the DUACS software on which the SL-TAC production is based has been consolidated. Thanks to this efficient infrastructure, 22 new significant versions, impacting the product portfolio and content, were implemented.

Beside evolutions implemented within the Service the SLTAC also has been preparing the next generation of products (with an enhanced special resolution) by integrating space Agencies (CNES, ESA) R&D. Demonstration samples, fully exploiting the altimetry content and with improved posting (1 km), have been proposed. Observation Simulation Experiment (OSE) tests carried out by MFCs showed a gain up to 10% in the model output assimilating these data (Benkiran et al., 2017). Similarly, the capacity of the high-resolution altimetry to improve the sampling of Sea level through “ice leads” was demonstrated. These studies allowed us to formulate recommendations to the Space Agencies concerning evolutions of upstream products. They also open very exciting prospects for the next phase of Copernicus Marine Service with new highresolution product lines for nadir altimetry and the future Surface Water Ocean Topography (SWOT) mission.

Then, the SL-TAC team worked on an extension and improvement of the product portfolio. The gridded products, highly popular among users, were added to the catalogue at the beginning of the project (2015), and several regional products were added after, such as the Europe (2019) product and Mediterranean & Black Sea Mean Dynamic Topography (MDT) (2020). A whole set of variables is now available in products, such as absolute dynamic topography, geostrophic currents, and all the components needed by Monitoring Forecasting Centres (MFCs) to adapt the content of Sea level to their model. The methodology of development in interaction with MFCs, initiated in the Tailored Altimeter Product for Assimilation Systems (TAPAS) group in MyOcean, has continued during the 6 years with annual meetings in order to achieve this evolution. This interaction

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 The SL-TAC has produced multi-mission altimetry Sea Level products in NRT and Delayed Time (REP/DT) for MFCs and a various range of oceanographic applications for external users such as: climate forecasting centers, geophysical and

#57

biological communities. These products have mainly consisted of directly usable and easy to manipulate Level 3 (along-track cross-calibrated Sea Level Anomalies (SLA)) and Level 4 products (multiple sensors merged as maps or timeseries) and have been available in global and regional versions over European Shelves. In 2015, the SL-TAC started the Copernicus Marine Service in demonstration mode, with only 6 main product lines. During Copernicus 1, the SL-TAC has been completely transformed to a fully operational system. This transformation has been carried out according three main axes.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- reinforcing and using the method and procedure previously defined to manage emergency situations like the different prolonged data outage of the reference altimeter mission that occurred in 2019 and 2020. The interruption of the measurements for a long period of more than 8 days affected the NRT product generation of the SL-TAC. To secure the NRT service continuity and to avoid significant offsets and biases in ocean topography products, Sentinel-3A was used as a temporary anchor in the SL-TAC system. While not a reference altimeter, it whas been stable enough over a few weeks to ensure a seamless transition over the Jason-3 safe hold mode period once NRT regional bias maps were estimated, - introducing the Sentinel-3A mission in the system early 2017, in collaboration with Eumetsat. In the same way, the Sentinel-3B mission was introduced in early 2019. The global Synthetic-Aperture Radar (SAR) mode coverage of Sentinel-3A and B enables more accurate observations of small mesoscale signals and make them valuable measurement for the SL-TAC production. Additionnaly, the tandem leads to a better signal sampling thanks to the optimal 140° orbit phasing (as requested a few years ago by the Copernicus Marine Service). Since 2019, with 6 different altimeters available, the Sentinel-3 tandem contributes to more than 37% of the gridded (L4) SL-TAC global product. This contribution can reach more than 50% when the number of other altimeters available is reduced (Figure 1). An OSE showed that the Sentinel-3 constellation provides a significant gain of energy especially in high variability regions and it reduces errors by 50% in in these areas (Figure 2). Thereby, the Sentinel-3 constellation appears to be crucial in the restitution on the ocean mesoscale eddies in Copernicus Marine Service systems, - assessing the quality of the Sentinel-3A mission thanks to an experiment in the Algerian Basin. It operated an ocean glider and a ship mission, along the same track and in synchronicity with an overpass of the Sentinel-3A mission. This provided three independent views of the ocean velocity field, along a section that encompassed three different oceanographic regimes. The results demonstrated the capacity of Sentinel3A to retrieve fine-scale oceanographic features (~20 km). The comparison with measurements from in situ platforms showed significant improvements, about 30% in resolution and 42% in velocity accuracy, using a SAR mode with respect to lower-resolution mode of conventional altimetry (Heslop et al, 2017), - introducing the Chinese HaiYang-2B (HY-2B) mission in 2020. It improves the mesoscale sampling over the global ocean and reinforces the SL-TAC system resilience when incidents occur on other altimeter missions,

1.1 Insure a robust production ingesting from 4 to 6 altimeters The first achievement has been the efficient and robust service provided along the 6 years, without interruption of the NRT production and with regular temporal extensions of the REP/DT series, nominally on a quarterly basis since 2015. To ensure this service and maintain the system’s resilience and product quality, the SL-TAC implemented 22 different versions within the 2015-2020 period corresponding to different evolutions implemented. These versions handled various upstream product versions changes and platform incidents of the altimeter constellation. Then, the consolidation of the upstream data ingestion has been a major priority. Early 2015, measurements from 4 different altimeters were used. The SL-TAC has integrated measurements from new altimeters as soon as they were available and nowadays, measurements from 6 different altimeters are operated. The capacity to manage different product versions changes and platform incidents of the altimeter constellation, made the SL-TAC able to maintain the system resilience and product quality. Main achievements in terms of constellation management are: - leveraging measurements from Jason-3 mission in NRT conditions since September 2016. Jason-3 is the so-called reference altimeter: it guarantees the large-scale accuracy of SL-TAC products in NRT and it is the backbone of the longterm stability of REP/DT products for Climate research. The introduction of these measurements in the system was thus particularly challenging. It was indeed necessary to take into account the global and regional sea level biases observed between Jason-3 and its predecessor Jason-2 measurements in order to ensure the continuity of the sea level and avoid any impact on users and applications. A specific procedure was defined to properly manage the temporal variability of these biases. It included a strong coordination with space agencies ensuring data availability for both altimeter missions during the CalVal phase. This collaboration has been necessary to estimate the sea level biases that SL-TAC has needed to manage,

#57

- managing Various orbit changes that occurred for some missions which required an adaptation of the SL-TAC system. In that way, the Jason-2 interleaved and longrepeat orbits phases were respectively implemented in 2016 and 2018 before the definitive deactivation of the altimeter in 2019; AltiKa drifting phase was implemented in 2016; Cryosat-2 new orbit was implemented in 2020, - managing various incidents and anomalies observed on different missions. Different algorithms have been improved to better take into account the quality

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

evolution of different measurements (e.g., specific data selection applied on AltiKa to avoid residual errors during important miss-pointing anomalies). Other missions were temporally disactivated from the NRT system during specific events (e.g., long safe hold mode events observed on Jason-2 during year 2019). After several serious incidents affecting HY-2A measurements quality, the satellite was deactivated from the SL-TAC NRT system in 2015.

Figure 1: Temporal evolution of the fraction of multi-mission merged product (L4) coming from each sensor: Degrees of Freedom of Signal (DFS).

Figure 2: Error reduction thanks to Sentinel-3 constellation (OSE studies).

#57

1.2 Product quality improvement and extension the portfolio

consistency of the altimeter gridded product, thanks to independent tide gauge and along-track altimeter measurement (about -10% variance reduction compared to the previous DT-2014 version; Table 1) (Taburet et al, 2019). - in 2021, a new version of the product is under development, with DT-2021 standards. As for the previous versions, it will include numerous changes at different stages of the processing (upstream standards, data selection, cross-calibration, mapping). These evolutions are expected to significantly reduce the signal error at long-wavelengths for the alongtrack product and at mesoscale for the gridded product (expected 17% reduction of the variance of differences between SLA gridded field and independent along-track data in high variability areas compared to the previous DT-2018 version; Table 1).

Different evolutions were also implemented with to improve the product quality and propose new products and variables that better fit users’ needs. The SL-TAC upgraded regularly the 30-year timeseries through whole reprocessing. They allow to benefit from: upto-date altimeter standards and corrections available in upstream L2P product; and improvements done in the processing of along-track (L3) and gridded (L4) products over global ocean or regional seas. All improvements were also simultaneously implemented in the NRT processing chain: - in 2014, the DT-2014 series were delivered and NRT chain upgraded following the DT-2014 standards. Implemented evolutions enabled significant improvements with a better restitution of the signal at wavelengths < 250 km (nearly 10% reduction of the variance of the differences between SLA gridded field and independent along-track data in high variability areas compared to the previous DT-2010 version; Table 1). The spatial coverage in high latitude areas was also improved. This version was also characterized by the use of the new 20-year altimeter reference period. It improved interannual signals with more relevant intensities and spatial signature (Pujol et al, 2016). - in 2018, various evolutions at each step of the processing were also implemented and defined the DT-2018 standards. They improved significantly the mesoscale signal in the sub-tropical band and in high variability areas (reduction of the variance of differences between geostrophic current derived from altimetry and independent drifter measurements by nearly 5% compared to the previous DT-2014 version). Coastal areas were also improved, with a better

Comparison gridded product vs independent TP interleaved (for DT-2014 & DT-2018) or S3B (for DT-2021) along-track

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

In order to better fit users’ needs and applications, new products and variables were introduced in the SL-TAC catalogue: - different key geophysical corrections were provided in the along-track products. They enabled users to change the physical content of the sea level altimeter field according to their application. First implemented in the regional Mediterranean Sea product, this change was generalized to all along-track products in 2019. In late 2021, a new geophysical correction, dedicated to the internal gravity wave signal, will also be introduced, - the absolute dynamic topography (ADT) and geostrophic currents derived from the altimeter measurement were introduced as new variables in gridded products in 2017. They have facilitated and diversified possible applications of this product, - The MDT field, necessary for ADT computation, was

DT2014 Vs DT2010

DT2018 Vs DT2014

DT2021 Vs DT2018 (preliminary)

Open Ocean / Low variability

-2.1%

-3%

-5%

Open Ocean / High variability

-9.9%

-3.1%

-17%

Coastal (dist < 200 km)

-4.1%

-10.1%

-1%

Table 1: Reduction of the budget error between 2 reprocessing versions. Statistics issued from comparison between gridded SLA product and independent along-track measurement. Results presented for DT-2021 are preliminary and should be consolidated during the final DT-2021 validation phase.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

the whole European Seas, including the North East Atlantic and Baltic Seas in order to better address different regional MFCs, - a specific field, giving an information of the formal mapping error of the geostrophic current was introduced in the gridded NRT products in 2020. This evolution will be completed in REP/DT series during year 2021, - the C3S altimeter products were introduced in the CMEMS catalogue in late 2018, - specific products’ format and nomenclature evolutions were implemented in order to contribute to the CMEMS catalogue standards homogenization. The SL-TAC products are now all delivered in NetCDFcf1.6 standards, - fromIn a user-oriented perspective, the collaboration with MFCs (the most direct users of SL-TAC), was ensured through the organization of meetings and discussions with MFCs assimilation experts. These meetings were a place to discuss about different issues regarding altimeter sea level products (e.g., products resolution, physical content, observation errors, reference surface) and to design and test new experimental products before possible implementation. 6 of such meetings were organized during the 2015-2020 period.

also added to the SL-TAC catalogue. The global CNES_ CLS18 MDT field was introduced in 2020 and used in the SL-TAC production since Dec 2019. Compared to the previous version (CNES-CLS13), the new CNES_ CLS18 MDT has showed improved performance everywhere and most significantly in coastal areas and in strong western boundary currents. The Root Mean Square Deviation (RMSD) have been reduced by 10% within 100 km of coasts for the CNES-CLS18 (Mulet et al., 2021). In 2020, two regional MDT fields were also specifically estimated for Copernicus Marine Service applications in the Mediterranean and Black Sea. They were, for the first time, computed using a first guess based on the geoid model rather than a numerical model as done in previous versions. The availability of additional in-situ measurements compared to previous versions, also enhanced the quality of new regional MDTs. Their error has been estimated between 5 and 6 cm in mean dynamic height and between 10.5 and 11.5 cm/s for the associated mean geostrophic current (Jousset et al., 2020). These regional MDTs have been used in the SLTAC regional production since May 2021, - monthly mean sea level anomalies gridded fields were also introduced in the REP series in 2020, - in 2019, regional products, initially defined for the Mediterranean Sea and Black Sea were extended to

Figure 2: Effective spatial resolution, in kilometers, of the DUACS-DT2018 maps for the Global Ocean gridded product. From Ballarotta et al., (2019).

#57

Another important achievement for the SL-TAC was the refinement of products’ errors description. The different validation diagnostics and procedures were refined in order to: -q uantify errors of different products at different physical scale, from climate to meso-scale. This estimation involved different diagnostics for different variables and implied different independent measurement (in situ and altimeter) a/o intermediate altimeter products. The error budget was fully estimated and documented after each significant change in the production (i.e update of the QUID for different products). Scientific peer-reviewed publications also completed the quality information after each full reprocessing activities (Pujol et al., 2016; Juza et al., 2016; Taburet et al., 2019). Specific diagnostics were implemented in order to quantify observing capability of gridded products at regional scale. This diagnostic was presented by Ballarotta et al. (2019) and is illustrated by Figure 2, -a ssess the quality of Sentinel-3 measurements. A methodology of comparison of the along-track products with in-situ tide gauge measurements available in the Copernicus Marine Service catalogue has been implemented. Results obtained with Sentinel-3A and Jason-3 measurements were compared over the European coast. They revealed the improved capability of the Sentinel-3A to capture the signal near the coast, with a RMSD reduction of about 13% compared to the Jason-3 capabilities (Sánchez-Román et al., 2020),

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- assess the quality of different corrections applied on along-track products. The comparison with in-situ tide gauge was also used to quantify the impact of the Long Wavelength Error. It showed that this correction reduced errors in altimetry enhancing the consistency between altimeter and in-situ datasets (reduction of the errors up to 10%) (Sánchez-Román et al., 2020). Finally, the SL-TAC also contributed to the cross-cutting activities, and more specifically the Ocean State Report (OSR) and associated OMI production. - The SL-TAC leaded a/o contributed to different sections of OSRs from 2015 to 2019. - The sea level trend OMI was introduced in the OMI catalogue, first for the global ocean in 2018, then for regional seas in 2019, in coordination with the C3S production. Data processing and corrections developed with R&D activities (e.g., ESA CCI) were also taken into account to improve the OMI pertinence and precision. In that way, a specific correction for the Topex-A Altimeter Instrumental Anomaly and drift was implemented in the Sea level trend global OMI in 2020 as illustrated on Figure 3. This issue was also described in the 2019 OSR. - The North Pacific Gyre oscillation index (Di Lorenzo et al., 2008) was introduced in the OSR for year 2018 and in the OMIs catalogue in 2019. - The Kuroshio extension phase index (Qiu and Chen, 2005) was introduced in the OMIs catalogue in 2020.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: OMI GLOBAL_OMI_SL_area_averaged_anomalies. Gives the evolution of the mean sea level since January 1993 (in cm) computed from satellite altimeter observations (L4 gridded product). Timeseries has been low-pass filtered, the annual and semi-annual periodic signals have been adjusted, and the curve has been corrected for the GIA using the ICE5G-VM2 GIA model (Peltier, 2004). During 1993-1998, the dashed line shows an estimate of the global mean sea level corrected for the TOPEX-A instrumental drift, based on comparisons between altimeter and tide gauges measurements (Ablain et al., 2017).

1.3 Preparation of next generation products

test. Enjoyably, the assimilation of this product significantly reduced the model forecast errors as discussed by Benkiran et al., (2017). Authors underlined a 9% reduction of RMSD between the model forecast and observations. This result was also enhanced (additional 2 to 5%) when the altimeter physical content is fully compatible with model characteristics, thanks to the different geophysical variables available. The operational processing chain able to generate the new L3 product in NRT conditions has been developed during years 20202021. The new corresponding NRT product line is expected early 2022 in the Copernicus Marine Service catalogue.

Finally, The SL-TAC also prepared the next generation of products by integrating the R&D from space Agencies, H2020 or Service evolution projects in order to better address end-user needs. The improvement of the product resolution and spatial coverage were notably addressed and supported by CNES R&D activities in synergy with CMEMS long term evolution plan. First, an along-track L3 product was defined with a higher spatial sampling (i.e., about 1 km) than the conventional (about 7 km) product already present in the SL-TAC catalogue. It leveraged the full rate altimeter measurement. Specific innovative processing and corrections were applied in order to reduce the residual noise signal and led to a more accurate observation of the small mesoscale signal (up to ~55 to ~35 km according to the mission considered) with this new product. Indeed, a reduction by at least one third compared to the capability of the conventional product over the North East Atlantic Ocean was achieved. This demonstrative production was proposed to different MFCs for assimilation a/o validation

Then, experimental gridded products dedicated to the Arctic areas were also developed with support from CNES. They aim to estimate the sea level height in areas covered by sea ice. This is possible thanks to specific processing that enables the exploitation of the altimeter measurement in “ice leads” areas. As usual, this product was proposed to Arctic and Global MFCs for independent assessment. In addition to the scientific work, SL-TAC was also active to formulate recommendations and requirements to upstream provider (Eumetsat) in order to prepare its future transition to operation. SL-TAC is notably represented in the Sentinel-3 Quality Working Group.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2.1 Post 2021 perspectives

2. STATUS AT THE END OF COPERNICUS 1

In the coming years, the SL-TAC system will first have to be updated following the evolution of the altimetry constellation to ensure a robust sampling with a minimum of 4 satellites. These changes will first include the integration of new altimetry missions: S6A (2021); J3-EOL (2021), HY2C (2021), HY2D (2022), SWOT’s nadir (2022), S3C (2023). The Interleaved SAR/ Low Resolution Measurement (LRM) radar measurement modes on S6A will allow to continue the 30-year LRM timeseries. Although, a longer-term objective will be to use the SAR processing on the Sentinel altimeter missions. S3C (SAR) and HY2C and D (LRM) will be integrated to complete the coverage.

In 2021, the SL-TAC is managing 4 product lines in NRT; 9 product lines in REP/DT, including the redistribution of C3S sea level products; 6 product lines for auxiliary and intermediate data. This represents nearly 125 different datasets. The SL-TAC NRT processing system acquires and processes measurements from 6 different altimeters, including the Sentinel-3A/B tandem. They improve the mesoscale signal sampling and also ensure the continuity of climate signal thanks to the cross-calibration processing using the Jason refence mission. The high number of altimeters available also ensure the system resilience in case of failure of one mission. Procedures used for the different system evolution and implementation were consolidated and allows the SL-TAC to face important necessary emergency evolutions, as the temporary absence of the reference mission encountered in 2019 and 2020, minimizing the impact to users and applications.

Then, strong efforts need to be done to improve the resolution of sea level products, today limited to wavelengths larger than ~70 km (Dufau et al., 2016, Vergara, 2019). This will be possible thanks to the improvement of instruments (SAR on board S3A/B vs historical LRM) and the associated processing. They will enable a reduction of the posting of the along track product to ~1 km and the provision to regional MFCs improved observation and associated error for assimilation.

SL-TAC continuously worked on the evolution of products and their quality improvement, using outcomes from different R&D projects (e.g., CNES DUACS-RD; ESA CCI). The content of SL-TAC products was completed with different fields, auxiliary or derived from altimeter measurement, that can be used for different applications. Products quality was also refined, as well as its description for users. Six additional scientific peer-reviewed publications were issued (Pujol et al., 2016; Juza et al., 2016; Heslop et al., 2017; Taburet et al., 2019; Ballarotta et al., 2019, SanchezRoman et al., 2020) and QUID documents have been upgraded with up-to-date quality results.

After 2022, the launch of SWOT mission will most probably open a new area with its unprecedented spatial resolution with 2D swath of sea level. The availability of high resolution along-track and sSwath products combined with new mapping methods, such as dynamic mapping (Ballarotta et al., 2020) or multiscale calculation, will enable a higher resolution of L4 mapped products at global and regional scale. Data driven method has also emerged these last years and could be an alternative in the coming years. Local improvement will be performed in areas where the coverage and quality of the sea level observations are strongly degraded, notably in polar and coastal areas.

SL-TAC products, based on conventional altimeter measurement, now reached their maturity for the observation of the large mesoscale signal in open ocean. Along-track a/o gridded products are currently used by all MFCs, for assimilation a/o validation purpose and are also a useful upstream product for the Multi Observations TAC. The gridded altimeter product defined for the global ocean is currently the second most downloaded product by external users, with about 2000 distinct users identified in NRT and nearly 3500 in REP/DT.

Finally, regular reprocessing will be crucial on all altimeter missions with up-to-date geophysical corrections (atmospheric fields, barotropic and internal tides, MSS, etc.) to improve and homogenize the processing to reach the accuracy needed by climate applications.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS This work would not have been possible without the collaboration of all SL-TAC members. The authors sincerely thank them for their daily commitment, their seriousness and their enthusiasm to develop and improve products quality. Authors particularly thank Stephanie Dupuy, Pascal Mambert, and Daoud Jahdou who has ensured the operational maintenance of the DUACS system and has implemented several evolutions with support from Marie-Hélène De Launay. Thanks also to Guillaume Taburet who has coordinated the REP/DT production and has ensured a regular update of these products. Authors also thank Antonio Sánchez-Román who has actively contributed to the qualification of SL-TAC products. Thanks to Françoise Mertz and Cathy Schgounn for the service desk and their contribution to different documentations. Authors also thanks Vinca Rosmorduc for her promotion of SL-TAC products through different trainings. Thanks to Maxime Ballarotta, Oscar Vergara, Pierre Prandi, Quentin Dagneaux, Chloé Durand, Gwenola Maillard, Jean-Francois Legeais who have participated in many studies leading to the evolution and qualification of the SL-TAC products. Finally, authors also thank the different space agencies, CNES, EUMETSAT, ESA, NOAA, ISRO and NSOAS, that have ensured the operational upstream data delivery necessary to the SL-TAC production.

REFERENCES: Ablain M., R. Jugier, L. Zawadki, and N. Taburet.: The TOPEX-A Drift and Impacts on GMSL Timeseries (poster), Ocean Surface Topography Science Team meeting, 2017. Available at: https://meetings.aviso.altimetry.fr/fileadmin/user_upload/ tx_ausyclsseminar/files/Poster_ OSTST17_GMSL_Drift_TOPEX-A. pdf. Last access: 16/01/2020. Ballarotta, M., Ubelmann, C., Pujol, M.-I., Taburet, G., Fournier, F., Legeais, J.-F., Faugere, Y., Delepoulle, A., Chelton, D., Dibarboure, G., and Picot, N., 2019. On the resolutions of ocean altimetry maps, Ocean Sci. 15, 1091-1109, https://doi. org/10.5194/os-15-1091-2019 Ballarotta et al. Dynamic Mapping of Along-Track Ocean Altimetry: Performance from real observations «, in prep for JTech Benkiran M., E. Remy, G. Reffray, 2017. Impact of the assimilation of high-frequency data in a regional model with high resolution. OSTST 2017, Miami, USA.

Di Lorenzo, E., N. Schneider, K.M. Cobb, K. Chhak, P.J.S. Franks, A.J. Miller, J.C. McWilliams, S.J. Bograd, H. Arango, E. Curchister, and others. 2008. North Pacific Gyre Oscillation links ocean climate and ecosystem change. Geophysical Research Letters 35, L08607, https://doi. org/10.1029/2007GL032838. Dufau, C., Orstynowicz, M., Dibarboure, G., Morrow, R., and Le Traon, P.-Y., 2016. Mesoscale Resolution Capability of altimetry: present & future, J. Geophys. Res, 121, 4910–4927, doi:10.1002/2015JC010904. Juza, M.; R. Escudier, A. Pascual, M.-I. Pujol, G. Taburet, Ch. Troupin, B. Mourre, J. Tintoré. Impacts of reprocessed altimetry on the surface circulation and variability of the Western Alboran Gyre. Advances in Space Research, vol. 58 Issue 3 Pages 277–288. 2016. doi:10.1016/j. asr.2016.05.026. Mulet, S., Rio, M.-H., Etienne, H., Artana, C., Cancet, M., Dibarboure, G., Feng, H., Husson, R., Picot, N., Provost, C., and Strub, P. T.: The new CNES-CLS18 Global Mean Dynamic Topography, Ocean Sci. Discuss. [preprint], https://doi.org/10.5194/os2020-117, in review, 2021.

Peltier R. (2004). Global glacial isostasy and the surface of the ice-age Earth: The ICE-5G (VM2) model and GRACE. Annual Review of Earth and Planetary Sciences (32:111-149). Pujol, M.-I., Faugère, Y., Taburet, G., Dupuy, S., Pelloquin, C., Ablain, M., and Picot, N.: DUACS DT2014: the new multi-mission altimeter data set reprocessed over 20 years, Ocean Sci., 12, 1067-1090, doi:10.5194/os-121067-2016, 2016 Qiu, B. and Chen, S.: Variability of the Kuroshio Extension Jet, Recirculation Gyre, and Mesoscale Eddies on Decadal Time Scales, J. Phys. Oceanogr., 35, 2090–2103, 2005. Sánchez-Román A., Ananda Pascual, Marie-Isabelle Pujol, Guillaume Taburet, Marta Marcos, Yannice Faugère (2020). Assessment of DUACS Sentinel-3A altimetry data in the coastal band of the European Seas: comparison with tide gauge measurements. Remote Sensing 2020, 12(23), 3970. https://doi.org/10.3390/ rs12233970

Taburet, G., Sanchez-Roman, A., Ballarotta, M., Pujol, M.-I., Legeais, J.-F., Fournier, F., Faugere, Y., and Dibarboure, G.: DUACS DT-2018: 25 years of reprocessed sea level altimeter products, Ocean Sci. 15, 12071224. https://doi.org/10.5194/ os-15-1207-2019. Vergara, O., Morrow, R., Pujol, I., Dibarboure, G., and Ubelmann, C. (2019). Revised global wavenumber spectra from recent altimeter observations. J. Geophys. Res. doi: 10.1029/2018JC014844 Heslop, E. E., A. Sánchez-Román, A. Pascual, D. Rodríguez, K.A. Reeve, Y. Faugère, M. Raynal. (2017). Sentinel-3A views ocean variability more accurately at finer resolution. Geophysical Research Letters, 44. https:// doi. org/10.1002/2017GL076244. Jousset S., S. Mulet. 2020. New regional Mean Dynamic Topography of Mediterranean and Black Seas from altimetry, gravity and in situ data. OSTST 2020. https://meetings. aviso.altimetry.fr/programs/ abstracts-details.html?tx_ausyclsseminar_pi2%5BobjAbstracte%5D=3034&cHash=bc6fd1239350c0263e0d1d79a6db9b53

FROM A PROTOTYPE TO A FULLY OPERATIONAL IN SITU SERVICE FOR COPERNICUS MARINE

POULIQUEN, S.1, CARVAL, T.1, TAROT, S.1, SZEKELY, T.2, GOURRION, J.2, LIEN, V.3, LINDERS, J.4, TAMM, S.5, DE ALFONSO, M 6, PERIVOLIS, L.7, MARINOVA, V.8, ROTLAN, P.9, TINTORE, J.9, WEDHE, H.3, MADER, J.10, PFEIL, B.11, VERBRUGGE, N.12 6

1 IFREMER, Plouzané, France, 2OceanScope, Brest, France, 3IMR, Bergen, Norway, 4SHMI, Stockholm, Sweden, 5BSH, Hamburg, Germany, Puertos Del Estado, Madrid, Spain, 7HCMR, Athens, Greece, 8IOBAS, Varna, Bulgaria, 9SOCIB, Mallorca, Spain, 10AZTI, San Sebastian, Spain, 11UiB, Bergen, Norway, 12Puertos Del Estado, Madrid, Spain,

70% of platforms provide temperature, 40% salinity, around 10% waves, sea level, currents or biogeochemical parameters.

According to international and European strategies (e.g., GOOS, EuroGOOS ROOSes, EOOS; Le Traon et al., 2019; Tanhua et al., 2019), Copernicus Marine Service In Situ Thematic Assembly Centre (TAC) provides scientifically-robust oceanographic data to support ocean health, climate change and operational ocean services. Collecting and integrating in situ observations from various platforms (e.g., profiling floats, gliders, drifters, saildrones, research vessels, ferry boxes, fixed stations, tides gauges, marine animals, high-frequency radars), In Situ TAC delivers a comprehensive set of physical and biogeochemical ocean parameters at local, regional and global scales, covering polar and coastal regions.

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 In Situ TAC is not operating in situ platforms. Then, close collaboration with operators of ocean observing systems has always been crucial. It implies to define clear interfaces with all actors across GOOS global networks (Argo, OceanSites, DBCP, SOT/SOOP, OceanGliders, GOSHIP), the EuroGOOS ROOSes in regional seas and small data providers with their valuable data.

These multi-platform data are quality-controlled in both delayed mode and near-real time for the global ocean and the six European Seas, composing a reliable multi-scale and multi-variate in situ observational dataset from 1950 to present. In Situ TAC focuses on parameters that are asked and relevant to users as well as necessary for Copernicus Monitoring and Forecasting Centres (MFC) namely: temperature, salinity, sea level, currents, waves, carbon, chlorophyll, fluorescence, oxygen, nutrients (full list of parameters is available here). The integration of in situ data into a unique database and then to the Copernicus Marine catalogue requires the definition and use of standard formats and agreed quality control procedures. Both must be in line with what is done at the international and European levels by the in situ research community and with what is requested by Mercator Ocean International for Copernicus.

In Situ TAC evolved from a data integrator performing data gathering and homogenisation in 2015 to a product provider for both forecasting (NRT) and reanalysis (REP) activities. Main achievements (Figure 1) have been: - implementation of additional quality assessment on data collected around the world, - development in partnership with EuroGOOS, EMODnet and SeaDataNet of closer collaboration with operators of the in situ observing system in Europe, - coordinated actions at the three main integrators level for European observations (3 Memorandum of Understanding signed) of data standardisation, enhancement of Quality Control (QC) procedures and feedback to providers on anomalies detected or data usage, - deep involvement in users support through training and outreach activities organised by Mercator Ocean International with material presently available from In Situ TAC www site.

Handling of in situ observations are also impacted by the measurement techniques and depends on the wide range of sensors that have been used to acquire them, which adds dimensions of complexity. About 7000 multi-parameter platforms from more than 300 observational institutions are integrated every month representing more than 45 000 platforms over 70 years managed by the In Situ TAC. About

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: CMEMS In Situ TAC portfolio development.

1.1 Main developments in the operational In Situ TAC Infrastructure

In addition to improvements done on regional multiparameter NRT products, a particular effort has also been put on REP products by topic: Temperature & Salinity, Waves, Currents, BioGeoChemistry, Carbon. These products are now updated twice a year, with a 6-month temporal extension, one of the 2 updates also being a full reprocessing of product timeseries.

1.1.1 From a prototype to a fully operational service Over the period 2015-2021, in a coordinated effort, production units have significantly improved their procedures and tools thanks to Key Performance Indicators (KPI) displayed on the in situ marine website and to the file format checker useful to monitor the production. In collaboration with the Central Service Desk, users are informed, as soon as possible, of every event that may affect them: incidents, planned service outages and planned improvements of products.

1.1.2 Main technical improvements Over the past six years, In Situ TAC product files have improved radically with enhanced format, higher quality content, wider visibility and FAIR metadata following international rules and Copernicus Marine Service ones. In 2019, files migrated from NetCDF3 to NetCDF4 (greater compression, no more 2Go size limit, richer and FAIR global discovery and variable attributes). Along with this migration, stricter and exhaustive format and content controls were introduced. The file format checker and the file content checker was routinely applied on 700,000 NetCDF files. These routines ensure that requirements from “Copernicus Marine In Situ NetCDF Format Manual’’ and “System Requirements Document” are met. The NetCDF format manual is regularly updated to address specificities from new providers, platforms and metadata. Moreover, there is a continuous action to homogenise data: list of parameters, list of possible attributes, operation procedures.

Additionally, InSitu TAC has also significantly improved the number of platforms and the number of data providers that allowed the integration of more data and then the densifying of the in situ products. This way, historical data gaps (both in space and in time) are being filled as much as possible. This work on data ingestion and integration is possible thanks to In Situ TAC implication in and collaboration with many programmes and projects at European and international levels. For some areas, conditions for data collection are challenging (Arctic Ocean, Black Sea, Southern Ocean…) due to environmental difficulties to set up and maintain observing systems or complicated collaboration with surrounding countries. Despite the efforts already done this is a never-ending journey that we plan to carry on in the future.

#57

In Situ TAC now exposes rich and FAIR data and metadata in NetCDF files, following guidelines and vocabularies maintained by SeaDataNet. Vocabularies are crucial for products homogeneity and consist of: - v ocabulary for physical parameters linked to SeaDataNet P09/P01 parameters, P06 units, - unambiguous data provider identification with European Directory of Marine Organisations codes, - REP and NRT data information, - Exhaustive data type identification, - Improved sea level metadata from tide gauges following EuroGOOS Tide Gauges Task Team recommendations.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

regular users has doubled and continues to increase.

1.1.4 Enhancing the Quality Thanks to the collaborative work, feedback mechanisms within the Product Quality Working Group (PQWG), improved methodology and development of new user-oriented metrics, the In Situ TAC has been able to enhance the product quality. In 2016, the quality of wave data at In Situ TAC was improved through development and implementation of specific NRT Quality Control (QC) procedures. Since 2019, sea level data have been assessed by an automatic QC software, based on GLOSS procedures and implemented at In Situ TAC IBI (Iberia, Biscay and Ireland) and MED (Mediterranean) regions.

1.1.3 Support to users

Concerning temperature and salinity observations, the MinMax method (Gourrion et al., 2020), initially developed for delayed-time QC, has been adapted to NRT production constraints and implemented operationally at In Situ TAC for Argo observations since January 2019. The method has demonstrated its ability to improve the early detection of Argo platforms affected by fast positive salinity drift, a major issue since 2018. Reference fields have been updated operationally in February 2020 and the method will continue to be applied to a larger set of platform types.

During the past six years In Situ TAC has invested itself in a set of activities aiming to enhance the in situ products visibility and, in turn, user engagement. The main activities are: - i nterface for in Situ data discovery, sub setting and visualisation. Initially released in 2017 and in continuous improvement/extension since then, this dashboard is now showcased as Advanced Tool for visualization in Copernicus Marine Service main site. Also, the In Situ dedicated site with shortcuts to several resources (documentation, news, services, training, partners, contact…) is useful for users, -d isplay of KPIs for checking and monitoring in situ NRT and REP data in terms of availability, delay, quality, quantity, etc, -g uided Jupyter Notebook tutorials with code snippets in python illustrating how to find, subset, download and plot programmatically in situ products in the catalogue, -p articipation as experts to 7 regional workshops per year. Since the remote attendance is enabled, the audience has skyrocketed by 325%.

Since 2020, new QC procedures for Oxygen and BGC data have been developed and implemented. For the carbon NRT product, since the release of the automatic QC software QuinCe in 2018, procedures have been refined to improve reliability and to detect a wider variety of issues across multiple input sensors. The validation of drifter’s velocity and temperature measurements relies on provider (AOML (REP) and MeteoFrance (NRT)) quality control. The wind slippage correction delivered in REP drifter product is compared to geostrophic current derived from altimetry. For research vessel REP ADCP observations, each cruise is processed, quality controlled and visually inspected. For NRT velocities from Argo floats, Argo trajectory files are quality controlled according to “Argo quality control manual for CTD and trajectory data”.The EU High Frequency Radar (HFR) Node oversees the validity of HFR data files and the NRT QC in compliance with the EU common Standard for HFR surface current data and metadata. For the quality of the REP products, a comprehensive examination of historical data series available and of the systems’ performance has been established in collaboration with data providers, with dedicated quality reports.

Over the 6-year period, the number of regular users of in situ NRT products increased from less than 20 to nearly 300 a-day. Since the dashboard, KPIs and web release (September 2017) and the reinforcement of contributions to training workshops (January 2019), the number of daily

As an additional activity, In Situ TAC initialised in 2020 an action to review, through a deep analysis, applied NRT QC procedures to evaluate and homogenise methodologies. All reviewed and improved In Situ TAC NRT QC documentations are available at the In Situ TAC dedicated site.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

quality measurements with a vertical and temporal subsetting, has also been provided more specifically to the ocean reanalysis community.

1.1.5 Contribution to OSR and OMI atlas Since the first OSR (2016), In Situ TAC has contributed alone and in collaboration with MFCs and other TACs to several sections. This includes the study and better understanding of ocean circulation variability (e.g., north-Atlantic cold-fresh blob anomaly, deep convection, mesoscale activity, circulation anomaly) and its impacts on the marine ecosystem variability. Another section of the report was dedicated to climate change impacts (e.g., cyclones, extreme waves, decreasing oxygen, water mass changes) on marine ecosystem and ocean circulation to which In Situ TAC contributed too. A specific section proposed developed and implemented operational applications and services (e.g., pollution risk, marine emergency and search-and-rescue, storm forecasts and alerts) where in situ data were used as well. Finally, In Situ TAC has participated in the CMEMS Ocean Monitoring Indicators reporting and on-line atlas. This catalogue gathers scientific indicators of ocean health and climate change (e.g., temperature, salinity, ocean heat content, water mass and heat exchanges), regularly updated to provide operational support for stakeholders and public information.

Finally, 3D gridded (easy to use for some user communities) temperature and salinity products using objective analysis methods were developed, both in real time and delayed time modes. The development enables the extension of timeseries of the REP product with the most recent months of the NRT one, a design adopted specifically to facilitate the production of OMIs-based in situ gridded products.

1.2.2 Current Since 2015, a real effort has been made by In Situ TAC to: - make oceanic currents (UV) datasets more visible in the Copernicus catalogue, - increase the types and number of observations delivered. Outputs from the INCREASE project (CMEMS Service Evolution 2016-2018) paved the way for the integration of High Frequency Radar (HFR) datasets. Moreover, a collaboration with the C-RAID (Copernicus Reprocessing and Access Improvement for Drifter data) European Environment Agency (EEA) initiative has also been led on the period 20192022. Its objective is to clean-up the entire data archive from the past deployed buoys and to reprocess Argos data and positions, focusing on “SVP” (surface) type buoys. Products have finally been archived on the Global Data Assembly Center (GDAC) and distributed in Copernicus Marine Service.

1.2 Main improvement in the product catalogue 1.2.1 Temperature and Salinity The strategy of the In Situ TAC concerning temperature and salinity datasets is built on three complementary axes. First, efforts have been made to improve products quality. New validation methods performed through scientific assessment have been developed and applied on the REP CORA product (global reprocessing T&S product). When such methods had proven their efficiency on the delayed time mode dataset, they were deployed on the NRT one (e.g., the minmax method first deployed in the delayed time mode dataset in 2017 and then successfully deployed in the NRT dataset in 2019, see Gourrion et al., 2020 and Szekely et al., 2020). Since the project started, an internal NRT product and an internal REP product dedicated to assimilation into ocean models have been delivered for which additional validation tests and subsampling of raw datasets were performed. Other validation tests are also exploited to alert the In Situ TAC Production Unit about any anomalies detected.

The evolution of the UV product from 2015 to 2021 in the Copernicus Marine Service catalogue can be summed up as: - in the REP product, users access to surface (when drogue lost) and 15-m depth (when drogue on) drifters velocities (SVP drifters data collected). The quality control and distribution have been ensured by the AOML Data Assembly Centre (DAC) with, in addition, a wind-slippage correction calculated by the In Situ TAC (CLS, Rio 2012). Users also access to HFR total and radial velocities and VM-ADCP (Vessel Mounted Acoustic Doppler Current Profiler) velocities. HFRs provide surface velocities within an integration depth ranging from tens of centimetres to 1-2 meters depending on the operating central frequency. The ADCP data consist of three-dimensional water current velocities over a depth range along a vessel underway trajectory. This has been a great enhancement as the initial REP dataset contained only current velocities measured on research vessels, - in the NRT product, users access to raw drifter velocities at 15-m depth that are pre-processed by the Marine meteorological Centre of Meteo-France (CMM), 3-day filtered drifters velocities, HFR total and radial velocities and velocities from Argo floats. More than 15,000 Argo floats are distributed by the GDAC, which allow calculating deep ocean current from floats drift at parking depth and surface current from float surface drift.

Then, the focus has been to enhance the spatial and temporal coverage of the REP CORA product, with additional data from the EN4 British MetOffice dataset (period covered: 19501990) and to enhance assessment procedures. For poorly sampled areas, additional data have been integrated such as Arctic profiles from the UDASH Arctic and Subarctic database (1980-2015) and the ship-based data instruments extracted from the EMODnet-Chemistry data and the World Ocean Database collections (2020-2021). Moreover, since 2019, EasyCORA, an extraction of CORA with only the best

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

and updated in 2020. These quality control procedures were implemented in 2016 and are based on range levels and spikes and stuck values detection. Applying these procedures in NRT, users benefit from data with an associated quality flag. In April 2017, the NRT product was launched with more than 400 wave platforms. During 2018, an important improvement was performed with networks integration at global level of two new sources: the Global Telecommunication System and the NDBC/USA. Also, a great effort was conducted to integrate coastal stations and to complete historical timeseries since 1980. The same year, the REP WAVE product was published in operations at global level with data validated and visualized, and flagged accordingly, by wave experts. In 2019, In Situ TAC started to work on wave spectra inclusion as requested by users, MFCs and WAVE satellite TAC. It was carried out on the March 2020 release for regional and global NRT products and on the December 2020 release for the WAVE REP product. The Real Time Quality Control for Waves manual was updated to include wave spectra and new tools were developed to visualize spectral information. At the end of Copernicus 1, figures demonstrate an evolution of integration efforts with more than 1500 platforms providing wave integrated parameters and wave spectra available in more than 150 platforms.

1.2.3 Sea Level Sea level is an important variable to monitor the ocean and coastal sea levels measured by tide gauges are a key source of information to assess sea level related hazards along the coast. By 2015, coastal sea level data was delivered in In Situ TAC NRT products, even when sea level was not yet included in the catalogue. In 2016, In Situ TAC started a fruitful collaboration with the EuroGOOS Tide gauges Task Team (TGTT) to agree on best practices and basic metadata about sea level data. This TGTT group published on May 2017 (updated October 2017) a document, in collaboration with Global Sea level Observing System (GLOSS) and In Situ TAC representatives. By 2019, In Situ TAC distributed, by github, a software for NRT quality control of tide gauge data based on GLOSS standards. The software has been implemented in IBI and MED regions; it is run every 15-min and includes spike detection and flagging, stability test, resampling and interpolation of short gaps and computation of filtered hourly values. Byproducts of this procedure include flagged original sampling data (as provided by national centres, typically 5-15 min) and filtered hourly values. Procedures and methods applied follow the best practices described in the last IOC Manual No. 83 and represent an important step to leverage tide gauge data in operational oceanography. In 2020, a new collaboration line has been opened with EuroGOOS TGTT, GLOSS and OceanOPS to work on new unique platform identifiers and mandatory metadata. This will improve interoperability and reliable gaps analysis that should be performed between existing tide gauge data portals.

1.2.5 Oxygen, Chlorophyll and Nutrients The In Situ TAC REP biogeochemical product offers highquality data on Chlorophyll, oxygen, and nutrients (nitrate, silicate and phosphate) collected across the globe from 1993 and up to present date. Measurements include both discrete bottle data and sensor data coming from CTD, ferryboxes, moorings, BGC Argo and gliders.

A great effort has been made by the In Situ TAC to integrate the European sea level measuring stations going from 383 in 2017 to 870 in 2021 operational tide gauges. However, in 2021, only NRT sea level data from European providers is distributed in NRT products. Finally, strong needs are identified such as a REP product of sea level and the integration of international networks at global level. Those are priorities for Copernicus 2.

Nevertheless, the quality control applied on data may be inconsistent between providers and sometimes also unknown to In Situ TAC partners that channel the data up to Copernicus Marine. To make sure that data of high or low quality are flagged accordingly, Copernicus partners developed new automated quality control procedures for REP biogeochemical data to identify questionable data before visual inspection. These procedures greatly enhanced the team’s delayed-mode quality control capabilities and could also be used to improve the NRT QC. The automated QC is parameter-dependent and based on statistical testing. Moreover, tests were applied when possible to check the data against physical constraints.

1.2.4 Wave In 2015, the In Situ TAC started the preparation of Wave products. Waves inclusion in the CMEMS catalogue was carried out not only by the In Situ TAC but also by MFCs. Then, the CMEMS WAVE Working Group, led by Mercator, was created to work on homogenized products at CMEMS level and the In Situ TAC was collaborating actively with this group to adopt standards from bodies such as IODE (International Oceanographic Data and Information Exchange) and CF (Climate and Forecast) international Conventions.

For Chlorophyll a purely statistical approach has been chosen. The world ocean is divided into coastal and pelagic regions and each region is divided into upper and deeper ocean. Then, the 99th percentile for Chlorophyll concentration is computed for each region individually and used as upper boundary for accepting data, i.e., any data point outside the 99th percentile of any given region is visually checked before being flagged as either ‘1 – good’ or ‘4 – bad’.

Additionally, an intense work was done to develop common automatic quality control procedures for NRT data. This work led to the Copernicus In Situ TAC, Real Time Quality Control for WAVES manual, published in 2016 and reviewed

#57

For Oxygen, on the other hand, concentrations are compared with the calculated maximum saturation of oxygen in seawater relative to the atmosphere (allowing for over-saturation in the upper ocean). In addition, there is a regional range test based on statistics within each region separated into 2 layers. Any data points that are above 100% saturation (but allowing for over-saturation in the surface layer) are rejected and flagged as ‘4 – bad’ data. Moreover, any data points that fall outside regional ranges are visually inspected before having its final quality flag decided. Oxygen product is available either with data acquisition unit, or in µmol/L for modelers, or in µmol/kg for oceanic application and monitoring purposes.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

control procedures follow those of SOCAT. Efforts are underway to include other inorganic carbon variables (pH) and sources (EMODnet-chemistry, GOA-ON, FerryBox).

2. STATUS AT THE END OF COPERNICUS MARINE 1 The main goal of the In Situ TAC is to collect in situ data, as much as possible, with the ultimate target of providing to users all possible data aggregated in the products. Succeeding in such a complex task depends on In Situ TAC activity but also on the support and engagement of observation providers. Thus, a clear strategy was followed during the last decade: offering more benefits to attract new providers by showing them that sharing data with In Situ TAC has many advantages. The most obvious is the increase of data visibility when included in the In Situ TAC as Copernicus Marine Service is one of the cutting-edge data services in Europe and one of the main sources of marine data used by the scientific community. It is on this proactive feedback to providers that In Situ TAC nowadays give them useful metrics. These last developments also allow In Situ TAC to become a warning entity about temporal unavailability, incorrect positions and other data problems of the providers’ day-to-day operative.

For Nutrients, a profile test is applied in addition to a statistics-based regional range test. The profile test identifies where the concentration in the surface layer exceeds the concentration at intermediate depths. Then, all profiles that fail the profile test are visually inspected before a final quality flag is applied. Note that the profile test is only applied in pelagic (non-coastal) regions to avoid potential impact of runoff from land.

1.2.6 Carbon In 2019, In Situ TAC started to integrate Inorganic Carbon data and products in the Copernicus Marine Service catalogue as global carbon NRT and REP products. The carbon REP In Situ product contains Surface Ocean CO2 Atlas (SOCAT) and Global Ocean Data Analysis Project (GLODAP) community data that contribute to a total over 30 million observations, covering more than six decades.

GLODAP includes inorganic carbon Essential Ocean Variables (Dissolved Inorganic Carbon, Total Alkalinity and pH), plus salinity, oxygen and nutrients data from the ocean interior collected on research cruises. Extensive quality control and subsequent calibration and adjustments to ensure internal consistency are carried out to generate the product. Observations data synthesis and gridded climatologies from GLODAPv2 (generated in 2016) were the first version included in the CMEMS catalogue.

In addition, a policy of networking is applied to propagate data to other important portals like SeaDataNet or EMODnet. Based on the EuroGOOS ROOSes approach, the In Situ TAC went a step further in this collaboration by establishing 3 Memorandum of Understanding with these main European data integrators. Indeed, the three groups agreed to join their efforts to develop a FAIR (Findable Accessible Interoperable Reusable) European data landscape for in situ observations and to enable an efficient service to a broader range of users. Such progress has been spread in the Atlantic community through the AtlantOS project and is further developed in the ENVRIFAIR H2020 project. Details can be found in the common SeaDataNet, EMODnet and CMEMS White Paper “Ocean FAIR data Services” (Tanhua, Pouliquen et al., 2019). These FAIR principles are now under implementation in several marine European Infrastructure Research Consortium (ERIC) to ease the data flow access with EMODnet and Copernicus Marine Service.

The generation of the carbon NRT product relies on ICOS (Integrated Carbon Observation System) Ocean Thematic Centre efforts. A state-of-the-art online-hosted software, QuinCe, is under development to perform automatic data reduction and quality control. At this stage, QuinCe handles surface fugacity of carbon dioxide and distributes automatically quality-controlled NRT data from ICOS stations with NRT transmission capabilities. The quality

The first main achievement of In Situ TAC was to produce homogeneous data. This was a big challenge considering the multi-source and multi-platform nature of the data and how heterogenous they could be initially. This homogenization process has always been based on up-to-date conventions and standards. Due to the growing quantity of data being integrated, it has become a continuous activity and the resulting reference documents are reviewed regularly.

SOCAT is a synthesis activity of quality-controlled global surface ocean fugacity of carbon dioxide with contributions from more than 100 international marine carbon scientists. A new version of the product (an observations synthesis and four gridded datasets with monthly, annual and decadal averages, plus a 0.25-degree monthly average only for coastal areas) is released annually.

#57

The second achievement was to develop a scientifically assessed version for the delayed mode products. These datasets aim to address user needs to reanalysis purposes and climate research activities. Developments were made to allow their temporal extension every 6 months and to ease the development of Ocean Indicators in partnership with CMEMS partners.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

performs automatic NRT QC on data and a delay mode QC procedure that includes visual inspection and comparison with other data sources. At times, providers even rely on In Situ TAC, delegating the important task of QC and become users of their own data to improve their local data repositories. In the last years, important cross-cutting activities and extra functions have been developed. A Dashboard was created to facilitate data discovery, file downloading and data visualization.

Probably the most recognizable achievement of In Situ TAC, as it offers lots of added-value to the whole in situ data gathered, is the harmonisation of QC procedures applied within all In Situ TAC centres. As a matter of facts, In Situ TAC

Figure 2: In Situ TAC latest 30 days of data (March 2021).

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Furthermore, the uptake of Copernicus Marine Service products by operational users has to be improved and there is a general need to enhance products visibility. Main priorities for the upcoming years lay on: - sustainability of existing in situ production with extension in time and space in partnership with European actors (EuroGOOS and ROOSes, EOOS) and Internationnal ones (GOOS/OCG and IOC) to: - intensify the integration of recently developed and upcoming ocean observing capabilities, - enhance activities focussing on coastal and polar regions, - develop, improve and integrate automatic methodologies to reduce operating costs and to gain increased knowledge, - support the generation of a REP product of coastal sea level data from tide gauges, in coordination with the EuroGOOS Tide Gauge Task Team and GLOSS initiatives, - cooperation with main other data Integrators in Europe (such as SeaDataNet and EMODnet) to jointly enhance in situ service for end users, improve quality control procedures and develop metrics for the uncertainty information, - support to user uptake through efficient contribution to CMEMS training and outreach activities, - enhanced integration of biogeochemical data to widen the view toward ocean health, to assist validation activities and to help boosting BGC modelling and assimilation capabilities.

3. POST 2021 PROSPECTS The main driver for the improvement of the In Situ TAC is to serve the overall need to understand and predict the ocean state and variability in relation to climate, ocean health, real time response and provision of products at the right time and the right place. While the understanding of the open ocean variability has improved significantly in recent years, there is an increased focus on understanding the variability and the underlying processes in coastal and polar seas. A different approach is necessary due to the high complexity and variability in small scales in coastal zones. For this new challenge, multi-platform in situ observations can significantly contribute to Copernicus Marine Service in the next phase. Indeed, today, a wide range of observation capabilities is available and different methodologies need to be integrated into the In Situ TAC products and datasets. There is also a need to develop fit for purpose In Situ products to improve systematic model validation in polar and coastal areas and to enhance data assimilation capabilities. The goal being the improvement of forecasting skills of numerical models.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS

The authors acknowledge the In Situ TAC team members, from the 17 institutes involved, who have contributed to the products, their quality evaluation and their monitoring.

THE OCEAN COLOUR THEMATIC ASSEMBLY CENTER: MULTI-SENSOR AND OLCI GLOBAL AND REGIONAL PRODUCTS SANTOLERI, R.1, BRANDO, V. E.1, COLELLA, S.1, VOLPE, G.1, DI CICCO, A.1, BÖHM, E.1,

CESARINI, C.1, FORNERIS, V.1, GARNESSON, P.2, MANGIN, A.2, CALTON, B.3, NETTING, J.3, KRASEMANN, H.4, HIERONYMI, M.4, D’ALIMONTE, D.5, STELZER, K.6, VAN DER ZANDE, D.7 CNR ISMAR, 2ACRI, 3PML, 4Hereon, 5AEQUORA, 6Brockmann Consult, 7RBINS

legacy OC sensors (Figure 1): SeaWiFS (NASA) and MERIS (ESA) (both no longer operational), MODIS/Aqua from NASA, VIIRS on NPP and NOAA-20 operated by NOAA, and OLCI on Sentinel-3A and 3B operated by EUMETSAT. In 2018-2021, OC data streams shifted from science missions (SeaWiFS, MERIS and MODIS) toward operational missions (two OLCI and two VIIRS sensors). Since mid 2020, OCTAC relies also on the operational high-resolution imagers: MSI on Sentinel-2A and B operated by ESA.

OVERVIEW The Ocean Colour Thematic Assembly Center (OCTAC) of the Copernicus Marine Environment Monitoring Service (CMEMS) provides high-quality core Ocean Colour products for the global ocean and European seas based on multiple Ocean Colour missions. The OCTAC relies on current and

Figure 1: Legacy, current and forthcoming (approved and planned) ocean colour sensors and high-resolution imagers (source CEOS). Red identifies science missions, Blue identifies operational OC missions, brown identifies high-resolution imagers.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Properties (IOPs), and PFTs/PSCs (Phytoplankton Functional Types and Size Classes), as well as the radiometry in itself (GOOS, 2018).

The OCTAC provides users across the scientific and operational oceanography communities, commercial providers focused on the use of marine resources, and public agencies focused on environmental monitoring, with interests in data across oceanic, shelf and coastal waters. Depending on their applications, these users require different spatial resolutions (i.e., ~1 km in ocean, 300 m over the shelf, down to 10’s of metres in coastal waters).

Since 2015, CMEMS has been providing single-sensor as well as multi-sensor products (Table 1) at 1 km resolution for European seas, and at 4 km resolution for the Global Ocean. In 2017-2018, the single-sensor MODIS and VIIRS datasets were retired and the single-sensor Sentinel-3A/ OLCI was introduced. In May 2021, OLCI datasets at 300 m resolution combining Sentinel-3A and B, as well as the Sentinel-2/MSI datasets at 100 m, were added to the catalogue (Figure 2).

To meet these needs, OCTAC provides in a timely and sustained manner a set of Essential Ocean Variables (EOVs) that can be retrieved from Ocean Colour radiometry. These include Chlorophyll-a concentration (CHL), Inherent Optical

Figure 2: Overview of the OCTAC catalogue evolution from 2015 to 2021.

The OCTAC comprises three OC production centres that generate global and regional higher-level combined products, thus bridging the gap between space agencies providing ocean colour data and users needing addedvalue information (not yet available from space agencies). These higher-level combined products provide addedvalue to standard products delivered by the space agencies, in terms of both harmonization and accuracy.

datasets with higher spatial coverage than the singlesensor data streams. Second, regional satellite products provide higher accuracy at basin level than the standard Ocean Colour data. Indeed, the regionalization of processing chains takes into account bio-optical characteristics of each regional sea (Atlantic, Arctic, Baltic, Mediterranean and Black Sea) for production and data validation (e.g., D’Alimonte and Zibordi, 2003; Brewin et al., 2018; Kajiyama et al., 2018; Pitarch et al., 2016; Volpe et al., 2007; 2019). Moreover, blended Chlorophyll products are generated with the appropriate algorithm across the open ocean and coastal waters based on the occurring water types (Hieronymi et al., 2015; Kajiyama et al., 2018; Le Traon et al., 2017; Volpe et al., 2019).

Firstly, observations from multiple missions are processed together to ensure homogenized and inter-calibrated

For each ocean region, the OCTAC delivers two types of products: CHL and OPTICS. CHL includes the phytoplankton

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 AND STATUS AT THE END OF COPERNICUS 1

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

updated, hindcast, ancillary information such as precision orbital data and meteorological fields used for atmospheric correction (Le Traon et al., 2015).

Chlorophyll concentration (Chl-a), the Primary Production (PP), and the Phytoplankton Functional Types. Whereas OPTICS refers to other variables retrieved from ocean colour sensors, and includes: - inherent Optical Properties (IOPs), such as absorption and scattering coefficients, - t he diffuse attenuation coefficient of light at 490 nm (Kd490), - s ecchi depth (transparency of water), - s pectral Remote Sensing Reflectance (Rrs), -p hotosynthetically available radiation (PAR), - c oloured Dissolved Organic Matter (CDOM), and - t he Suspended Particulate Matter (SPM).

Generally, once a year, the full data timeseries undergoes reprocessing to ensure the most recent findings to be consistently applied and back-propagated to all data. These MY timeseries are produced with a consolidated and consistent input dataset and processing software version, resulting in a dataset suitable for long-term analyses and climate studies (Von Schuckman et al., 2017, Sathyendranath et al., 2017 and references therein). Since June 2020, REP timeseries are extended automatically every six months and superseded NRT datasets for those months are removed from the catalogue.

1.1 Processing levels and timeliness

1.2 The OCTAC catalogue evolution 2015-2021

The CMEMS data are produced in near-real time (NRT), delayed time (DT) and as reprocessed (REP) or Multi-Year (MY) data. These are delivered as daily and monthly datasets (Level 3, L3) and as “gap-free” products (Level 4, L4) to overcome cloud cover in subsequent oceanographic analyses (Le Traon et al., 2017; Le Traon et al., 2019).

In 2015, the OCTAC catalogue was composed of 30 OC Products (184 datasets) covering the Global Ocean and European Regional Seas (Table 1). In NRT, all regional products were based on single-sensors (MODIS and VIIRS) and only the global product was based on sensor merging. For the REP, the consistent re-processed multi-sensor timeseries from 1997 to 2012 were based on the ESA OCCCI v2 including SeaWIFS, MODIS and MERIS (Sathyendranath et al., 2019). The L4 “gap-free” products were available only for a few products (GLO, MED and BS) and they were based on Optimal Interpolation and variants of the DINEOF procedure (Beckers and Rixen, 2003).

Within OCTAC, NRT products are operationally produced every day to provide the best estimate of ocean colour variables at the time. These products are generated using available satellite data as well as climatological ancillary data (meteorological and ozone data for atmospheric correction, and predicted attitude and ephemerides for data geolocation). Improved DT products are also produced operationally by reprocessing OC upstream data using

Ocean Region

NRT L3

NRT L4

REP L3

REP L4

Global Ocean

Multi (MODIS+VIIRS) GSM (Maritorena, et al., 2010)

Multi (MODIS+VIIRS)

Multi (OC-CCI 4km): SeaWIFS+MODIS+MERIS ChlOC5, ChlOC4

Arctic Ocean

single M0DIS & VIIRS OC488 (Taberner et al, 2014)

Atlantic Ocean

single M0DIS & VIIRS OC488 (Taberner et al., 2014)

Baltic Sea

single MODIS HZG

Mediterranean Sea

single M0DIS & VIIRS MedOC4 (Volpe et al., 2007)

single M0DIS & VIIRS DINEOF scheme

Multi (OC-CCI 4km) SeaWIFS+MODIS+MERIS MedOC4 (Volpe et al., 2007)

Based on L3 Multi DINEOF scheme

Black Sea

Single M0DIS & VIIRS (Kopelevich et al., 2013)

single M0DIS & VIIRS DINEOF scheme

Multi (OC-CCI 4km) SeaWIFS+MODIS+MERIS

Based on L3 Multi DINEOF scheme

European Sea

Single (MODIS)

Multi (OC-CCI 1km) SeaWIFS+MODIS+MERIS OC488 (Taberner et al., 2014) Multi (MODIS+VIIRS) OI scheme

Multi (OC-CCI 1km) SeaWIFS+MODIS+MERIS OC488 (Taberner et al., 2014) Multi (OC-CCI 4km) SeaWIFS+MODIS+MERIS (Pitarch et al. 2016)

Table 1: Summary of the upstream and processing chains in the OCTAC catalogue at the start of Copernicus 1.

#57

Figure 2 and Table 2 summarize the evolution of the OCTAC catalogue to 2021 focused on: - development and improvements of the NRT and REP multi-sensor processing chains, - inclusion of the Copernicus Sentinel-2 and Sentinel-3 in the single-sensor and multi-sensor datasets, - improvements in the algorithm for Chlorophyll retrieval based on optical characteristics of the basin and round-robin procedures, - development of new datasets on Phytoplankton Functional Groups and community structure and on Primary Production, - i ncrease of the number of L4 “gap free” datasets.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

global ocean colour time-series (Sathyendranath et al., 2017; Sathyendranath et al., 2019). The last version of OC-CCI timeseries is ingested by OCTAC and converted into CMEMS OC format to generate the global OC REP at 4 km resolution. In the REP processing chains, the OC-CCI processor is used to generate consistent timeseries of reflectances of the Arctic, Atlantic and Baltic Seas at 1 km. The same processor is also used in NRT for the Arctic, and Atlantic ocean from 2018. Since 2020, in Mediterranean and Black Seas, the REP processing chain became identical with the CNR NRT multi-sensor processor (Volpe et al., 2019) and thus is no longer based on the OC-CCI L3. Both the NRT and REP processing chains involve the pre-processing of L2 data from space sensors with different wavelengths that are merged over a common set of wavelengths corresponding to the SeaWIFS bandset (Volpe et al., 2019).

1.2.1 Multi-sensor products From 2015 to date, OCTAC members put a great effort to develop and further improve NRT and REP multi-sensor products (Figure 2, Table 2). For Global NRT and REP products, the Copernicus-GlobColour processor used data from different sensors including: SeaWiFS, MODIS Aqua, MODIS Terra, MERIS, VIIRS NPP, VIIRS-JPSS1 OLCI-S3A and S3B (Garnesson et al., 2019).

The use of multiple sensors permitted to significantly increase the spatial coverage of the daily observations. For instance,Figure 3 shows the effect of the merging of two sensors (VIIIRS and MODIS Aqua) and the successive introduction of OLCI in 2019 and then of NOAA VIIIRS 20 in 2020. The number of clear-sky pixels for the Multi product is larger by 20 – 40 % than products from a single-sensor. Alas, the incremental effect was of ~10% for the third sensor and 4% when the fourth sensor was added.

Several OCTAC products are generated taking the advantage of the ESA OC-CCI initiative targeting climate quality consistency with a minimal inter-sensor bias to produce consistent long term multi-sensors (SeaWIFS, MODIS, MERIS, VIIRS and OLCI)

Products

Input data

L3 processor

Chlorophyll Algorithm

L4 method

Global NRT+REP

L2 SeaWiFS (REP), MERIS (REP), MODIS, VIIRS NPP and JPSS1, OLCI S3A and S3B

GlobColour (Garnesson et al 2019)

Blended (Garnesson et al., 2019) OC5 (Gohin et al., 2002) and CI (Hu et al., 2012)

Monthly average Advanced Optimal Interpolation (variant of Saulquin et al., 2010)

Global REP

L1 SeaWiFS, MODIS, MERIS, VIIRS, OLCI

OC-CCI v5 (OC-CCI, 2020)

OC3, OC4, OC5 and CI, depending on pixel water type (OC-CCI 2014)

Monthly average DINEOF

Arctic NRT+REP

L1 SeaWiFS, MODIS, MERIS, VIIRS, OLCI

OC-CCI v5 upgraded to 1 km full resolution

OC5CI developed by PML: Case 1: CI (Hu et al., 2012) Case 2: OC5 (Gohin et al., 2002)

Monthly average

Atlantic NRT+REP

L1 SeaWiFS, MODIS, MERIS, VIIRS, OLCI

OC-CCI v5upgraded to 1 km full resolution

OC5CI developed by PML: Case 1: CI (Hu et al., 2012) Case 2: OC5 (Gohin et al., 2002)

Monthly average

Mediterranean NRT+REP

L2 SeaWiFS (REP), MERIS (REP), MODIS, VIIRS NPP ad JPSS1, OLCI S3A and S3B

CNR MED+BS Processor (Volpe et al 2019)

Blend of Case1 (MedOC: Volpe et al., 2007, 2019) and Case 2 (Ad4: Berthon &. Zibordi, 2004).

Variant of DINEOF (Volpe et al., 2018) & Monthly means

Black Sea NRT+REP

L2 SeaWiFS (REP), MERIS (REP), MODIS, VIIRS NPP ad JPSS1, OLCI S3A and S3B

CNR MED+BS Processor (Volpe et al 2019)

Merging of BS_OC2 and MLP (Kajiyama et al., 2018)

Variant of DINEOF (Volpe et al., 2018) & Monthly means

Baltic NRT

L1 OLCI S3A and S3B

CNR HZG processor

OLCI Neural Network Swarm (Hieronymi et al., 2015)

Only Monthly means

Baltic NRT

SeaWiFS, MODIS, MERIS, VIIRS, OLCI L3 merged Rrs

OC-CCI v4.2 upgraded to 1 km full resolution

MLP ensemble (Brando et al., 2021)

Only Monthly means

Table 2: Overview of processing chains for global ocean and regional seas in 2021.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Increase in basin coverage in the Mediterranean Sea with the NRT/REP multi sensor processor. A) status at 2018 as in Volpe et al., 2019. B) introduction of OLCI S3A in 2019; C) introduction of NOAA VIIIRS 20 in 2019.

1.2.2 Sentinel data streams

Sentinel-2 MSI datasets products are available in the catalogue since May 2021. These high-resolution products are delivered on a daily and monthly basis for coastal waters (20 km strip from the coastline) for all European Seas with a spatial resolution of 100 m (Figure 5). European Seas have very specific optical characteristics and challenges. Diverse atmospheric conditions and fast changing water types in space and time are addressed by combining different algorithms for Atmospheric Correction (Brockmann et al., 2016; Keukelaere et al., 2018; Novoa et al., 2018, Vanhellemont, 2019) and in-water retrieval (Gons et al., 2002; Nechad et al., 2010). The selection and merging of algorithms are based on criteria that characterize the water colour (Lavigne et al. 2021). Delivered parameters are water-leaving reflectances, turbidity (e.g., Figure 5), suspended matter concentration, backscatter and Chlorophyll concentration.

Since 2018, the Sentinel-3A/OLCI datasets have been added to the CMEMS catalogue with a spatial resolution of 1 km (consistent with the other OCTAC datasets) and with the first 11 of the 21 OLCI spectral bands. The OLCI data stream was included in the NRT/DT/REP processing chains of multisensor products for Global Ocean and Regional Seas (Figure 2). In May 2021, OLCI datasets at 300 m resolution were added to the catalogue combining the Sentinel-3A and B missions. This increased the product coverage and resolution of the whole regional seas, as well as the global level data up to 200 km along from the coastline (Figure 4). For the OLCI “single-sensor” product and the 300 meters product, the Global Chlorophyll-a product stored in the EUMETSAT level-2 is used (i.e., OC4Me algorithm). In February 2021, EUMETSAT switched for NRT product to a new algorithm based on a blended OC4Me/CI-Hu approach.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Example of the 300 meters global merged product OLCI-S3A and OLCI-S3B. A) Global coverage for September 15th, 2020, B) focus on Baranof Island, Gulf of Alaska (56°57’05’’N, 134°56’52’’W).

Figure 5: Example of the 100 meters Sentinel-2 monthly Turbidity product based on MSI-S2A and MSI-S2B. Coverage of the European Sea for April 2020 and inset focusing on French and British coastal waters.

#57

1.2.3 Chlorophyll Algorithm improvements

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The merging scheme (Kajiyama et al., 2018) has been designed to use the band-ratio algorithm and the MLP neural net in waters exhibiting lower and higher optical complexity, respectively.

Algorithms improvements for Chlorophyll retrieval were carried out based on optical characteristics of the basin and round-robin procedures. It produced blended Chl-a maps applying appropriate algorithms across the open ocean and coastal waters depending on the occurring water types. The same regional bio-optical algorithms used in NRT production are applied to generate REP Chlorophyll in all basins with the sole exception of the Baltic Sea (Table 2).

In the Baltic Sea, since 2015, NRT products were based only on MODIS-Aqua data, while the Sentinel-3A/OLCI data stream was introduced in 2017. Then, the MODIS-Aqua data-stream was retired in December 2019. The OLCI Neural Network Swarm (ONNS) algorithm is used to retrieve Chlorophyll concentration. In ONNS, the C2RCC atmospherically corrected remote sensing reflectance is classified into 13 optical water types (OWT), then different OWT-optimized neural networks (NNs) are deployed, and in the end, the results of individual NNs are blended according to their OWT fuzzy logic weights (Hieronymi et al., 2017). For REP timeseries, an ensemble MLP algorithm specifically developed for the Baltic Sea was introduced in 2021. Similarly to the Black Sea merging scheme (Kajiyama et al., 2018), the CHL is retrieved by combining results from individual MLPs based on different Rrs spectral subsets, weighting their contribution through the corresponding novelty index (Brando et al., 2021). This ensemble approach substituted the regional recalibration of the OC4v6 with in situ data by Pitarch et al., (2016).

Within the Copernicus-GlobColour processor, the Chlorophyll-a (CHL) multi-sensor daily product merged Chlorophyll-a values are recomputed using an equivalent scheme for each sensor (Garnesson et al., 2019). For oligotrophic waters, the product relied on the CI algorithm (Hu et al., 2012). For mesotrophic and coastal waters, it relied on the OC5 algorithm (Gohin et al., 2002) tuned for each sensor. The blended OC5 and CI was obtained using the same approach as NASA with a transition between concentration from 0.15 to 0.2 mg/m3 to ensure a smooth merging. In the Arctic and Atlantic Seas the regional Chlorophyll algorithm is OC5CCI, a variation of OC5 (Gohin et al., 2002), developed by IFREMER and PML. To this end, an OC5CCI look-up table was generated specifically for application over OC-CCI daily-merged remote sensing reflectances. The resulting OC5CCI algorithm was tested and selected after a calibration exercise and sensibility analysis of the existing algorithms (OC3, OC4, OCI, OC5CI, OC5, OC5CCI) that included a round robin quantitative performance assessment against in situ data.

1.2.4 Phytoplankton type variables and Primary Production New datasets on Phytoplankton type variables were introduced in the OCTAC catalogue from 2019 for the global ocean and all regional seas. Also, in 2020 a Primary Production (PP) dataset was introduced for the global Ocean. Phytoplankton Size Classes (PSCs) and Phytoplankton Functional Types (PFTs) are expressed as Chlorophyll_a concentration (mg m-3). Phytoplankton Size Classes (PSCs) include Micro-phytoplankton (Micro), Nanophytoplankton (Nano) and Pico-phytoplankton (Pico).

In the Mediterranean Sea, the blended Chlorophyll product are based on two algorithms: the MedOC4, an updated version of the regionally parameterized Maximum Band Ratio (Volpe et al., 2007, 2019) for Open Ocean waters (Case I) and the ADOC4 algorithm (D’Alimonte and Zibordi, 2003) for optically complex waters (Case II domain).

The PP is retrieved based on Antoine and Morel (1996) algorithm, using: - ocean colour products (merged Chlorophyll-a, PAR [photosynthetically active radiation], diffuse attenuation coefficient [Kd]), - Sea-Surface Temperature (SST) from OSTIA (SST_ GLO_SST_L4_REP_OBSERVATIONS_010_011), - mixed layer depth climatology, estimated according to the definition from CMEMS (GLOBAL_ANALYSIS_ FORECAST_PHY_001_024).

Since 2020, the determination of the water type accounts specifically for waters with high Chlorophyll concentration due to phytoplankton blooms (e.g., Gulf of Lions) or mixing (e.g., Alborán Sea) that can be erroneously identified as Case II waters. In the Black Sea, the retrieval of the Chl concentration is based on two different regional algorithms: - a band-ratio algorithm based on two wavelengths (490 and 555 nm) (Zibordi et al., 2015), - a Multilayer Perceptron (MLP) neural net based on Rrs values at three wavelengths (490, 510 and 555 nm) that features interpolation capabilities helpful to fit data non-linearities.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Products

PSCs

PFTs **

Algorithm

Global NRT and REP L3 and L4

MICRO, NANO, PICO

DIATO, DINO, GREEN, HAPTO, PROCHLO, PROKAR

Xi et al., (2020, 2021)

Global REP

MICRO, NANO, PICO

Brewin et al., 2015

Arctic REP

MICRO, NANO, PICO

Brewin et al., 2017

Atlantic REP

MICRO, NANO, PICO

DIATO, DINO

Brewin et al., 2017

Mediterranean REP

MICRO, NANO, PICO

CRYPTO, DIATO, DINO, GREEN, HAPTO, PROKAR

Di Cicco et al., 2017

Black Sea REP

MICRO, NANO, PICO

Baltic REP

MICRO, NANO, PICO

Di Cicco et al., in prep. CRYPTO, DIATO, DINO, GREEN, PROKAR

Di Cicco et al., in prep.

Table 3: Overview of Phytoplankton type variables for the global ocean and each of the regional seas in 2021. **Diatoms, Dinophytes (or Dinoflagellates), Cryptophytes, Green algae + Prochlorophytes and Prokaryotes. Haptophytes (or Coccolitophores).

For Mediterranean, Black and Baltic Seas CNR – PSC and PFT datasets are produced applying regional algorithms (Di Cicco et al., 2017; Di Cicco et al., in preparation) to CHL REP datasets of each basin. These models are empirical functions developed following the global abundance approach of Hirata et al., (2011) and are based on the wellknown relationship between phytoplankton types and cell size and the trophic status of the environment). All three datasets provide daily data (REP) at 1 km resolution and include PSCs, while Mediterranean and Baltic datasets include several PFTs (Table 3).

(Brewin et al., 2017) to account for the influence of SST (i.e., combining the method with an ecological approach), and to partition the micro-phytoplankton into diatoms and dinoflagellates. Also, Nano- and Pico- groups were provided. In 2021, the global methodology was extended to the Arctic Sea to derive PSCs at 1 km resolution (Table 3). A Global PFTs and PSCs dataset were added to the CMEMS catalogue by ACRI to be delivered on a monthly and daily basis (both NRT and REP) with 4 km resolution (Table 3). PFTs and PSCs concentrations are estimated with the algorithm of Xi et al., (2020), which was initially implemented using S3A OLCI reflectance in the visible spectrum (bands comprised between 400 and 681 nm) applying the empirical orthogonal function (EOF). The algorithm has been tuned to additionally handle SeaWiFS, MODIS, VIIRS and MERIS sensors, and to consider the SST as an input in addition to spectral remote-sensing reflectance values. In May 2021, an uncertainty field is also provided for each variable by an analytical error propagation accounting for uncertainties of SST and reflectance data, as well as those attributable to parameters of the algorithm (Xi et al., 2021).

A Global PFT dataset (4 km resolution) was provided by PML with the abundance-based method re-tuned to estimate the Chlorophyll concentration of the three PSCs in the global oceans (Brewin et al., 2015) from OC-CCI CHL data and operationalized to use the latest version of the OC-CCI processor. At a regional scale, the development of PFT variables for the Atlantic is based on results of the CMEMS-funded TOSCA (Toward Operational Size-class Chlorophyll Assimilation) project instead. The abundancebased method (Brewin et al., 2015) was implemented

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

coastal applications require specific processing as the upstream processing chain does not provide yet the same product suite as the Ocean Colour sensors. R&D studies are needed to investigate data fusion between the two observational classes to provide multi-resolution products with different resolution depending on location (i.e., ~1 km in the ocean, 300 m over the shelf, down to tens of metres in coastal waters). This may lead to the development of experimental daily-products at 10 m resolution, combining spectral and temporal resolution of VIIRS and OLCI with the spatial resolution of S2/MSI +L8/OLI.

2. POST 2021 PROSPECTS The OCTAC will continue to improve the accuracy at the basin level of the existing EOVs, i.e., CHL, IOPs and PFTs/PSCs, with particular attention to optically complex waters occurring in shelf and coastal zone. To this aim, R&D will need to test and expand the use of semi-analytical models instead of empirical models and foster the use of optical water type-specific algorithms. In addition, the list of products will be expanded incorporating R&D, performed either within the consortium or externally, to include new biogeochemical EOVs related to the Carbon cycle (e.g., POC, PIC, phytoplankton biomass, PSD, regional PP, regional PFTs/PSCs NRT/daily).

Furthermore, dedicated R&D should focus on the preparation for the NASA PACE science mission currently planned for a 2024 launch and for the Copernicus Sentinel 10 CHIME currently planned for a 2026 launch. In 20212024, exploratory studies to exploit hyperspectral capabilities should be carried out using the ASI PRISMA, or DLR ENMAP spaceborne data. This is expected to improve accuracy of retrieved biogeochemical quantities, novel retrieval approaches and new products maps.

Users and agencies are interested in products that cover oceanic, shelf and coastal waters, to monitor for example compliance to EU’s Water and Marine Strategy Framework Directives or coastal hazards and their drivers. However, the higher spatial-resolution data stream needed for

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS

All the work presented in this article would not have been possible without the invaluable contribution made by: Javier Alonso Concha, Michela Sammartino, Marco Bracaglia, Mario Benincasa and Flavio La Padula (CNR); Odile Hembise Fanton d’Andon and Marine Bretagnon (ACRI); Silvia Pardo, Thomas Jackson and Ben Howey (PML); Jenni Attila, Seppo Kaitala and Sampsa Koponen (SYKE); João Felipe Cardoso Dos Santos and Quinten Vanhellemont (RBINS); Martin Böttcher and Carole Lebreton (BC); and Sindy Sterxck (VITO).

REFERENCES: Antoine, D., & Morel, A. (1996). Oceanic primary production: 1. Adaptation of a spectral light-photosynthesis model in view of application to satellite Chlorophyll observations. Global biogeochemical cycles, 10(1), 43-55. Brando, V.E.; Sammartino, M; Colella, S.; Bracaglia, M.; Di Cicco, A; D’Alimonte, D.; Kajiyama, T., Kaitala, S., Attila, J. Phytoplankton bloom dynamics in the Baltic Sea using a consistently reprocessed timeseries of multisensor reflectance and novel Chlorophyll-a retrievals. Remote Sens. 2021, 13(16), 3071 https://doi.org/10.3390/ rs13163071. Brewin, R. J., Sathyendranath, S., Hirata, T., Lavender, S. J., Barciela, R. M., & HardmanMountford, N. J. (2010). A three-component model of phytoplankton size class for the Atlantic Ocean. Ecological Modelling, 221(11), 1472- 1483. Brewin, R.J.W., Sathyendranath, S., Jackson, T., Barlow, R., Brotas, V., Airs, R., Lamont, T. (2015) Influence of light in the mixed-layer on the parameters of a three-component model of phytoplankton size class, Remote Sensing of Environment 168 (2015) 437–450 http:// dx.doi.org/10.1016/j. rse.2015.07.004

Brewin, R.J., Ciavatta, S., Sathyendranath, S., Jackson, T., Tilstone, G., Curran, K., Airs, R.L., Cummings, D., Brotas,V., Organelli, E. and Dall’Olmo, G., (2017). Uncertainty in OceanColor Estimates of Chlorophyll for Phytoplankton Groups. Frontiers in Marine Science, 4, p.104. Brockmann, C., Doerffer, R., Peters, M., Stelzer, K., Embacher, S., & Ruescas, A. (2016): Evolution of the C2RCC neural network for Sentinel-2 and 3 for the retrieval of ocean colour products in normal and extreme optically complex waters, In Proc. of the Living Planet Symposium 2016, ESA SP-740. D’Alimonte, D. and Zibordi, G.: Phytoplankton determination in an optically complex coastal region using a multilayer perceptron neural network, IEEE T. Geosci. Remote, 41, 2861– 2868, https://doi.org/10.1109/ TGRS.2003.817682, 2003. Di Cicco, A., Sammartino, M., Marullo, S. and Santoleri, R., (2017) Regional Empirical Algorithms for an Improved Identification of Phytoplankton Functional Types and Size Classes in the Mediterranean Sea Using Satellite Data. Front. Mar. Sci. 4:126. doi: 10.3389/ fmars.2017.00126

Di Cicco, A., Canuti, E., et al., Regional empirical algorithms for the retrieval of Phytoplankton Functional Types and Size Classes in the Black Sea from in-situ and satellite data (in preparation). Di Cicco, A., Canuti, E., StońEgiert, J., et al., Regional empirical algorithms for the retrieval of Phytoplankton Functional Types and Size Classes in the Baltic Sea from in-situ and satellite data (in preparation) Doerffer, R., and Schiller, H. (2007). The MERIS Case 2 water algorithm. International Journal of Remote Sensing, 28(3-4), 517-535 Garnesson, P., Mangin, A., Fanton d’Andon, O., Demaria, J., and Bretagnon, M.: The CMEMS GlobColour Chlorophyll a product based on satellite observation: multi-sensor merging and flagging strategies, Ocean Sci., 15, 819-830, https:// doi.org/10.5194/os-15-8192019, 2019. Gohin, F., Saulquin, B., OgerJeanneret, H., Lozac’h, L., Lampert, L., Lefebvre, A., Riou, P., and Bruchon, F.: Toward a better assessment of the ecological status of coastal waters using satellite-derived Chlorophyll-a concentrations, Remote Sensing of Environment, 112, 3329-3340, 10.1016/j. rse.2008.02.014, 2008.

Gons, H.J.; Rijkeboer, M.; Ruddick, K.G. (2002). A Chlorophyll-retrieval algorithm for satellite imagery (Medium Resolution Imaging Spectrometer) of inland and coastal waters, J. Plankton Res., v. 24, n. 9, p 947-951. GOOS, 2018. Global Ocean Observing System, Essential Ocean Variables Specification Sheet: Ocean colour https://www.goosocean. org/index.php?option=com_ oe&task=viewDocument Record&docID=19959. Hirata, T., Hardman-Mountford, N. J., Brewin, R. J. W., Aiken, J., Barlow, R., Suzuki, K., Isada, T., Howell, E., Hashioka, T., Noguchi-Aita, M., Yamanaka, Y., 2011. Synoptic relationships between surface Chlorophyll-a and diagnostic pigments specific to phytoplankton functional types. Biogeosciences, 8(2), 311-327. Hieronymi, M., Müller, D., & Doerffer, R. (2017). The OLCI Neural Network Swarm (ONNS): a bio-geo-optical algorithm for open ocean and coastal waters. Frontiers in Marine Science, 4, 140. Hu, C., Z. Lee, and B.A. Franz (2012). Chlorophyll-a algorithms for oligotrophic oceans: A novel approach based on threeband reflectance difference, J. Geophys. Res., 117, C01011, doi:10.1029/2011JC007395.

#57

Kajiyama T., D. D’Alimonte, and G. Zibordi, “Algorithms merging for the determination of Chlorophyll-a concentration in the Black Sea,” IEEE Geoscience and Remote Sensing Letters, 2018. [Online]. Available: https://-www.doi.org /¬10.1109/¬LGRS.2018.2883539 Keukelaere, L. De, S. Sterckx, S. Adriaensen, E. Knaeps, I. Reusen, C. Giardino, M. Bresciani, et al., 2018. “Atmospheric Correction of Landsat-8/OLI and Sentinel-2/ MSI Data Using ICOR Algorithm: Validation for Coastal and Inland Waters.” European Journal of Remote Sensing 51 (1): 525–42. https://doi.org/10.1080/227972 54.2018.1457937. Lavigne, H.; Van Der Zande, D.; Ruddick, K.; Cardoso dos Santos, J.; Gohin, F.; Brotas, V.; Kratzer, S. (2021). Qualitycontrol tests for OC4, OC5 and NIR-red satellite Chlorophyll-a algorithms applied to coastal waters, Remote Sensing of Environment, in press. Nechad, B.; Ruddick, K.G.; Park, Y. (2010). Calibration and validation of a generic multisensor algorithm for mapping of total suspended matter in turbid waters, Remote Sensing of the Environment, v. 114, p. 854-866.

Novoa, S.; Doxaran, D.; Ody, A.; Vanhellement, Q.; Lafon, V.; Lubac, B.; Gernez, P. (2017). Atmospheric corrections and multi-conditional algorithm for multi-sensor remote sensing of suspended particulate matter in low-to-high turbidity levels coastal waters. Remote Sens., v. 9, 61, DOI 10.3390/rs9010061 Le Traon, P. Y., Ali, A., Alvarez Fanjul, E., Aouf, L., Axell, L., Aznar, R., et al., (2017). The copernicus marine environmental monitoring service: main scientific achievements and future prospects. Mercator. Ocn. J. 56, 101. doi: 10.25575/56 Le Traon, P. Y., Reppucci, A., et al., (2019) From Observation to Information and Users: The Copernicus Marine Service Perspective. Front. Mar. Sci., 6, 234. https://doi.org/10.3389/ fmars.2019.00234. Pitarch, J., Volpe, G., Colella, S., Krasemann, H., and Santoleri, R.: Remote sensing of Chlorophyll in the Baltic Sea at basin scale from 1997 to 2012 using merged multi-sensor data, Ocean Sci., 12, 379-389, 10.5194/os-12-379-2016, 2016.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Sathyendranath, S., Brewin, R. J. W., Jackson, T., Mélin, F., and Platt, T.: Ocean-colour products for climate-change studies: What are their ideal characteristics?, Remote Sens. Environ., 203, 125–138, https://doi.org/10.1016/j. rse.2017.04.017, 2017. Sathyendranath, S. et al., (2019). An Ocean-Colour Timeseries for Use in Climate Studies: The Experience of the Ocean-Colour Climate Change Initiative (OC-CCI). Sensors 19 (19), 4285. https://doi. org/10.3390/s19194285. Vanhellemont, Q., (2019). Adaptation of the dark spectrum fitting atmospheric correction for aquatic applications of the Landsat and Sentinel-2 archives. Remote Sensing of Environment 225, 175–192. doi:10.1016/j. rse.2019.03.010. Volpe, G., Santoleri, R., Vellucci, V., Ribera d’Alcalà, M., Marullo, S., and D’Ortenzio, F.: The colour of the Mediterranean Sea: Global versus regional biooptical algorithms evaluation and implication for satellite Chlorophyll estimates, Remote Sens. Environ., 107, 625–638, https://doi.org/10.1016/j. rse.2006.10.017, 2007. Volpe, G., Buongiorno Nardelli, B., Colella, S., Pisano, A., and Santoleri, R.: An Operational Interpolated Ocean Colour Product in the Mediterranean Sea, in: New Frontiers in Operational Oceanography, edited by: Chassignet, E. P., Pascual, A., Tintorè, J., and Verron, J., GODAE OceanView, 227–244, 2018.

Volpe, G., Colella, S., Brando, V. E., Forneris, V., Padula, F. L., Cicco, A. D., ... & Santoleri, R. (2019). Mediterranean ocean colour Level 3 operational multi-sensor processing. Ocean Science, 15(1), 127-146. Xi H., Losa S. N., Mangin A., Soppa M. A., Garnesson P., Demaria J., Liu Y., Hembise Fanton d’Andon O., Bracher A. (2020) Global retrieval of phytoplankton functional types based on empirical orthogonal functions using CMEMS GlobColour merged products and further extension to OLCI data. Remote Sensing of Environment, 240, 111704, https://doi.org/10.1016/j. rse.2020.111704 Xi H., Losa S. N., Mangin A., Garnesson P., Bretagnon M., Demaria J., Soppa A.M., Hembise Fanton d’Andon A., Bracher A., Global Chlorophyll a concentrations of phytoplankton functional types with detailed uncertainty assessment using multi-sensor ocean color and sea surface temperature satellite products, Journal of Geophysical Research – Oceans, 2021, https://doi.org/10.1002/ essoar.10505775.1

THE SEA SURFACE TEMPERATURE THEMATIC ASSEMBLY CENTER

AUTRET, E.1, BUONGIORNO NARDELLI, B.2, GOOD, S.A.3, HØYER, J.4, PAYET, J.M.5, PIOLLÉ, J.F.1, PISANO, A.2, SAUX-PICART, E.5 Ifremer, France - 2CNR, Italy - 3MetOffice, United Kingdom - 4DMI, Denmark - 5MeteoFrance, France

A continuous effort was dedicated to the inclusion of all relevant new sources of data in the processing systems as they became available. The main developments were: - the ingestion of Sea and Land Surface Temperature Radiometer (SLSTR) data acquired by ESA Sentinel3A and Sentinel-3B missions, - the release of more homogeneous and accurate REP products and information on the ocean state with respect to phase 1, by integrating the upstream high quality climatic records by the ESA CCI/C3S initiatives in all multi-year (MY) processing chains.

OVERVIEW Since the beginning of Copernicus Marine Service phase 2, in 2018, the Sea Surface Temperature Thematic Assembly Centre (SST-TAC) has been in charge of the near real-time (NRT) and delayed mode (REP) processing of SST products based (primarily) on satellite observations. During the first phase (2015-2018), this service was operated as the Ocean and Sea Ice Thematic Assembly Centre (OSI TAC). The SSTTAC has produced both single and multi-sensor merged data (L3C/L3S) and interpolated data (L4), delivered as global and regional products. The regional products have been specifically designed for European seas. Ocean Monitoring Indicators (OMIs), needed to provide consistent descriptions of the ocean state over the past decades, have also been developed. The SST-TAC activities have been carried out in close relationship with main satellite data and product providers: EUMETSAT, the OSI SAF, NOAA, REMSS and ESA. Additionally, strong links have been developed with specific programs and initiatives such as: the ESA Climate Change Initiative (CCI), the Copernicus Climate Change Service (C3S), and the Group for High Resolution Sea Surface Temperature (GHRSST), which has provided international coordination among the institutions and agencies involved in satellite SST data production.

These major evolutions are summarized in the following subsections.

1.1 Ocean products (GLOB) The SST-TAC produced a number of global L4 products to serve different user needs. Those produced in NRT were: - a daily foundation SST product (commonly known as the Operational Sea Surface Temperature and Ice Analysis: OSTIA; Donlon et al., 2012; Good et al., 2020), - an hourly diurnal skin SST product (While et al., 2017), - an ensemble of SST analyses produced around the world (Martin et al., 2012).

1. MAIN ACHIEVEMENTS

Complementing the NRT foundation SST product is a MY dataset. In addition, CMEMS disseminated a L4 climate data record representing daily average SST at 20 cm depth from ESA CCI (Merchant et al., 2014; Merchant et al., 2019) and its extension produced by C3S.

The SST-TAC product portfolio has been continuously evolving during the service through: - t he update of existing products, - t he development of new products and datasets, - t he removal of products which exhibited degraded performances and/or were superseded.

The NRT foundation SST production system was the subject of particularly significant changes over the course of CMEMS phase 2. In early 2018, the production system transitioned to use the NEMOVAR data assimilation scheme (Mogenson et al., 2009). The benefit emerged from ability to adapt background error covariance length scales in the region of high SST gradients (Fiedler et al., 2019) (these define for the analysis system how errors in the

#57

first guess SST field correlate spatially, and consequently, how far information from observations should be spread). A further change occurred in 2019, when the background error covariance length scales were adjusted to improve the feature resolution of the analyses (Fiedler et al., 2019). The impact of this change is illustrated in Figure 1.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

reference satellite and in situ data. SLSTR on Sentinel-3B was also added on 2019, along with NOAA20 VIIRS. Finally, following assessment of the impact, in 2020, SLSTR sensors were added to the reference dataset used to correct other satellites’ biases. This reduced the cold bias of the analyses compared to independent near surface Argo data. For example, in June 2020 tests, the global bias was reduced from -0.08°C to -0.04°C (see CMEMS Quality Information Document [QUID] for more details).

Simultaneously, SLSTR data from the Sentinel-3A satellite were included in the analysis. However, it had a limited impact because its data were bias corrected toward a

Figure 1: The gulf stream region as represented in the global L4 foundation SST analyses on 11 and 12 March, respectively before and after the background error covariance length scales adjustment. The visible sharper transitions between warmer and colder water were the result of an upgrade to improve the feature resolution of the product.

To complement the updated NRT foundation SST production system, a new reprocessed foundation SST dataset was generated to supersede an existing dataset based on an old version of the OSTIA system (Donlon et al., 2012). The new dataset was generated using a configuration of the NRT OSTIA system described in Good et al., (2020), and input data included the climate data records from ESA CCI and interim climate data records from C3S. The resulting dataset covers 1981 to 2020, but routine updates were planned to extend the dataset forward in time. It was designed for users of the NRT product who need reliable historical data. The reprocessed foundation SST product was complemented by the ESA CCI and C3S product, which served climate users. Within CMEMS, this has been used to generate ocean monitoring indicators of global average SST, trends and annual anomalies.

1/10° horizontal resolution. This daily product resulted from the merging of several satellite SST level 2 data. Beforehand, data passed a significant number of quality controls and were inter-calibrated through an inter-sensor bias correction procedure. The procedure exploited a median field generated from a set of “best quality” sensors, to provide an estimate of the night time SST based on original SST observations. New sources of data were included in the L3S global product. In 2019, observations from Suomi National Polar-orbiting Partnership (SNPP) Visible/Infrared Imager Radiometer Suite (VIIRS) and GMI were included. In 2020, the coverage was significantly increased with the integration of sea surface temperature data from N20, Sentinel-3A and S3B, GOES16, GOES17, Himawari, MSG Indian Ocean (Figure 2).

To complement global L4 products, SST-TAC provided a multi-sensor level 3 product covering the global ocean at

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 2: Example of the daily L3S Sea Surface Temperature product (SST_GLO_SST_L3S_NRT_OBSERVATIONS_010_010) in December 2020, including satellite observations from Seviri on MSG, VIIRS NPP, VIIRS N20, SLSTRs on Sentinel-3A and S3B, GOES16, GOES17, Himawari, MSG Indian Ocean, AVHRR on Metop-B, AMSRE 2 and GMI.

1.2 Atlantic (European North West Shelf and Iberia Biscay Irish seas) products

Kalman smoother. In 2019, the area was extended to the European North West Shelf and Iberia Biscay Irish seas. This spatial extension enabled the coverage of the Canary upwelling system. In 2020, timeseries were reprocessed from 1982 to 2019 with a new input dataset: ESA CCI data (11 AVHRRs and 3 ATSRs) and C3S data (AVHRRs and SLSTR). Prior to this interpolation, an input data intercalibration step was added.

A new reprocessed foundation SST dataset was generated over the North West Shelves in 2017 and updated in 2018 (NWS). The SST Level 4 analysis product was reconstructed daily from the global Advanced Very High Resolution Radiometer (AVHRR) Pathfinder dataset for the 1982-2014 time period. This includes global AVHRR observations from NOAA polar orbiting satellites. The spatial resolution is 1/25°. The absence of valid SST values was mainly due to clouds. The data used in our reprocessing were the daily files from 1982 to 2014. During the 2015-2017 period, operated data were the real time AVHRR L2 products. The reconstruction of the daily gap-free SST timeseries relied on the interpolation at each point of the timeseries with a

In addition to the reprocessed product, an operational Level 4 product was delivered daily over the European North West Shelf and Iberia Biscay Irish seas with a spatial resolution of 1/50° (Figure 3). The input data have evolved continuously as in the global L3S product. Initially, the product covered the European North West Shelves. The area was then extended to the European North West Shelf and Iberia Biscay Irish seas in 2019.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Example of the daily L4 Sea Surface Temperature product in 2020, including satellite observations from Seviri on MSG, VIIRS NPP, VIIRS N20, SLSTRs on Sentinel-3A and S3B, GOES16, GOES17, Himawari, MSG Indian Ocean, AVHRR on Metop-B, AMSRE 2 and GMI.

1.3 Baltic Sea products

and the Sentinel-3B was included in 2019 together with NOAA 20 SST observations. The inclusion of additional satellite SST products has led to a significant increase in the performance of the operational product over the last 5 years, when compared to in situ observations.

The operational Baltic Sea SST product is a Level 4 gap-free field covering the North Sea and Baltic Sea with a spatial resolution of 1/50°. In 2015, inputs to the operational product were level 2, swath based Metop, NOAA AVHRR SSTs observations supplemented with Geostationary SEVIRI observations. A dynamical and spatially varying bias adjustment scheme was introduced in 2015 when SST products were referenced to each other, resulting in improved performance for SST fields.

A new product was introduced in the Baltic Sea SST portfolio in 2019 as per user requests from the CMEMS Baltic Sea Monitoring Forecasting Center (BAL-MFC). The product was a level 3 supercollated (L3S) field consisting of aggregated and referenced satellite observations that were averaged but not interpolated to fill gaps. The product is now being routinely assimilated in the CMEMS BAL-MFC ocean model runs. An example of a L3S field is shown in Figure 4.

The launch of the Copernicus Sentinel-3A and 3B satellites provided a significant amount of additional SST observations. Sentinel-3A data were included in 2018

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Example of the daily L3S over the Baltic Sea (SST_BAL_SST_L3S_NRT_OBSERVATIONS_010_032).

sensors were added to L3 EUR NRT products: VIIRS, SLSTR-A and SLSTR-B.

A new L4 SST reprocessed product has been produced and made available in late 2020. The product is similar to the operational product in terms of spatial coverage and resolutions and daily fields were available from 1982 to 2019. The L4 reanalysis was created by using infrared satellite products for SST from Copernicus Climate Change Service (C3S), ESA Climate Change Initiative (CCI) and high-resolution sea ice products from the Swedish Meteorological and Hydrological Institute and Finnish Meteorological Institute.

The ingestion of these new sensors has improved the quality of products and, in particular, the mono-sensor L3S super-collated product (covering European Seas with a 1/50° resolution and generated from the merging of various satellite SST L2P data) (Figure 6). The new daily L4 NRT product, based on the super-collated product, has been also impacted by the integration of these sensors (Figure 6). This new product has been an optimally interpolated gap-free (level 4) multi-sensor SST product over European Seas at 1/50° resolution. This product has been built using the EUR L3S product derived from biascorrected European Seas L3C mono-sensor products. This analysis has used the analysis of the previous day as first guess field.

1.4 European sea products During the 2015-2021 period, the European Seas (EUR) production unit focused on the creation of new products and the integration of the new SLSTR sensor from both Sentinel-3A and 3B satellites. Additionally, three new

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 5: SST from the mono-sensor VIIRS NOAA-20 (a), SLSTR-A (b), SLSTR-B (c) on the 1st March 2021.

Figure 6: SST from L3S multi-sensor (AVHRR_METOP_B, VIIRS_NPP, SLSTR-A, SLSTR-B, MODIS_A, MODIS_T, SEVIRI, AMSR2 and VIIRS_ NOAA20) (left), and L4 SST (estimate of the SST sub-skin based on the NADA optimal software) (right), on the 1st March 2021.

1.5 Mediterranean and Black Sea products

During CMEMS phase 2, the NRT production has known significant evolutions: - the update of some upstream L2 SST data, such as the change from meteosat-10 to meteosat-11, - the integration of Sentinel SLSTR-3A/-3B sensors, - the release of two new diurnal products.

Over the Mediterranean Sea (MED) and Black Sea (BS), the SST-TAC provides NRT and REP SST products. The former consists of daily (night-time) merged multi-sensor (L3S) and optimally interpolated (L4) foundation SST fields provided at high (1/16°) and ultra-high (1/100°) spatial resolution, covering the period from 2008 to present (Buongiorno Nardelli et al., 2013). The reprocessed MED and BS datasets complemented the NRT production, providing stable long-term L4 foundation SST data records, currently covering the period 1982-2019 (Pisano et al., 2016). Finally, two (MED/BS) new L4 diurnal subskin SST products have been released in May 2021.

The integration of both SLSTR sensors began with the first data availability, in 2017 and 2019 (SLSTR-3A/-3B, respectively). The integration assessment showed that the best accuracy was achieved when using the dual view of SLSTR-3A as reference sensor while leaving the nadirview of SLSTR-3A/-3B (as well as the dual view of SLSTR3B) to the bias-adjustment procedure. With this sensors’ configuration, SLSTR sensors have been fully operational

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

model data during the central warming hours (Figure 7). Similar results have characterized the Black Sea diurnal product.

since March 2018 and September 2019 (SLSTR-3A/-3B, respectively). Overall, these new sensors improved the accuracy of both MED and BS NRT products and led to a significant decrease in the mean bias and root-meansquare difference (RMSD) to in situ data.

The MED and BS reprocessed products have known two evolutions. These datasets, originally created by using the Pathfinder SST dataset v.5.2 (PFV52), were regenerated in September 2018 by using the updated Pathfinder v.5.3 (PFV53). In April 2020, existing datasets were substituted with two new products based on a reprocessing of climate data records provided by ESA CCI (Merchant et al., 2019) and the C3S. These new MED and BS reprocessed products have provided 38 years (1982-2019) at 1/20° grid, and routine updates are planned to extend datasets forward in time.

The new MED and BS diurnal products will provide daily L4 maps of hourly mean subskin SST at 1/16° grid resolution, by combining infrared satellite data (SEVIRI) and model data from Copernicus MED-MFC and BS-MFC (Marullo et al., 2014). MED and BS diurnal products will cover the period from 1st January 2019 and 1st January 2020 up to NRT, respectively. The reconstructed Mediterranean diurnal cycle has showed a satisfying accuracy when compared to in situ data and has performed better than the CMEMS

Figure 7: Mean diurnal cycle computed for the Mediterranean Sea diurnal optimally interpolated SST product (DOISST, blue line), model (purple line) and drifters’ data (red line) over a matchups’ dataset covering the complete year 2019.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. PRESENT STATUS OF SST-TAC

3. POST 2021 PERSPECTIVES

The main driver of SST-TAC system evolutions has been the improvement of global and regional processing chains to guarantee that CMEMS provides state-of-the-art (in terms of scientific quality) and consistent L3/L4 products for each target area and consolidated set of users. During Copernicus 1, the SST-TAC carried out a critical review of the product portfolio by: -a ddressing the new requirements from users (including requests from CMEMS MFCs), - i mplementing operational improvements from research and development (R&D) activities carried out either within the CMEMS framework or within other projects at European, national and international levels.

Post 2021 priorities are related to the different status of each production line, as regional and global products have been developed following different approaches and configurations, in order to account for each region specifically. Overall, improvements will generally lead to product evolutions in terms of timeliness, effective spatial resolution, output frequency, update cycle and accuracy. The three main evolution lines include: - the modification and tuning of interpolation algorithms to resolve more accurately small scales, the diurnal cycle, or other specific processes (e.g., coastal upwelling), - optimization of the data offer to deliver users a simplified catalogue and improved validation strategies, - the test and ingestion of any new reliable source of data that may improve the quality of the Copernicus Marine Service output products, such as new satellites and sensors. This concerns specifically data from Copernicus Programme satellites, such as the SLSTR data from the future Sentinel-3C and D missions, and high resolution microwave data from CIMR.

Other major improvements are to be noted: - the homogenization of the upstream data input to the MY REP product (now coming from ESA-CCI and C3S), - the provision of non-interpolated merged multisensor data covering all regions (L3S, to be used for data assimilation), - the development of new interpolated products resolving the diurnal cycle.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

REFERENCES Buongiorno Nardelli, B., Tronconi, C., Pisano, A., Santoleri, R. (2013). High and Ultra-High resolution processing of satellite Sea Surface Temperature data over Southern European Seas in the framework of MyOcean project. Remote Sensing of Environment, 129, 1-16. Donlon, C.J., Martin, M., Stark, J., Roberts-Jones, J., Fiedler, E., Wimmer, W., The Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) system, Remote Sensing of Environment, Volume 116, 2012, Pages 140-158, ISSN 00344257, https://doi.org/10.1016/j. rse.2010.10.017. Fiedler, EK, Mao, C, Good, SA, Waters, J, Martin, MJ. Improvements to feature resolution in the OSTIA sea surface temperature analysis using the NEMOVAR assimilation scheme. Q J R Meteorol Soc. 2019; 145: 3609– 3625. https:// doi.org/10.1002/qj.3644.

Good, S.; Fiedler, E.; Mao, C.; Martin, M.J.; Maycock, A.; Reid, R.; Roberts-Jones, J.; Searle, T.; Waters, J.; While, J.; Worsfold, M. The Current Configuration of the OSTIA System for Operational Production of Foundation Sea Surface Temperature and Ice Concentration Analyses. Remote Sens. 2020, 12, 720. https://doi. org/10.3390/rs12040720. Martin, M., Dash, P., Ignatov, A., Banzon, V., Beggs, H., Brasnett, B., Cayula, J.-F., Cummings, J., Donlon, C., Gentemann, C., Grumbine, R., Ishizaki, S., Maturi, E., Reynolds, R.W., Roberts-Jones, J., Group for High Resolution Sea Surface temperature (GHRSST) analysis fields Intercomparisons. Part 1: A GHRSST multi-product ensemble (GMPE), Deep Sea Research Part II: Topical Studies in Oceanography, Volumes 77–80, 2012, Pages 21-30, ISSN 0967-0645, https://doi. org/10.1016/j.dsr2.2012.04.013.

Marullo, S., Santoleri, R., Ciani, D., Le Borgne, P., Péré, S., Pinardi, N., ... Nardone, G. (2014). Combining model and geostationary satellite data to reconstruct hourly SST field over the Mediterranean Sea. Remote sensing of environment, 146, 11-23. Merchant, C.J., Embury, O., Roberts-Jones, J., Fiedler, E., Bulgin, C.E., Corlett, G.K., Good, S., McLaren, A., Rayner, N., Morak-Bozzo, S. and Donlon, C. (2014), Sea surface temperature datasets for climate applications from Phase 1 of the European Space Agency Climate Change Initiative (SST CCI). Geosci. Data J., 1: 179-191. https://doi. org/10.1002/gdj3.20 Merchant, C.J., Embury, O., Bulgin, C.E. et al. Satellitebased time-series of seasurface temperature since 1981 for climate applications. Sci Data 6, 223 (2019). https:// doi.org/10.1038/s41597-0190236-x.

Mogensen, K.; AlonsoBalmaseda, M.; Weaver, A.; Martin, M.; Vidard, A. NEMOVAR: A variational data assimilation system for the NEMO ocean model. ECMWF Newsl. 2009, 120, 17–21. Pisano, A., Buongiorno Nardelli, B., Tronconi, C., Santoleri, R. (2016). The new Mediterranean optimally interpolated pathfinder AVHRR SST Dataset (1982–2012). Remote Sensing of Environment, 176, 107-116. While, J., Mao, C., Martin, M.J., Roberts-Jones, J., Sykes, P.A., Good, S.A. and McLaren, A.J. (2017), An operational analysis system for the global diurnal cycle of sea surface temperature: implementation and validation. Q.J.R. Meteorol. Soc., 143: 1787-1803. https:// doi.org/10.1002/qj.3036.

THE SEA ICE THEMATIC ASSEMBLY CENTER

GIRARD-ARDHUIN, F.5, BRANDT-KREINER, M.2, BUUS-HINKLER, J.2, DINESSEN, F.6, FLEMING, A.1, HØYER, J.2, KARVONEN, J.4, SALDO, R.3 British Antarctic Survey (BAS), United Kingdom - 2Danish Meteorological Institute (DMI), Denmark Technical University of Denmark (DTU), Denmark - 4Finnish Meteorological Institute (FMI), Finland 5 French Institute for Ocean Science (Ifremer), Brest, France - 6Norwegian Meteorological Institute (MET), Norway 1

(at that time MyOcean) iceberg number-density product became operational on 2 April, 2010 and was based on SAR imagery from Canadian Radarsat satellites.

1. OVERVIEW

However, when the Sentinel-1 program began (with the launches of Sentinel-1A in late 2014 and Sentinel-1B in 2016) much more SAR imagery became available. This was a game changer and led to a dramatic increase in the number of available iceberg products. In particular by the end of 2016 when both satellites were in orbit and fully operational.

The Copernicus Marine Service Sea Ice Thematic Assembly Centre (Sea Ice TAC) provides elaborated operational (Level 3 and 4) observational multi-mission sea ice products derived from upstream satellite earth observation (L2) data. Sea ice products range from operational products delivered in Near Real Time (NRT) to timeseries utilizing satellites observations since 1979.

Iceberg number densities are provided in a gridded format at 10 km spatial resolution. Each grid cell is associated with the number of icebergs detected within its area.

During the period 2015-2021, the Sea Ice TAC has expanded its portfolio to include: - individual icebergs detection, - NRT and multiyear (MY) sea ice thickness in the Arctic, - sea ice type based on Sentinel-1 Synthetic Aperture Radar (SAR) data, - automated products such as sea ice concentration in the Arctic and Baltic.

To meet different user needs the iceberg product-line today consists of six individual datasets: 1. single scenes EW (Extra Wide Swath), 2. single scenes IW (Interferometric Wide Swath), 3. four-day mosaics EW, 4. seven-day mosaics IW, 5. four-day EW mosaics with individual iceberg positions, 6. seven-day IW mosaics with individual iceberg positions.

The spatial resolution of several products has increased, as well as the coverage. Timeseries for MY products and Ocean Monitoring Indicators have been continuously extended as near as possible up to the present. Sentinel-2 and Sentinel-3 have been incorporated in the processing chains for ice charts services and used both in the production and as part of the validation.

The last two datasets are the most recently developed and came into service in July 2020. They are available in ESRI Shape file format, which makes them adequate for use and display in Geographic Information System (GIS) and are often used by national ice services and sometimes by sea navigators directly onboard their vessels. The GIS allows users to visualize their own data as a map.

This article presents the main achievements of the Sea Ice TAC, from 2015 to 2021, and perspectives for the next years. Since products are based on several different algorithms, sensors, and even satellites, these points are presented separately for each product.

Iceberg detection in radar imagery is sometimes connected with significant uncertainty due to high winds (that leads to a rough sea state) and other factors affecting the signal to noise ratio. Thus, datasets 5 and 6 hold information on detection quality (iceberg score and detection significance) for each individual iceberg.

1.1 Iceberg product Iceberg products by DMI are based on target-detection in SAR satellite imagery. The first version of the Copernicus

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

but netCDF files contain statistical layers with anomalies indicating whether observations are normal or unusual (i.e., more or less extreme counts).

Furthermore, size parameters (iceberg length, iceberg area and more) can be obtained from attribute tables in the shape file. The number density products (datasets 1 to 4) naturally do not contain information on individual icebergs

Figure 1: icebergs detected in Sentinel-1 EW mode during 2020-12-28 to 2021-01-01 with zoomed view also showing iceberg number densities.

a subjective analysis of the latest available observations from: satellite imagery, weather and oceanographic information, visual observations from ships. The main satellite source used is the SAR satellite imagery, due to its high resolution, all-weather day and night ability.

During the coming years more and more satellite SAR imagery will become available. To assimilate new data, new and improved iceberg detection techniques will be developed. One of the most early next steps to be implemented is the automation of ships removal (falsely identified as icebergs). This will be based on retrieval and matchup with AIS ship tracking data.

With the start of Sentinel-1A data in 2015 and Sentinel1B in 2016 there has been a significant improvement in the temporal and spatial coverage of SAR. Sea ice analysts leverage both Sentinel-1 data in EW and IW mode, as well as SAR imagery from the Copernicus Contributing Missions (CCM); CosmoSkyMed, TerraSar-X and Radarsat-2. Analysts also use optical/IR data from Sentinel-2 and Sentinel-3, when cloud free and timely available and AMSR2 microwave radiometer.

1.2 The ice chart products The Sea Ice TAC ice chart products are based on ice charts from the national ice services at MET Norway, DMI and FMI covering European regions of the Arctic and the Baltic. Ice charts are produced by skilled sea ice analysts based on

#57

The overall increase in satellite imagery available to the Sea Ice TAC has improved the quality of the product due to an improved coverage of high spatial resolution SAR data. This has also led to an increased number of Greenland ice charts being produced over the period of the Copernicus-1

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

phase: from 23 on average per month in 2015 to 59 in 2020. Also, new parameters in the Greenland ice charts were introduced: ice stage-of-development (ice types) and ice floe sizes (which have been included in Copernicus Marine Service since late 2020).

Figure 2: Example ice charts from national ice service at MET, DMI and FMI, respectively.

1.4 Antarctic ice edge

For all ice services, the vast amount of new Copernicus satellite imagery has been a driver for developing new methods, based on deep learning, for retrieval of sea ice information from SAR imagery and complementary satellite data. At the end of Copernicus phase-1 these automated sea-ice information products became inputs to the DMI ice analysts and thus made the ice charting process more effective. The development of a deep learning methodology for automated sea ice mapping is an upside from the close collaboration within Sea Ice TAC partners working on this field: MET, FMI, NERSC and DTU. The development will continue in the coming years, exploiting and preparing to use the next generation of Copernicus Sentinel satellites; the Copernicus Imaging Microwave Radiometer (CIMR) and Rose-L missions.

Since 2015, the BAS gridded ice edge position information has been derived by manual interpretation of SAR data, primarily utilising Sentinel-1 imagery. The product provides a high-resolution ice edge to aid maritime users in the most heavily trafficked parts of the Antarctic Peninsula and Weddell Sea. It is used to validate the OSI SAF Southern Ocean sea ice classification product following a methodology established as part of the OSI SAF CDOP Visiting Scientist scheme. The product has evolved since 2015 to utilise both Radarsat2, Sentinel-1a and Sentinel-1b image data. The area of data coverage has extended to include the Bellingshausen and Amundsen Seas, providing greater coverage of validation data. The automatic NRT validation of the OSI SAF product has been fully implemented and is accessible at Polarview.

1.3 Sentinel-1 L3-product In support of users of Sentinel-1 data a new Sentinel-1 product was introduced in Copernicus Marine Service in 2020. This product contains separate Sentinel-1 datasets in GeoTiFF format. Datasets include sigma0 calibrated and thermal noise corrected Sentinel-1 Extra Wide dual polarization (HH/HV) data, a land mask and information of the local incident angle. Several individual datasets are delivered daily.

Continued evolution of Southern Ocean products is driven by both: - the need for high quality hemispheric coverage of sea ice data to drive global forecast models, - maritime users needs to support safe operations as mandated by IMO Polar Code.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ice type (stage of development) product from Sentinel-1 SAR data using a convolutional neural network (CNN) (Boulze et al., 2020). It provides classification of SAR data into four different types: open water, young ice, first-year ice, multi-year ice. The CNN was trained on manual ice charts prepared by the National Ice Center NOAA. CNN also provides probability of classification, which can be used as a measure of uncertainty. The daily NRT production was included in the MET-Norway operating system and made available to the Copernicus Marine Service as a daily mosaic product covering the European part of the Arctic.

These ongoing requirements will underpin the growing interest for these products and the need to expand the area covered. The availability of more Sentinel-1 imagery means new automatic methods to derive ice edge position information need to be developed, supported by current widespread adoption of Artificial Intelligence (AI) and Machine Learning (ML) methods.

1.5 Arctic Sentinel-1 ice type The Nansen Environmental and Remote Sensing Center (NERSC) developed a new high-resolution automatic sea

Figure 3: Example of Sentinel-1 sea ice type product from 13. January 2021.

1.6 Arctic multisensor-based sea ice concentration product

Each dataset is first classified as a SIC product before they are merged into one product using a variational merging process. The SAR data are first classified into ice/water at a 40m resolution. After the separation, the concentration is calculated from the surface area of ice within a 1x1 km area. The AMSR2 SIC processing is based on the OSISAF / ESA CCI+ Hybrid Dynamic (OSHD) algorithm which provides a spatial resolution of ~5 km. Before merging, the AMSR2 SIC product is subsampled into 1x1 km resolution to match the SAR SIC. An example of the final product is shown below.

MET Norway has developed an automatic multisensor sea ice concentration (SIC) product for the Arctic which is (by the end of Copernicus-1) provided as a daily SIC product. The new product has estimated SIC based on a combination of JAXA’s AMSR2 microwave radiometer and C-band dualpolarized (HH/HV polarization combination) SAR data from Sentinel-1.

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: MET-multisensor SIC estimate from 12 January 2021.

1.7 Baltic SAR-based sea ice concentration, thickness, drift

Another new automated Baltic Sea product for the winter 2020-2021 is the sea ice thickness (SIT) based on Sentinel-1 IW mode dual-polarized (VV/VH) SAR data. This product is produced every time after a Sentinel-1 IW mode image over Baltic Sea is received at FMI (downloaded from ESA’s cophub or a dedicated ftp server). The resolution of the product is 500 m and the covered area corresponds to the received SAR area. The algorithm consists of SAR segmentation and SAR texture feature extraction. The FMI digitized ice charts are used as a background field as in the Sentinel-1 EW mode ice thickness product (Karvonen et al., 2003).

The automatic Baltic SIC is provided in CMEMS as a daily mosaic covering the Baltic Sea since December 2020. The product resolution is 500 m and each grid cell contains the most recent value at hand, defined by the most recent SAR measurement. The SIC estimates are based on a combination of JAXA’s AMSR2 microwave radiometer and C-band dual-polarized (HH/HV polarization combination) SAR data from ESA’s Sentinel-1 and CSA’s Radarsat-2. The algorithm utilizes a multilayer perceptron neural network documented in detail in (Karvonen 2017).

Figure 5: Sentinel-1 IW mode Sea Ice Thickness estimate, covering a large part of the Gulf of Bothnia on 10 February at 16:06:24 UTC.

1.7 Global NRT sea ice drift

The Baltic Sea ice drift (SID) estimation algorithm (Karvonen 2012) was replaced in December 2021 by a new algorithm utilizing feature matching with the ORB algorithm (Rublee et al., 2011) in low resolution and optical flow (Lucas and Kanade, 1981). The first phase included thorough tests to ensure stability and quality, and then refiningthe drift estimation using pattern matching by optical flow. The new ice drift algorithm has been validated by using the winter 2017-2018 SAR data (Radarsat-2 ScanSAR wide mode HH/ HV and Sentinel-1 GRDM EW HH/HV) against drift from ice drifter buoys.

In 2015, the evolution of the NRT SID product was mostly driven by the new Sentinel 1A SAR platform, following a rather long period of sparse data from Radarsat2 after the demise of Envisat ASAR instrument in 2012. At the time, it consisted of NRT high accuracy high detail north or south pole sea ice covering measurement patches. The steadily increasing data volume gave a much improved coverage both spatially and temporally. This is crucial, given that the discrete tracking of sea ice requires spatial coverage of the same approximate area, with a temporal difference, in order to capture the displacement before the observed surface of the sea ice decorrelates significantly.

The daily Baltic Sea SIT mosaic product merges the most recent Sentinel-1 EW HH/HV and Radarsat-2 ScanSAR wide HH/HV mode SAR Baltic Sea SIT products into daily mosaics covering the whole Baltic Sea, with the most recent SIT data available at each grid cell. The product has been available since January 2016.

#57

With the introduction of Sentinel-1B satellite in mid2016, the combined data volume from the two active Sentinel-1 SAR instruments enabled the introduction of aggregatable daily composite products (mosaics) on fixed grids, as opposed to the low latency high detail NRT product, which was complicated to handle for the general user. The composite product combines the smaller NRT sea ice drift data patches over a 24 hour period into a manageable fixed grid.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

1.9 Arctic NRT Sea Ice Thickness based on merged Cryosat2 and SMOS data The NRT sea ice thickness (SIT) product provides a weekly mean SIT over the Arctic, and is a product based on merged SMOS and CryoSat2 data. The SMOS mission provides L-band observations. Adjacently, the ice thickness-dependency of brightness temperature enables to estimate of the sea-ice thickness for thin ice regimes. CryoSat2 uses radar altimetry to measure the height of the ice surface above the water level, which can be converted into sea ice thickness. As SMOS loses sensitivity near 1 meter ice thickness and the SIRAL altimeter of CryoSat2 at below 1 m (Wang et al., 2016), in the merged SIT the two methods complement each other.

At the end of 2019, the composite product was enhanced with derived vector flow divergence, vorticity, shear layers, thus including information for the user describing the sea ice dynamics in more detail than raw measurements. Given the derivative nature of the vector flow, it quickly became obvious that even though the produced drift values were reliable, even small noise contributions had a significant influence on divergence values included in the 24h composite product. This resulted in the introduction of an intermediate step where the accuracy of each NRT drift measurement was recalculated using data at higher spatial resolution going from the original 300 m image data down to 100 m, thereby much improving the quality of divergence values of the composite.

The merging algorithm, developed in Alfred Wegener Institute (Ricker et al., 2017), leverages daily Soil Moisture and Ocean Salinity (SMOS) observations and CryoSat2 measurements of one week to produce SIT fields in a 25 km EASE2 grid. The sensors have significantly different swath widths, surface resolutions and revisit times. To merge the satellite data with different update rates of thickness observations, the algorithm uses an optimal interpolation (OI) scheme. SMOS data are rejected over multiyear ice and when uncertainties are more than 1 m.

1.8 Arctic multiyear sea ice drift In 2015, Ifremer products were the Arctic sea ice drift at low resolution for the period from 1992 up to the present, in the ice season from October until April. This product is inferred from the timeseries data available at CERSAT/Ifremer. SSM/I radiometer data is merged with several scatterometer sensors such as QuikSCAT/ASCAT-A/ASCAT-B, and processed at 3- and 6-day lags. This product was delivered once a year as a long-term reanalysis product.

Due to poorly known properties of melting ice and snow the SIT retrieval is only performed during the freezing season between October and April. FMI has produced the weekly merged SIT product for CMEMS since April 2020. Since 2021, the update frequency of thickness fields has been increased, and a rolling weekly mean SIT is provided daily.

1.10 Arctic Sea and Ice Surface Temperature

One evolution of the product was to add the months of September and May to the estimate of sea ice drift. These months are not easy to handle because it is the time for freeze and melt of sea ice in the Arctic. Microwaves sensors are indeed very sensitive to these processes. The merging of radiometer and scatterometer data is in particular interesting to use for these months because it enables having at least 80% more drift vectors on the sea ice cover.

The operational Arctic sea and sea ice surface temperature (SST/IST) product is a Level 4 (L4) gap-free field covering surface temperatures of the sea ice, the marginal ice zone and the ocean north of 58° northern latitude. Temperatures have a spatial resolution of 1/20°. In 2015, the input to the operational product consisted of Level 2, swath-based Metop AVHRR SST/IST observations from the OSI SAF project, supplemented with operational NOAA SST products. A dynamical and spatially varying bias adjustment scheme was introduced in 2017 where the SST products were referenced to each other, resulting in improved performance for SST fields.

Ifremer developed an algorithm to estimate sea ice drift at medium resolution from the AMSR series. This timeseries (2002-present) is provided at 2-, 3- and 6-day lags from October until April. The benefit of the resolution enables a higher angle resolution, and is useful in particular for high magnitude drift (Fram strait for example) thanks to the detection at 2 day-lag.

The launch of Sentinel 3-A and B provided a significant amount of additional SST observations. The Sentinel 3A data were included in April, 2019 and the Sentinel 3B was included in December, 2019 together with NOAA 20 SST observations and a new SST/IST product from the NPP_ VIIRS satellite.

Sea ice drift in the Arctic at low and medium resolution are now processed, qualified and provided monthly to CMEMS. New sensors such as ASCAT-C and CFOSAT have been tested and will be integrated to continue the timeseries.

#57

The inclusion of the additional satellite surface temperature products and the dynamical adjustment of SST products has led to a significant increase in the performance of the operational product over the last 5 years, when compared against in situ observations.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

free data set enables the investigation of both general temperature tendencies for the whole covered area and regional differences in temperature changes. Today’s data set only covers the Arctic but developed algorithms can also be applied to the Antarctic.

A new L4 SST/IST reprocessed product has been produced and released in 2021. The product is similar to the operational product in terms of spatial coverage and resolutions and daily fields are available from1982 to 2019.

1.11 Ocean Monitoring Indicators The Arctic Sea Ice Extent was introduced in 2018, as an Ocean Monitoring Indicator (OMI) covering the period from 1979 up to the present. Obtaining knowledge about sea ice cover changes is essential for monitoring the health of the Earth as sea ice is one of the most highly sensitive natural environments. In 2019, the Antarctic equivalent OMI was introduced, the Antarctic Sea Ice Extent, covering the same period as in the Arctic. The OMIs show trends in sea ice extent in the Arctic and the Antarctic, and are highly valued by climate researchers and policy makers alike.

The reprocessed L4 was created by using infrared satellite products for surface temperatures from Copernicus Climate Change Service (C3S), ESA Climate Change Initiative (CCI) and DMIs own products, which have been validated against each other, as well as in situ observations available. The data set contains a consistent climate indicator as it consists of both sea and sea ice temperatures, which can be used to analyze recent situations and trends. This gap-

ACKNOWLEDGMENTS We would like to acknowledge all contributing SI TAC team members here: Andrew Fleming (BAS), Matilde Brandt-Kreiner, Jørgen Buus-Hinkler, Jacob Høyer and Wiebke Kolbe (DMI), , Roberto Saldo (DTU), Juha Karvonen and Jaakko Seppänen (FMI), Fanny Girard-Ardhuin and Cedric Prevost (IFREMER), Frode Dinessen, Thomas Lavergne, Signe Aabøe and Cecilie Wettre (MET), Anton Korosov and Mohamed Babiker (NERSC).

REFERENCES J. Karvonen, Baltic Sea Ice Concentration Estimation Using SENTINEL-1 SAR and AMSR2 Microwave Radiometer Data, IEEE Transactions on Geoscience and Remote Sensing, DOI: 10.1109/TGRS.2017.2655567, 2017. J. Karvonen, M. Similä, I. Heiler, Ice Thickness Estimation Using SAR Data and Ice Thickness History, Proceedings of the 2003 IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2003), vol. I, pp. 74-76, 2003.

B. D. Lucas, T. Kanade, An iterative image registration technique with an application to stereo vision. Proceedings of Imaging Understanding Workshop, pp. 121--130, 1981. http://cseweb.ucsd. edu/classes/sp02/cse252/ lucaskanade81.pdf.

J. Karvonen, Operational SARbased sea ice drift monitoring over the Baltic Sea, Ocean Science, v. 8, pp. 473–483, 2012, https://doi.org/10.5194/os-8473-2012. E. Rublee, V. Rabaud, K. Konolige, G. Bradski, ORB: An efficient alternative to SIFT or SURF, Proc. 2011 International Conference on Computer Vision, 2011, DOI: 10.1109/ ICCV.2011.6126544.

R. Ricker, S. Hendricks, L. Kaleschke, X. Tian-Kunze, J. King and C. Haas, A weekly Arctic sea-ice thickness data record from merged CryoSat-2 and SMOS satellite data, The Cryosphere, 11, 1607-1623, https://doi.org/10.5194/tc-111607-2017, 2017.

X. Wang, J. Key, R. Kwok, and J. Zhang, Comparison of Arctic Sea Ice Thickness from Satellites, Aircraft, and PIOMAS Data, Remote Sensing, 8, 713, https://doi.org/10.3390/ rs8090713, 2016. Boulze, H. Korosov, A. Brajard, J. Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks. Remote Sens. 2020, 12, 2165. https://doi.org/10.3390/ rs12132165.

THE WAVE THEMATIC ASSEMBLY CENTER

CHARLES, E.1, HUSSON, R.2, DODET, G.3, MOUCHE, A.3 AND WAVE-TAC TEAM Collecte Localisation Satellites (CLS), Ramonville Saint-Agne, France 2Collecte Localisation Satellites (CLS), Brest, France 3 Univ Brest, CNRS, Ifremer, IRD, Laboratoire d’Océanographie Physique et Spatiale (LOPS), IUEM, Plouzané, France.

creation of a dedicated Wave Thematic Assembly Center (WAVE-TAC) in 2018, a wider range of products and the consolidation of the satellite constellation.

OVERVIEW

The range of wave products increased significantly by integrating for the first time Level-3 spectral (SPC) products derived from Sentinel-1 Synthetic Aperture Radar (SAR) missions. This product provides not only the significant wave height (SWH) but also the wave period and direction for different observed swell systems. Additionally, waves backward and forward propagation in space/time along

The dissemination of the first Copernicus Marine Service operational wave products started mid-2017 within the Sea Level Thematic Assembly Centre (SL-TAC) using observations derived from Jason-3 and Sentinel-3A altimeter missions. During the following four years, the wave satellite service has considerably evolved with: the

Figure 1: Along-track SWH over 48 hours for a 2-satellite constellation (Jason3 and Sentinel-3A) and a 7-satellite constellation (Jason-3, Sentinel-3A, SARAL/AltiKa, CryoSat-2, Sentinel-3B, CFOSAT and HaiYang-2B).

#57

great circles is included. Also, a Level-4 gridded product was added to the catalog to provide SWH maps based on 7 altimetry missions merged together. Finally, the relevance of the along-track SWH for data assimilation was strengthened by adding a collocated wind speed derived from altimeter measurements.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Today, the wave service is processing measurements onthe-fly, derived from a constellation of up to 9 satellites. This operational system requires numerous adjustments to adapt to the upstream product evolutions and potential incidents. Constant monitoring and adaptation are essential to guarantee consistent and qualified products delivered on time for data assimilation and other NRT applications.

The spatial and temporal coverage of measurements is critical for users and one of the main assignments to the WAVE-TAC is to integrate new missions as soon as they have passed the Calibration/Validation phase. The constellation was consolidated with up to 9 missions currently available in Near-Real-Time (NRT): Jason-3, Sentinel-3A, SARAL/AltiKa, CryoSat-2, Sentinel-1A & 1B, Sentinel-3B, CFOSAT and HaiYang-2B. The new exploratory French-Chinese mission CFOSAT, with the Surface Wave Investigation and Monitoring (SWIM) instrument onboard, is dedicated to the observation of surface waves. Nadir measurements were integrated in 2020 in the Level 3 SWH product, and the expected added-value of CFOSAT mission will soon be highlighted with the provision of Level-3 partitioned wave spectra by the end of 2021.

1.2 Altimeter wave products Altimetry measurements are processed in two types of products: Level-3 along-track SWH and multi-mission Level-4 gridded SWH. Level-3 SWH product consists in along-track SWH measurements, derived from 7 altimetry missions. Measured values are edited using threshold and flags criteria to remove erroneous values. Measurements are then cross-calibrated onto the reference mission Jason-3 to ensure consistent and bias-free values across all missions. This cross-calibration depends on SWH values and is computed by comparing values at crossovers for at least a year (when possible). This product benefited from several scientific evolutions since the beginning of the wave service.

Finally, the WAVE-TAC also has had a role to investigate and to communicate on product quality. Continuous efforts have been made to implement new methods of validation and to facilitate the uptake of Copernicus Marine Service wave products by both internal and external users.

First, the editing was reinforced by adding a criterion based on a maximum SWH root mean square dispersion as a function of SWH to eliminate potential erroneous data, as explained in Queffeulou (2016).

1. MAIN ACHIEVEMENTS 2017-2021

Then, another evolution reduced the noise of measurements by applying a filter based on Empirical Mode Decomposition (EMD). EMD consists in decomposing the signal into socalled intrinsic mode functions. These functions are computed with an iterative process on the input signal (they are therefore signal-dependent). This is particularly well suited for signals that present noise distribution variability, such as SWH. The implementation of this new denoising method followed the work of Kopsinis and McLaughlin (2009) and the specific tuning for SWH signal proposed by Quilfen and Chapron (2019).

1.1 Consolidation of the constellation The production and dissemination of the first Copernicus Marine Service altimeter wave products started mid-2017 within the SL-TAC using observations derived from Jason-3 and Sentinel-3A altimeter missions. Since 2018, and the creation of the WAVE-TAC, the constellation has been constantly growing. First, in 2018, two secondary drifting missions SARAL/AltiKa and Cryosat-2 were added to the system. Then, the Copernicus Sentinel-3B mission, interleaved with Sentinel-3A, was integrated in 2019, optimizing the spatial sampling of measurements. Finally, CFOSAT nadir and HaiYang-2B were added to the system in 2020. As shown on Figure 1, the consolidation of the altimetry constellation resulted in a significant increase of the spatial and temporal density of observations since 2017.

The comparison of Level-3 SWH with in-situ measurements shows a satisfying match, with a root mean square difference (RMSD) ranging from 4 to 10 cm, depending on the mission and a correlation coefficient above 0.98. Finally, wind speed derived from altimeter measurements was added as a new field in the Level-3 SWH datasets. This evolution addressed a specific user need to get wind speed collocated to wave height values. It led to a better assimilation in wave models and a better understanding of sources of errors, whether they depend on the forcing or on the wave model. As for SWH, wind speed measurements are edited and cross-calibrated onto the reference mission Jason-3.

In 2018, SAR missions Sentinel-1A and 1B were integrated in Copernicus Marine Service wave products for the first time to provide Level-3 SAR-derived ocean wave spectra, referred to as Level-3 SPC product.

#57

The NRT gridded Level-4 SWH product benefited from quality improvements of Level-3 datasets and from the dense coverage of altimeter observations. This product provides easy-to-use gridded observational SWH data and two types of wave height fields are available. First, daily SWH statistics (mean, maximum and standard deviation) merging altimeter observations over a 24-hour time window onto 2°x2° grid cells. Then, an estimate of the instantaneous SWH field at 12UTC each day. A weighted average of nearby observations is computed, based on the

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

temporal distance between the time of the map and observations. Empty cells are filled using spatial interpolation. This estimate was improved by adapting the selection time window to the local wave climatology. With a 7-satellite constellation, the daily instantaneous SWH field filled the gaps left with along-track coverage and shows a reasonable agreement with the in-situ buoy network, with a RMSD of about 15 cm and a correlation coefficient above 0.92.

Figure 2: Pacific: SAR-derived swell conditions (size of dots indicates the wavelength, colour the wave height and tail the upcoming direction); Atlantic: Altimetry-derived along-track (lines) and gridded (contours) significant wave height.

#57

1.3 Toward more wave parameters

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

interleaved missions. Only valid measurements are disseminated, and the consistency between measurements is guaranteed by the cross-calibration onto the reference mission (Jason-3) and onto in-situ measurements. NRT data are usually available 3 to 4 hours after the actual acquisition, enabling their assimilation in forecasting systems. Also, the high number of qualified measurements and associated wind field directly benefit wave simulations and ensure an increased resilience of the system when a mission is temporarily unavailable.

SAR measurements are processed in a single type of product: Level-3 SPC. It consists in both along-track and along swellpath wave integral parameters, derived from Sentinel-1A and -1B missions. The dataset parameters include partition SWH, partition peak period and partition peak or principal direction given along swell propagation path in space and time at a 3-hour timestep, from source to land. This product is proposed in two different formats: observations grouped according to the initial storm event they originate from or grouped in 3-hour files according to their observation time. This last format was introduced in 2019 as required by numerical wave modelers for data assimilation.

The gridded Level-4 SWH product also benefited from the high number of Level-3 measurements and provides userfriendly maps, containing different daily statistics and an instantaneous estimate of the mid-day wave height field. The Level-3 SPC product based on Sentinel-1’s constellation currently delivers partitioned wave spectra at observation time as well as integral parameters (partition significant wave height, partition peak period and partition peak or principal direction) given along swell propagation path in space and time at a 3-hour timestep, from source to land. Quality flags are also included for each parameter and indicate the valid time steps along propagation (e.g., no propagation for SWH close to a storm source or any integral parameter when reaching the land). The integral parameters at observation points are also available together with a quality flag based on the consistency between each propagated observation and the overall swell field.

In 2019, a better characterization of long swell was also introduced in the Level-3 processing to deal with the Level-2 inherited inability to properly estimate the swell propagation direction for the longest swell systems. An important work was also initiated to integrate spectral parameters derived from CFOSAT off-nadir SWIM instrument. This new Level-3 dataset should become operational to all users in 2021. This new source of data will complement the Sentinel-1 mission already existing products by providing swell observations on all ocean basins, including North Atlantic, and with different limitations than those of SAR (especially shortest and azimuth propagating swells). This will be the first inclusion of measurements from another mission concept in the L3 SPC products, which will require a careful intercomparison and inter-calibration.

Since the creation of the WAVE-TAC, the quality and quantity of data has been continuously improved. New fields (quality flags, uncertainty, wind speed) were added to match evolving user’s needs. The validation approach was reinforced by comparing satellite measurements not only with other satellites, but also with in-situ measurements and simulation outputs. Facilitating the uptake of Copernicus Marine Service wave products is essential for a broader utilization by both internal and external users. Along-track altimeter and SAR products are well known in the assimilation community, but their spatio-temporal sampling requires some expertise for new users. Therefore, the WAVE-TAC took the opportunity to communicate on the wave products and their evolution and quality at international scientific conferences and during a dedicated Wave Group / TWAPAS (Tailored WAve Products for Assimilation Systems) meeting.

2. STATUS AT THE END OF COPERNICUS 1

At the end of Copernicus 1, the WAVE-TAC is delivering three NRT wave products (see Figure 2), including SWH and spectral parameters, derived from a constellation of nine satellites. The spatio-temporal coverage of the Level-3 SWH product was significantly increased with the integration of a total of 7 altimeters, including the Copernicus Sentinel-3A/B

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

and calibration/validation of wave hindcasts. Long-term reprocessing of altimetry missions will be disseminated by the end of 2021 based on the CCI Level-2P Sea state products and will be extended to SAR and other missions in the coming years.

3. POST 2021 PROSPECTS/ OUTLOOK

The main drivers of the system evolution for coming years are following user needs, and include the continuity of the current service and the products improvement.

Finally, a particular focus will be on proposing higher resolution products thanks to new data processing. The increase of along-track sampling will highly rely on: - the noise reduction in upstream products (improved re-tracking, etc.), - the improvement of along-track data selection to improve density and accuracy when approaching coasts, ice shelves or other complex areas, - the processing of along-track 20-Hz upstream products into 5-Hz products (~1.4 km resolution), with adapted subsampling and filtering methods.

The WAVE-TAC system will evolve based on the evolution of the constellations of altimetry, SAR and other radar missions measuring wave spectra (e.g., CFOSAT/SWIM off-nadir) to ensure a robust NRT sampling. Regarding the altimetry constellation, the following missions will be integrated: Sentinel-6A (2022), HaiYang-2C (2022), HaiYang-2D (2022), SWOT’s nadir (2023), Sentinel-3C (2024). Concerning the SAR and other wave spectra measuring device constellation, the following missions will be integrated: CFOSAT/SWIM offnadir (end 2021) and Sentinel-1C (2023). A particular effort will be made to improve the coverage of spectral wave parameter measurements along European coasts, using Sentinel-1 wide swath modes and other missions.

All these proposed developments will be coordinated and discussed with end-users during organized dedicated meetings (TWAPAS). These will contribute to coordinate the research and development activities upstream of the different products as well as gather user’s experience and needs on the already distributed ones.

Access to homogeneous long-term wave measurements is crucial for climate applications, assimilation in reanalyses

REFERENCES: Kopsinis, Y., & McLaughlin, S. (2009). Development of EMD-based denoising methods inspired by wavelet thresholding. IEEE Transactions on Signal Processing, 57(4), 1351–1362. https://doi. org/10.1109/TSP.2009.2013885

Queffeulou P., OSTST 2016, Validation of Jason-3 altimeter wave height measurements.

Quilfen, Y., & Chapron, B. 2019. Ocean surface wave-current signatures from satellite altimeter measurements. Geophysical Research Letters, 46, 253–261. https://doi. org/10.1029/ 2018GL081029

WIND THEMATIC ASSEMBLY CENTER ACHIEVEMENTS STOFFELEN, A.1, GIESEN, R.1, BENTAMY, A.2,

1 Royal Netherlands Meteorological Institute (KNMI), De Bilt, The Netherlands Institut Français de Recherche pour l’Exploitation de la Mer (IFREMER), Plouzané, France

been leveraged to identify biases between observed and modelled wind fields and develop a locally bias-corrected ECMWF ocean forcing product.

OVERVIEW Wind stress on the ocean surface forces ocean dynamics and plays an essential role in the heat, momentum and gases exchange at the air-sea interface. Winds are highly variable at all temporal and spatial scales and not well captured on ocean eddy scales. Even with the growing constellation of scatterometers and other satellite wind instruments, sampling remains incomplete. On the other hand, users need wind forcing products at kilometric scale (ocean eddy scale) with global coverage and high temporal frequency.

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 1.1 Wind product evolutions L3 wind products are constrained by upstream satellite data availability and associated L2 input products from OSI SAF. New datasets are added only when data quality and temporal coverage are stable. Metop-A and Metop-B ASCAT datasets in the L3 NRT product were complemented by three new scatterometer datasets over the period 2015-2021, ScatSat-1 OSCAT in 2018, Metop-C ASCAT in 2019 and HY-2B HSCAT in 2020. The total number of available daily scatterometer observations over the global ocean has thereby increased from around 4 million in 2015 to around 8 million in 2021 (Figure 1).

The Wind Thematic Assembly Centre (TAC) developed a unique repository of L3 and L4 surface wind and wind stress vector observation products of unmatched quality for both operational and climate purposes. The WindTAC product evolution relies on the evolving satellite constellation delivering basic data exploited for CMEMS products, mainly through the EUMETSAT Ocean and Sea Ice Satellite Application Facility (OSI SAF). Wind-TAC products contain wind information from scatterometers, radiometers and the European Centre for Medium-Range Weather Forecasts (ECMWF) model. All L3 wind products contain ECMWF model winds that are collocated in space and time with scatterometer observations at L2. The L2 scatterometer and ECMWF model winds are subsequently sampled and processed to L3 wind products in the same way. Therefore, they are subject to identical spatial and temporal sampling errors, which are evaluated against nominally gridded ECMWF products.

Next to the sea surface vector winds and their latitudinal and meridional components, L3 and L4 products contain several derived variables that are dedicated to downstream users. All products contain surface wind stress and its latitudinal and meridional components. In addition, L3 daily and L4 6-hourly products include: - the divergence and rotation of the wind speed vector field, - the divergence and rotation of the wind stress vector field. All L3 wind datasets include ECMWF model winds collocated in space and time with each scatterometer observation. Since scatterometers sense the ocean roughness, measurements contain no atmospheric information. To achieve the closest similarity to scatterometer winds, L2 ECMWF model winds are converted to 10 m stress-equivalent winds (U10S) by taking out effects of atmospheric stability and air mass density (de Kloe et al., 2017).

Over the period 2015-2021, the L3 NRT (near-real time) wind product in the CMEMS catalogue has been updated with three newly available scatterometer datasets. Reprocessed (REP) L3 and L4 products have been introduced to complement NRT products, extending the time coverage back to 1992. The collocated stressequivalent scatterometer and ECWMF model winds have

#57

L3 and L4 NRT wind products have been complemented by equivalent L3 and L4 REP wind products. The covered time range stretches from 1992 to the year before present. Initially, the L3 REP product included collocated ECMWF ERA-Interim stress-equivalent model winds. Since ERAInterim was discontinued in 2019 and users have switched to ERA5, ERA-Interim model winds in the L3 REP product

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

were replaced by ERA5 model winds in 2020. The L3 REP product update frequency was increased to quarterly in 2020 and will be further increased to monthly in 2021, to include observations up to three months before present. The update frequency of L4 REP products will be increased to at least 6-monthly in 2021.

Figure 1: Daily global number of scatterometer observations within the L3 NRT product from January 1st 2016 to March 31st 2021. The number shown is the sum of all scatterometer datasets included in the product at each specific day. Comments indicate significant changes in this number, caused by either introduction of new datasets or upstream data anomalies longer than 2 days.

1.2 Scatterometer versus ECMWF model surface winds

reduced mean poleward flow at mid-latitudes and weaker trade winds in tropics (Figure 2a). The spatial distribution of the systematic model biases is similar for ERA-Interim and ERA5, but generally about 20% smaller in ERA5. A comparison of transient winds reveals that wind speed variability is significantly reduced in ERA models compared to scatterometer observations (Figure 2b). Inability of the atmospheric model to reproduce higher-frequency wind variability implies an underestimation of atmospheric forcing at the air-sea boundary. This has detrimental consequences for ocean forcing and the representation of air-sea interaction in coupled models.

Many marine forecasting centers use ECMWF-based wind inputs for ocean model forcing, including their associated biases. To characterize differences between observed and modelled surface wind fields, scatterometer observations (Metop-A ASCAT) were compared to the collocated ECMWF ERA-Interim and ERA5 wind fields in the L3 REP wind product (Belmonte Rivas and Stoffelen, 2019). The model mean zonal winds are larger than scatterometer winds, pointing to stronger subtropical easterlies and mid-latitude westerlies. On the other hand, the mean meridional model winds speeds are too low, resulting in

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 2: Annual mean meridional (a) wind speed difference and (b) transient wind speed difference between scatterometer (Metop-A ASCAT) and collocated ECMWF ERA5 for 2018.

1.3 Improved ocean wind forcing products

and a temporal averaging window of two to three days were found to be the best balance between preserving small-scale temporal and spatial variability and smooth correction fields. Further development and testing of ERA* is ongoing to produce improved wind forcing products, that can be added to the CMEMS catalogue. Improved hourly wind vector fields can be used for wave models, storm surge prediction and ocean forcing alike. They, furthermore, overcome problems with scatterometer data assimilation, atmospheric model dynamical closure and weak ocean coupling of atmospheric models. In addition, a comparison and validation of ERA5* and Ifremer L4 products is being carried out.

In order to reduce systematic biases in the numerical weather prediction model (NWP) wind fields, a scatterometer-based correction was derived. Temporallyaveraged differences between geolocated scatterometer wind data and ECMWF re-analysis fields (Figure 3) were applied to original ECMWF winds to produce a bias-corrected ocean wind forcing product, ERA* (ERA star, Trindade et al., 2020). Verified against independent observations, the variance of differences was reduced by 20% in ERA* compared to uncorrected NWP fields. A combination of at least two scatterometers with complimentary orbits

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Collocated differences between scatterometer (Metop-A, Metop-B and Metop-C ASCAT) and ERA5 U10S meridional wind components, accumulated over a 3-day temporal window (1-3 June 2019).

1.4 Ocean Monitoring Indicators

constellation of scatterometers to the CMEMS catalogue, in order to improve the temporal coverage of satellite wind vector data. Possible candidates are Oceansat-3, MetopSG, FY-3E WindRad, CFOSAT and additional satellites in the HY2 series, provided that these datasets pass the data quality and stability criteria as required by OSI SAF at L2.

Wind pattern anomalies are connected to changes in ocean variables (such as waves, currents, sea surface temperature and salinity) and global climate indicators (like the North Atlantic Oscillation, El Niño Southern Oscillation, or the Indian Dipole index). The Wind-TAC has introduced three Ocean Monitoring Indicators that exhibit annual anomalies in the mean wind, the wind variability and the wind-induced Ekman upwelling. Global anomaly maps display wind pattern information to a wide range of users.

New satellite wind sources will be considered as input data for current L4 wind products, such as passive microwave radiometers. Furthermore, developments will focus on improving the selection procedure of remotely sensed wind observations in space and time to limit detrimental effects in downstream applications due to sampling issues. As the spatial and temporal resolution of ocean models increases, users require higher-resolution ocean wind forcing products. The spatial and temporal resolution of the current L4 wind product is expected to be enhanced at regional scales to hourly wind analyses at a 0.125° horizontal resolution. These developments will exploit spatial wind structures based on wind retrievals from SAR onboard Sentinel-1a and 1b within selected ocean regions (e.g., Mediterranean Sea, North Atlantic, Indian Ocean and the main eastern boundary upwelling systems).

1.5 Status at the end of Copernicus 1 The Wind-TAC portfolio now contains five wind products, two NRT L3 and L4 products, their REP counterparts and a L4 monthly average product. Over the six-year period, the amount of available NRT scatterometer observations has nearly doubled. The update frequency of the REP products has increased allowing users to perform their analyses over the longest possible period. The collocated ECMWF model winds in the products have been leveraged to identify systematic and variance biases in the model. This achievement led to the current development of a scatterometer-corrected ocean forcing product.

The ongoing development of ERA* is expected to culminate with the introduction of a new L4 wind product, providing hourly scatterometer-corrected ECMWF model wind fields at 0.125° horizontal resolution. Extensive validation of the method will be performed to select configurations that provide the best reduction of scatterometer-model biases. The Wind-TAC encourages close collaboration with CMEMS MFCs, C3S and other users to assist them in the exploitation of improved L3 and L4 wind forcing products for wave, current and air-sea interaction modelling.

1.6 Post 2021 perspectives The Wind-TAC will continue to introduce newly available L3 scatterometer datasets from the international virtual

#57

The improvement of coastal processing is a main theme in the EUMETSAT OSI SAF wind activities in upcoming years and will be propagated to CMEMS products. Diurnal wind cycles may be substantial near land-sea interfaces and affect local ocean conditions, such as coastal erosion, upwelling, primary production, etc. The growing virtual scatterometer constellation will provide observations at different times of day and will thereby be capable of measuring diurnal wind cycles. SAR snapshots collocated with scatterometers will be used to refine winds in coastal zones.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Last but not least, users show interest in extreme winds and ocean drag. As for now, in-situ moored buoy and dropsondes wind speed measurements are inconsistent under extreme conditions (Stoffelen et al., 2020). Thus, a common and consolidated reference is necessary to calibrate satellite wind speed products. In particular to estimate hurricane intensity and to tune atmospheric surface drag parameterizations and associated ocean mixing. Future calibration developments of high and extreme wind speeds will be integrated into the CMEMS products.

ACKNOWLEDGEMENTS Authors acknowledge Wind-TAC team members who contribute to the production, monitoring and quality evaluation of CMEMS wind products, including: Anton Verhoef, Cédric Prevost, Maria Belmonte Rivas, Jos de Kloe and Gerd-Jan van Zadelhoff.

REFERENCES: Belmonte Rivas, M. and A. Stoffelen (2019), Characterizing ERA-Interim and ERA5 surface wind biases using ASCAT, Ocean Sci., 15, 831–852, doi: 10.5194/os-15-831-2019.

Kloe, J. de, A. Stoffelen and A. Verhoef (2017), Improved use of scatterometer measurements by using stress-equivalent reference winds, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 10 (5), doi: 10.1109/ JSTARS.2017.2685242.

Stoffelen, A., A. Mouche, F. Polverari, G.-J. van Zadelhoff, J. Sapp, M. Portabella, P. Chang, W. Lin and Z. Jelenak (2020), C-band High and Extreme-Force Speeds (CHEFS) - Final Report, EUMETSAT project report, KNMI, https://www-cdn.eumetsat.int/ files/2020-06/pdf_ss_chefs_ final_rep.pdf

Trindade, A., M. Portabella, A. Stoffelen, W. Lin and A. Verhoef (2020), ERAstar: A high-resolution ocean forcing product, IEEE Trans. Geosci. Remote Sens., 1-11, doi: 10.1109/TGRS.2019.2946019.

THE MULTI OBSERVATIONS THEMATIC ASSEMBLY CENTRE

GUINEHUT, S.1, BUONGIORNO NARDELLI, B.2, CHAU, T.3, CLAUSTRE, H.4, ETIENNE, H.1, GEHLEN, M.3, GREINER, E.1, MULET, S.1, SAUZÈDE, R.4, VERBRUGGE, N.1 CLS, Ramonville Saint-Agne, France - 2CNR, Naples, Italy - 3LSCE, Gif sur Yvette Cedex, France - 4LOV, Villefranche-Sur-Mer, France

OVERVIEW Complementary to ocean state estimates provided by modelling and assimilation systems, a multi observationsbased approach is available through the MULTIOBSERVATIONS (MULTIOBS) Thematic Assembly Center (TAC) of the European Copernicus Marine Environment Monitoring Service (CMEMS). Copernicus Marine Service MULTIOBS TAC provides multi-observations ocean products at global scale derived from the combination of two or more different sensors, satellite and in situ, and using state-of-the-art data fusion techniques. These products cover the blue ocean for physics and the green ocean for the carbonate system and biogeochemical variables. MULTIOBS products are available in Near-Real-Time (NRT) or as Multi-Year Products (MYP) for the past 25 to 35 years with regular temporal extensions. MULTIOBS TAC provides also associated Ocean Monitoring Indicators (OMIs). It uses mostly input from other TACs. MULTIOBS TAC delivers 4 physical, 1 carbon and 2 biogeochemical products and 3 associated OMIs that are dedicated to the following: - provide global ocean state-estimates of variables still critically poorly sampled at all scales, - take advantage of the strength of the Global Ocean Observing System (in situ and satellite), - s tay close to the observations (i.e., unbiased), - resolve mesoscale structures at the right place (when eddy permitting), - provide long stable timeseries enhancing ocean climate and ocean health monitoring capabilities. MULTIOBS TAC was created in 2018 and has 4 main partners: Collecte Localisation Satellites (CLS), Consiglio Nazionale delle la Ricerche (CNR), Laboratoire des Sciences du Climat et de l’Environnement (LSCE) and Laboratoire d’Océanographie de Villefrance, Institut de la Mer de Villefrance (LOV/IMEV). However, a multi observations

component called GLO-OBS (Global Observations) existed since the year 2015 as a sub-component of Copernicus Marine Service Global Monitoring and Forecasting Center (GLO-MFC) (Le Traon et al., 2017). Most of MULTIOBS TAC products are very recent and have been in Copernicus Marine Service catalog for just about 2 years. Those new products leverage R&D from EU H2020 project and other projects funded by space agencies (CNES and ESA). Main achievements from 2015 to 2021 are listed in section 1 by product. MULTIOBS TAC status at the end of Copernicus-1 is then described (section 2) before post 2021 perspectives (section 3).

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 1.1 SSS/SSD Starting in 2015, CNR developed a global 2D sea surface salinity (SSS) and sea surface density (SSD) L4 product by interpolating in situ SSS and SSD data with the multidimensional Optimal Interpolation (OI) technique originally introduced by Buongiorno Nardelli (2012). This method extracts information on surface patterns from satellite Sea Surface Temperature (SST) L4 data, increasing the effective resolution of interpolated fields. The technique, originally developed to interpolate SSS, has been modified by Droghei et al. (2016) to provide dynamically consistent SSD field. The first version of this product was provided on a 1/4° regular grid at weekly sampling (monthly averaged fields are also available) as a multi-year timeseries in April 2017. In 2018, two major evolutions have been developed: first the use of SMOS satellite fields as additional forcing, then the processing chain adaptation to provide consistent NRT fields. Continuous improvements were then implemented.

101

#57

A new version of both MY and NRT processing chains was released in April 2019 to include new SSS upstream satellite product from SMOS and internal climatological fields as background. The last improvement was implemented in December 2020 to include SST upstream data changes with the exploitation of Copernicus Marine Service OSTIA SST. The MY timeseries covers now the 1993-2019 period and is extended by the NRT.

1.2 ARMOR3D During MyOcean projects (MyOcean1, MyOcean2 and MyOcean-FO), CLS developed a global 3D physical ocean state product (namely ARMOR3D). ARMOR3D provides on a 1/4° horizontal grid at weekly sampling (monthly averaged fields are also available) 3D global fields of temperature, salinity, geopotential heights and geostrophic currents down to the bottom. This L4 product is available as a MY timeseries since 1993 and in NRT. It is obtained by combining satellite (SLA, geostrophic currents, SST, SSS) and in situ (T/S profiles) observations through statistical methods (Guinehut et al., 2012; Mulet et al., 2012). ARMOR3D has a major role in MULTIOBS TAC ocean stateestimates as it uses MULTIOBS SSS product (see previous section) as input and as it is used by almost all other products (OMEGA3D, Copernicus-Globcurrent, surface carbon, 3D POC/Chla; see following sections).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Main achievements of the last 6 years are the continuous improvement of the product with: - the addition of information to answer user needs (mixed-layer depth variable, monthly mean field, increase in vertical resolution), - the use of new or updated upstream (first guess climatology, MULTIOBS SSS, OSTIA SST, DUACS18 L4 altimeter sea level anomaly, CNES-CLS18 MDT), - the improvement of the method (correction of nonsteric signal in altimeter measurements, updated and denoised covariances used to project vertically downwards the information provided by surface fields), - the provision of fully consistent MY and NRT timeseries. ARMOR3D contributes directly to two OMIs for ocean warming. The first one is the Global Ocean Heat Content (production under GLO-MFC responsibility). The second one, online since September 2019, is the Global Cumulative Trend of zonal mean Subsurface Temperature (production under MULTIOBS TAC responsibility) and was published in issues 1 and 2 of Copernicus marine Service Ocean State Report (Guinehut et al., 2017, Mulet et al., 2018). Both are computed from a multi-product approach using fields from ARMOR3D, 4 global reanalyses (GREP) and the in situ gridded field CORA. Estimation of the robustness of both indicators is provided from the multi-product approach. From ARMOR3D solution, confirmed by all solutions, the warming appears to be significant in almost all parts of the ocean and at depths of up to 800 m depth (Figure 1).

Figure 1: ARMOR3D Depth/Latitude global mean temperature cumulative trend over 1993-2019 (in °C).

102

#57

1.3 OMEGA3D A new MY product called OMEGA3D developed by CNR is available online since April 2020. It provides observationbased 3D quasi-geostrophic vertical and horizontal ocean currents over 75 levels from the surface to 1500 m depth, at 1/4° horizontal resolution, from January 1993 to December 2018. Current velocities are obtained by solving a Q-vector formulation of the Omega equation, including diabatic forcing terms (Buongiorno Nardelli, 2020). This product is

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

based on a combination of different fields: temperature, salinity and geostrophic currents provided by ARMOR3D and ERA-Interim surface fluxes. Vertical velocities are now included in the list of parameters available through Copernicus Marine Service catalogue. An illustration of OMEGA3D quasi-geostrophic vertical currents at 50 m is given in Figure 2. It shows vertical exchanges about 10 meters per day associated with the Gulf Stream mesoscale activities but also the effect of Hurricane Florence.

Figure 2: OMEGA3D vertical velocity at 50m depth (in m/day) for the 12/09/2018 (left) and MODIS Aqua image illustrating Florence Hurricane position (right).

1.4 Surface and near-surface current (Copernicus-Globcurrent) It is well recognized that surface currents in the ocean are not the simple addition of different current components. Although, a simple approximation of ocean currents can be made by combining geostrophic currents (derived from altimeter Absolute Dynamic Topography fields) with an estimate of the Ekman response to wind forcing. CLS has been producing for internal use, global fields of such combined geostrophic+Ekman currents which have been continuously improved (Rio et al., 2014). In 2014, ESA funded the GlobCurrent project (2014-2017) under the Data User Element program in which global combined surface and near-surface ocean current based on CLS work have been provided in delayed-time and in NRT. In 2018, Copernicus Marine Service took over the production and distribution of this product now called CopernicusGlobcurrent with a first version online since July 2018. Copernicus-Globcurrent product consists in total velocity fields (zonal and meridional) at 0 m and 15 m depth, at 6 h frequency in NRT and at 3 h frequency for MY timeseries. Daily and monthly means are also available.

The geostrophic currents are calculated through the geostrophic approximation applied to the sum of altimeter sea level anomalies (SLA) and a mean dynamic topography (MDT), both coming from Copernicus Marine Service Sea Level TAC. Ekman currents are computed at two depths (0m and 15m) applying an empirical Ekman model updated from Rio et al., (2014) to ECMWF wind stress fields. Parameters of the empirical Ekman model are computed using in situ observations from Argo drifts at the surface and SVP-type drifters at 15 m. Two versions of the product have been successively released. For its 1st publication in July 2018, geostrophic currents were from DUACS DT-2018 and Ekman currents were computed using ERAinterim wind stress fields. For its 2nd publication in December 2019, both MY and NRT products were improved using new upstream fields such as the CNES-CLS18 mean dynamic topography and wind stress fields from ERA5 which is the latest climate atmospheric reanalysis produced by ECMWF. The 2nd publication also used an estimation of the ocean stratification coming from the mixed-layer depth of ARMOR3D in a new configuration of the empirical Ekman model. The MY timeseries have been extended twice a year, as soon as upstream fields were available. It currently covers the 1993 – May 2020 period.

103

#57

Furthermore, a new preliminary multiyear product of 2D surface current has been available internally since July 2020, as a demonstration product. This product is inherited from ESA GlobCurrent and CNES DUACS-MR projects. The method is implemented to enhance the effective spatiotemporal resolution of the satellite altimeter-derived gridded geostrophic current. It uses high resolution satellite images of ocean tracers (e.g., sea surface temperature, SST) which relates to ocean surface dynamics. The approach is based on the inversion of the heat conservation equation for horizontal velocities using altimeter geostrophic velocities as background and prescribing source and sinks terms a priori values and associated errors from successive SST fields (Rio and Santoleri, 2018). The product is provided at 1/10° horizontal resolution along the 1993-2018 period.

1.5 CMEMS-FFNN Surface Carbon A new surface carbon multiyear product, including air-sea flux of CO2 (fgco2), partial pressure of CO2 (spco2) and pH has been available since April 2019. The model called CMEMSFFNN is inherited from the EU H2020 AtlantOS project and is developed by LSCE. It relies on the implementation of feedforward neural network models (FFNN) for the interpolation of sparse carbon system measurements to basin-wide maps on a 1° horizontal resolution grid at a monthly period (DenvilSommer et al., 2019). This method establishes non-linear relationships between chosen drivers or predictors such as

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

SST, SSS, Chlorophyll, mixed layer depth, surface height, pCO2 climatology and surface ocean pCO2 measurements from SOCAT (https://www.socat.info/). Reconstructed variables (fgco2, spco2, pH) are distributed with associated uncertainties derived from the 100-member CMEMS-FFNN ensemble. Related OMIs for ocean carbon sink (i.e., global yearly integrated air-sea flux of CO2) and ocean acidification (i.e., global mean sea water pH) are also available. The latter was published in issue 4 of Copernicus Marine Service Ocean State Report (Gehlen et al., 2020). Three versions of the product have been successively released. For its 1st publication in April 2019, the MY timeseries and the ocean acidification OMI covered the 2001-2017 period. The 2nd publication in April 2020 included a new version of the ensemble based FFNN and a timeseries covering the 1985-2018 period together with the two OMIs. The timeseries (product and OMIs) was further extended in December 2020 to cover the 19852019 period. CMEMS-FFNN air-sea CO2 flux product have been contributing since 2019 to the yearly assessment of the Global Carbon Budget (Friedlingstein et al., 2020; Hauck et al., 2020). The combination of spco2 and surface ocean alkalinity allows to reconstructing surface ocean pH from 1985 onward. The global ocean surface pH is decreasing (Figure 3), as a direct consequence of the uptake by the ocean of carbon dioxide emitted by human activities such as e.g., fossil fuel burning and land-use change.

Figure 3: Yearly mean surface sea water pH reported on total scale.

104

#57

1.6 CANYON Nutrient profiles A new biogeochemical MY product for nutrients has been available since April 2019. The so-called method CANYON-B (CArbonate system and Nutrients concentration from hYdrological properties and Oxygen using a Neuralnetwork) is inherited from EU H2020 AtlantOS project and R.Sauzède PhD thesis and is developed by LOV/ IMEV. It relies on a neural-network method to derive, from simple and cost-effective measured parameters from Biogeochemical-Argo (BGC-Argo) profiling floats, some more complex biogeochemical measurements, not yet easily or cost-effectively amenable to robotic detection (Sauzède et al., 2017; Bittig et al., 2018). CANYON-B uses measurements of temperature, salinity, pressure, and O2 together with sampling latitude, longitude, and date to retrieve concentrations of three nutrients including nitrates, phosphates and silicates. It has been trained on high quality nutrient data collected over the last 30 years

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

and made available through the GLODAPv2 database. It is then applied to all available delayed-mode qualified BGCArgo profiling floats equipped with an oxygen sensor. Three versions of the product have been successively released. Each time, new profiles according to the availability of new BGC-Argo-O2 delayed-mode data were included. The 1st publication in April 2019 covered the June 2004 to November 2018 period with 30 338 profiles. The 2nd publication in April 2020 covered the June 2004 to June 2019 period with 44 934 profiles. The 3rd publication in May 2021 extended from September 2002 until December 2020 with a total of 98 789 profiles. Mediterranean Sea profiles were also added in this last release using a specifically developed regional method called CANYON-MED (Fourrier et al., 2020). An example located in the North Western Mediterranean Sea shows winter mixing in input fields (temperature, salinity, oxygen) in February/March 2013 and associated uplift of reconstructed nutrients (Figure 4).

Figure 4: Depth/Time Temperature (in °C), Salinity (in psu) and Oxygen (in µmol kg-1) measurements from Argo float WMO 6901467 located in the North Western Mediterranean Sea and associated Nitrate, Phosphate and Silicate (in µmol kg-1) profiles as reconstructed by CANYONMED (Fourrier et al., 2020).

105

#57

1.7 SOCA3D POC/bbp/Chla Since July 2020, a new biogeochemical multiyear product have been providing global gridded fields of Particulate Organic Carbon (POC) and Chlorophyll-a concentration (Chla) at depth, two biogeochemical variables still currently critically under sampled at all scales. The method relies first on a neural network method called SOCA (Satellite Ocean-Color merged with Argo) which is inherited from the EU H2020 AtlantOS project and R.Sauzède PhD thesis. SOCA estimates vertical profiles of backscattering coefficient (bbp), a bio-optical proxy for POC, from surface ocean color satellite measurement of bbp and additional physical drivers (Sauzède et al., 2016). It has been trained on high quality bbp data collected from BGC-Argo floats. Then, an empirical parameterization is used to infer the vertical distribution of Chla from surface ocean color satellite observations of Chla and the relative position of the mixed layer and euphotic depths (Uitz et al., 2006). This parameterization was established from a database of Chl profiles acquired by High Performance Liquid Chromatography (HPLC), the reference method for such measurements. Both methods are developed at LOV/IMEV. Two versions of the product have been successively released. The 1st publication in July 2020 included POC and Chla variables, and associated errors on a 1/4° horizontal grid over 19 vertical levels from the surface to 1000 m depth, at a weekly frequency and for the 1998-2018 period. An associated monthly mean climatology is also available. The 2nd publication in May 2021 is extended with the year 2019 and provides bbp with its associated error as new variables, an improved bbp to POC relationship and a refined vertical resolution with 36 levels.

2. STATUS AT THE END OF COPERNICUS At the end of Copernicus-1, 4 physical, 1 carbon and 2 biogeochemical products implemented by four groups (CLS, CNR, LSCE, LOV/IMEV) composed the MULTIOBS TAC. The following products, briefly described in the previous section are provided at global scale: -2 D sea surface salinity and sea surface density fields, both in NRT and as MYP, - 3D temperature, salinity, geopotential height and geostrophic current fields, 2D mixed-layer depth both in NRT and as MYP, -3 D vertical current as MYP, - 2D total surface and near-surface currents, both in NRT and as MYP, -2 D surface carbon fields of flux of CO2, pCO2 and pH as MYP, - nutrient vertical distribution (nitrate, phosphate and silicate) profiles as MYP,

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- 3D backscattering coefficient (bbp), Particulate Organic Carbon (POC) and Chlorophyll-a (Chl-a) fields as MYP. Service continuity is first ensured by fully operational processing chains and then by daily or weekly operation tasks which are performed to produce and then push NRT products to the centralized Dissemination Unit. MYP are regularly extended, once or twice a year, consistently with upstream timeseries temporal extension. MULTIOBS TAC products are further used to derive OMIs for ocean warming, ocean acidification and ocean carbon sink which timeseries are also regularly extended. All products are fully validated and a dedicated QUID (Quality Information Document) is associated with each product/OMI. Results of NRT validation metrics for MULTIOBS TAC products are also available through Copernicus Marine Service product quality Dashboard. In 2020, 1344 users downloaded MULTIOBS TAC products.

3. POST 2021 PERSPECTIVES

In the continuity of activities carried out these last years, the first perspective will be to ensure the service continuity by maintaining the 7 processing chains of the MULTIOBS TAC with the four partners. System evolution roadmap includes: - to implement state-of-the-art methods and upstream data for which continuous monitoring is performed, - to incorporate and test new or improved products from TAC (Sea Level, SST, Ocean Color, Wind, in situ) and products currently outside Copernicus Marine Service, such as SSS from Aquarius/SMAP and from the future Copernicus CIMR (Copernicus Imaging Microwave Radiometer) mission, - to include measurement error and processing uncertainty for predictors to reconstruct Surface Carbon, - to test and include additional constraint from climate index if possible in ARMOR3D, - to improve information content of products by increasing resolution in space (horizontal & vertical) and in time for all MULTIOBS products, starting with the physical ones, requiring higher resolution (L3/L4) products from TAC, - to extend OMEGA3D, nutrient profiles and 3D POC/ Chla products currently available only as multiyear timeseries to near-real-time, providing fully consistent multi-year and near-real timeseries for almost all MULTIOBS products, - to improve the representation of physical processes such as ageostrophic component (wind-driven, stokes),

106

#57

higher frequency processes such as tides, and mesoscale processes such as the vertical extension of eddies, - to add new variables such as phytoplankton functional types, total primary production or vertical profiles of carbonate chemistry (alkalinity, pH, DIC), and to provide derived quantities such as Ocean Heat Content (ECV),

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- to provide uncertainties estimate for all variables for both direct use, validation purposes, and data assimilation, - to propose new products (physical and biogeochemical) at regional scale (e.g., Mediterranean Sea, Atlantic Ocean), requiring to first gather dedicated regional upstream observations and observation products and then develop specific configurations at regional scale.

ACKNOWLEDGEMENTS The authors acknowledge all MULTI OBSERVATIONS TAC team members who have contributed to products, their quality evaluation and monitoring, including: F. Chevallier from LSCE, D. Ciani from CNR, A. Conchon, S. Jousset and F. Briol from CLS.

REFERENCES: Bittig, H. C., Steinhoff, T., Claustre, H., Fiedler, B., Williams, N. L., Sauzède, R., Körtzinger, A. and Gattuso, J.P., 2018: An Alternative to Static Climatologies: Robust Estimation of Open Ocean CO2 Variables and Nutrient Concentrations From T, S, and O2 Data Using Bayesian Neural Networks, Front. Mar. Sci., 5, 328, doi:10.3389/fmars.2018.00328. Buongiorno Nardelli, B., 2012: A Novel Approach for the High-Resolution Interpolation of In Situ Sea Surface Salinity. J. Atmos. Oceanic Technol., 29, 867–879, doi:10.1175/ JTECH-D-11-00099.1. Buongiorno Nardelli, B., 2020: A Multi-Year Timeseries of Observation-Based 3D Horizontal and Vertical Quasi-Geostrophic Global Ocean Currents. Earth Syst. Sci. Data, No. 12, 1711–1723. https://doi.org/10.5194/essd12-1711-2020.

Denvil-Sommer, A., Gehlen, M., Vrac, M., and Mejia, C., 2019: LSCE-FFNN-v1: a twostep neural network model for the reconstruction of surface ocean pCO2 over the global ocean. Geosci. Model Dev. 12, 2091–2105. doi: 10.5194/gmd12-2091-2019. Droghei, R., B. Buongiorno Nardelli, and R. Santoleri, 2016: Combining in situ and satellite observations to retrieve salinity and density at the ocean surface. J. Atmos. Oceanic Technol. doi:10.1175/JTECH-D-15-0194.1. Ferry, N., L. Parent, G. Garric, B. Barnier, J.-M. Molines, S. Guinehut, S. Mulet, K. Haines, M. Valdivieso, S. Masina and A. Storto, 2012: MyOcean eddy-permitting global ocean reanalysis products: description and results. Proceedings of 20 Years of progress in Radar Altimetry Symposium, ESA Special Publication SP-710.

Fourrier, M., L. Coppola, H. Claustre, F. D’Ortenzio, R. Sauzède and J.-P. Gattuso, 2020: A regional neural network approach to estimate water-column nutrient concentrations and carbonate system variables in the Mediterranean Sea: CANYON-MED. Front. Mar. Sci., https://doi.org/10.3389/ fmars.2020.00620.

Guinehut S., S. Simoncelli, S. Mulet, N. Verbrugge and K. von Schuckmann, 2017: Chapter 1: Essential variables, Section 1.2 Subsurface temperature. In: The Copernicus Marine Environment Monitoring Service Ocean State Report, Journal of Operational Oceanography, s235-s320, DOI: 10.1080 /1755876X.2016. 1273446.

Friedlingstein P., et al, 2020 : Global Carbon Budget 2020, Earth System Science Data, 12, 3269–3340, DOI: 10.5194/essd12-3269-2020.

Hauck, J., M. Zeising, C. Le Quéré, N. Gruber, D.C. Bakker, L. Bopp, T.T.T. Chau, Ö. Gürses, T. Ilyina, P. Landschützer, A. Lenton, L. Resplandy, C. Rodenbeck, J. Schwinger, R. Séférian, 2020 : Consistency and Challenges in the Ocean Carbon Sink Estimate for the Global Carbon Budget. Frontiers in Marine Science, 7, 852, DOI: 10.3389/fmars.2020.571720.

Gehlen M., T. T. T. Chau, A. Conchon, A. Denvil-Sommer, F. Chevallier, M. Vrac and C. Mejia, 2020: Ocean acidification, Copernicus Marine Service Ocean State Report, Issue 4, Journal of Operational Oceanography, 13:sup1, s64, DOI: 10.1080 /1755876X.2020.1785097. Guinehut S., A.-L. Dhomps, G. Larnicol and P.-Y. Le Traon, 2012: High resolution 3D temperature and salinity fields derived from in situ and satellite observations. Ocean Sci., 8, 845-857, doi:10.5194/os-8-845-2012

107

Le Traon, P.-Y., et al, 2017: The Copernicus Marine Environmental Monitoring Service: Main Scientific Achievements and Future Prospects. Special Issue Mercator Océan Journal #56. https://doi.org/10.25575/56.

#57

Mulet, S., M.-H. Rio, A. Mignot, S. Guinehut and R. Morrow, 2012: A new estimate of the global 3D geostrophic ocean circulation based on satellite data and in situ measurements. Deep-Sea Res. II., 77-80, 70-81, doi:10.1016/j.dsr2.2012.04.012. Mulet S, Buongiorno Nardelli B, Good S, A. Pisano A, E. Greiner, Monier M, 2018: Ocean temperature and salinity. In: Copernicus Marine Service Ocean State Report, Journal of Operational Oceanography, 11:sup1, S1-S142, DOI: 10.1080 /1755876X.2018.1489208.

Rio M.-H., S. Mulet and N. Picot, 2014: Beyond GOCE fo the ocean circulation estimate: Synergetic use of altimetry, gravimetry, and in situ data proceeds new insight into geostrophic and Ekman currents. Geophys. Res. Lett., 41, doi :10.1002/2014GL061773. Rio, M.-H. and R. Santoleri, 2018: Improved global surface currents from the merging of altimetry and Sea Surface Temperature data. Remote Sensing of Environment, 216, 770–785.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Sauzède, R., Bittig, H. C., Claustre, H., de Fommervault, O. P., Gattuso, J.-P., Legendre, L. and Johnson, K. S.: Estimates of water-column nutrient concentrations and carbonate system parameters in the global ocean, 2017: A novel approach based on neural networks, Front. Mar. Sci., 4(128), doi:10.3389/ fmars.2017.00128. Sauzède, R., Claustre, H., Uitz, J., Jamet, C., Dall’Olmo, G., D’Ortenzio, F., Gentili, B., Poteau, A. and Schmechtig, C., 2016: A neural networkbased method for merging ocean color and Argo data to extend surface bio-optical properties to depth: Retrieval of the particulate backscattering coefficient, J. Geophys. Res. Ocean., 121(4), 2552–2571, doi:10.1002/2015JC011408.

108

Uitz, J., Claustre, H., Morel, A. and Hooker, S. B., 2006: Vertical distribution of phytoplankton communities in open ocean: An assessment based on surface Chlorophyll, J. Geophys. Res., 111(C8), C08005, doi:10.1029/2005JC003207.

GLOBAL MONITORING AND FORECASTING CENTER, CMEMS ACHIEVEMENTS

DRILLET, Y.1, LELLOUCHE, J.M.1, LAMOUROUX, J.1, LAW, CHUNE, S.1, DRÉVILLON, M.1, BOURDALLÉ BADIE, R.1, PERRUCHE, C.1, AOUF, L.3, TITAUD, O.2, SAMSON, G.1, RUGGIERO, G.1, BRICAUD, C.1, GARRIC, G.1, ZUO, H4, CIPOLLONE, A.5, RENSHAW, R.6 Mercator Ocean International, 2CLS, 3Météo France, 4ECMWF, 5CMCC, 6MetOffice

OVERVIEW The Global Monitoring and Forecasting Center (GLO MFC) is coordinated by Mercator Ocean International (MOi) who is in charge of research and development, system development and integration, calibration and validation of products, evolution and operation for most systems operated in this framework. This is achieved thanks to a strong partnership with several institutes involved in different stages in Copernicus Marine Service development phases from R&D to production. For R&D activities (including new development and system evolution), main partners are CNRS, BSC, INRIA and CNR. On the other hand, CLS, Meteo France, MetOffice, CMCC and ECMWF are involved in production of reanalysis products and also wave forecast. From 2015 to 2021, many developments have been performed in the GLO MFC with important milestones. For the operational near real time forecasting systems main updates are: - the global physical 1/12° system including new Surface Merge Ocean Current datasets (SMOC) and 3D 6-hourly frequency fields for main physical variables, - a wave forecast system with an increase of resolution, - the integration of sentinel data in physical, biogeochemical and waves forecasting system, - the data assimilation of ocean colour to initialise near real time biogeochemistry forecast. Two new reanalysis systems have been developed consistently with the near real time system: - the Global 1/12° physical reanalysis and its monthly climatology, and - the Global wave reanalysis.

Original products are now available in Copernicus Marine Service catalogue with added value to monitor ocean state and variability: - updated version of the biogeochemistry reanalysis and new micronekton reanalysis, - global multi system ensemble reanalysis at 1/4°, and - ocean Monitoring Indicators (OMis), which are: maps, timeseries, and trends for several global Essential Ocean Variables (EOV) including uncertainty information. In this paper, details about the main achievements during Copernicus 1 period are provided for the global high resolution physical system, the waves forecast, the biogeochemistry and the ocean reanalysis. Then, current status of available products delivered by the GLO MFC is summarised. The last section illustrates a few examples of ongoing developments that will be integrated in future Copernicus Marine Service systems.

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 1.1 Global high resolution physical system Since October 2016, GLO MFC has been delivering realtime daily services (weekly analyses and daily 10-day forecasts) with a new global 1/12° system called GLO12v3 fully described in Lellouche et al. (2018). This new system was initially deployed over the October 2006 - October 2016 period, assimilating “reprocessed” or “near real time” observations (in situ temperature and salinity vertical profiles, along track altimeter observations, satellite

108

#57

sea surface temperature and sea ice concentration observations). Its main purpose was to improve ocean circulation at meso scale, surface currents and also largescale biases and interannual variability. To reach this goal, the following main updates were introduced in the system: - c orrection of atmospheric forcing fields at large-scale with satellite data, -a ddition of freshwater runoff from ice sheets melting to river runoffs, -a ddition of a time varying global average steric effect to the model sea level, - improvement of the Mean Dynamic Topography used for altimeter data assimilation taking into account the last version of the GOCE geoid, - introduction of an adaptive tuning on some of the observational errors, - addition of a dynamic height criteria to the quality control of the assimilated temperature and salinity vertical profiles,

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- assimilation of satellite sea ice concentrations, and - assimilation of climatological temperature and salinity in the deep ocean below 2000 m to prevent drifts in those very sparsely observed depths. The most satisfying outcome, illustrated in Lellouche et al., (2018), was the great improvement of models’ accuracy for water masses and in particular the salinity property with a decrease of the global RMS error from 0.1 psu. Moreover, the Surface and Merged Ocean Currents (SMOC) product was developed, based on the GLO12v3 system, specifically for surface drift applications. It includes wave (Stokes drift) and tidal currents in addition to physical system ocean currents. Figure 1 shows the reduction of Lagrangian forecast errors of 18.7% on average for the global area. Locally, and especially in large-scale wind circulations, improvements are much larger and can reach up to 200% (like in the Antarctic Circumpolar Current).

Figure 1: Illustration of error reduction using SMOC, a surface current product containing effects of waves and tides. Maps illustrate error (in km) for a 72 h advection using the standard surface current (top panel) and the SMOC total current (bottom panel). More precisely, the map of separation distance compares 72 hours of Lagrangian forecasts with drifting buoys from the Global Drifter Program. Numerical trajectories were computed with u0: physical model currents alone (top panel), compared to u_total, the SMOC total current (bottom panel). The results were averaged per 2° boxes.

109

#57

Based on GLO12v3 operational system, the GLORYS12v1 reanalysis at 1/12° has been developed to cover the altimetry period. The main specificities in GLORYS12v1 stand in: -u sing atmospheric reanalysis (ERAinterim and then ERA5), - r eprocessed observation data set for the full period, -b enefit from some changes in the system, notably on observation errors (e.g., 3D T/S in situ seasonal observations errors have been computed from GLO12v3). Homogeneity of the reanalysis system, atmospheric forcing and assimilated data set allow representation of ocean variability and trend and associated uncertainties and errors. These results are presented in Lellouche et al., (2021) as, for example, the satisfying agreement between global reanalysis and satellite observations for regional sea level trends. Indeed, discrepancies between the GLORYS12v1 reanalysis and reference altimetric datasets remain small as they do not exceed +/- 2 mm/yr in the majority of the ocean observed by altimetry

1.2 Wave forecast and reanalysis Wave products were added to Copernicus catalogue in April 2017. The global wave forecasting system of CMEMS is developed and operated by Météo-France. It is based on state-of-the-art MFWAM model (JCOMM systematic intercomparison, Bidlot et al., 2006). The first version of near real time system (WAVEv1) leverages the ECWAM-IFS38R2 computing code with a dissipation term developed by Ardhuin et al., (2010). WAVEv1 was operated at 1/5° of resolution, with 6-hourly analysis and 3-hourly forecasted wind forcing from the IFS-ECMWF atmospheric system. WAVEv1 assimilated every 6 hours the significant wave height (SWH) observed by satellite (thanks to Jason 2 & 3, Saral and Cryosat altimeters). Forecasts were provided up to 5 days. Wave heights, period and directions of the total sea and its partitions (wind-sea, primary and secondary swells) were (and are still) distributed at a frequency of 3 h. In March 2018, the system was upgraded (to WAVEv2) with the IFS-41R2 computing code and several other features: - t he first one concerns the resolution, which is improved from 1/5° to 1/10°, - the second concerns physics, with an adjustment of the dissipation term and the use of a Phillips spectrum tail to constrain the high frequency part of the spectrum, - dispersion by oceanic currents is also introduced by forcing the system with daily surface currents from the GLO12v3 physical system (see Figure 1 for an indication of error on surface currents), - the last point concerns the assimilation of Sentinel1A and 1B wave spectra (Aouf et al, 2021). This is the first time that this type of data is assimilated into an

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

operational wave model, which in particular, makes possible to more accurately constrain sub-polar swells, - Sentinel-3A is also added in the assimilated altimetry constellation. The validation of the global wave prediction system is performed using wave buoys and the independent altimeter HY2A, which shows values of scatter index of around 14% for SWH , and a SWH bias of 1 cm. Since December 2020, the system has been assimilating also wave heights from SWIM nadir CFOSAT, a Franco-Chinese satellite mission (WAVEv2.1). This improves scores by nearly 10%, especially for high latitudes. The forecast range is also extended to 10 days. A new multi-year product (called WAVERYSv1) was also created to cover altimetry period from 1993 to present. It shares the same specificities as the real-time system WAVEv2, except that its horizontal resolution is 1/5°. WAVERYSv1 has been shown to outperform the wave dataset of the ERA5 climate reanalysis, notably thanks to its better dissipation physics, the introduction of ocean currents and the assimilation of wave spectra over the last few years (Law-Chune et al, 2021).

1.3 Biogeochemistry data assimilation The biogeochemistry system, in its 1/4° configuration, was first commissioned at the beginning of CMEMS, in late 2014. Since then, three major evolutions have been carried out: 1. the upgrade of the dynamical forcing ocean, from the historical 1/4° to the current GLO12v3 1/12° (cf. section 2.1), 2. the upgrade of the NEMO-PISCES model to version 3.6 that includes new biogeochemical parameterizations (e.g., nitrogen fixation and impact of day length on phytoplankton growth), 3. the activation of an Ocean Colour data assimilation embedded system, allowing a better control and confidence into the model outputs. This section provides a short synthesis of the last 2 points. The data assimilation (hereafter DA) of satellite Ocean Colour maps capability has been effective since July 2019. It is based on the MOi data assimilation tools (reduced order Kalman filter, based on the Singular Evolutive Extended Kalman filter formulation). The system thus, operationally assimilates daily L4 remotely sensed surface Chlorophyll, and produces a surface correction field of Chlorophyll and Nitrates (Lamouroux et al., in prep.). This correction is then projected vertically all along the local mixed layer. In this first version of the biogeochemical assimilative system, only large-scale corrections (>500 km) are applied to the modelled Chlorophyll and nitrate. It has been indeed preferred to let the model develop its own

110

#57

small-scale biogeochemical dynamics, considering that an efficient correction of small scales would probably require an online coupled physical-biogeochemical DA system. In parallel, a climatological relaxation is applied to nutrients (NO3, PO4, Si, and Fe), dissolved oxygen and variables of the carbonate system (DIC, DOC and Alkalinity). This relaxation mitigates the physical DA impact on the offline coupled hydrodynamic-biogeochemical system. Indeed, this DA caused significant rise of nutrients in the Equatorial Belt area and resulted in an unrealistic drift of various biogeochemical variables, e.g., Chlorophyll, nitrate, phosphate (Gasparin et al., 2020). The time scale associated with this climatological damping is set to 1 year and enables a smooth constraint that has been shown to be efficient enough to reduce the model drift, and to let the model developing its own interannual variability. In its last - and current - version, this near real time system then produces high-frequency daily outputs for Chlorophyll concentration, phytoplankton concentration, dissolved oxygen, nitrates, phosphates, silicates, iron, as well as surface pCO2 and pH. A global 3D quasi-independent dataset of BGC-Argo profiles complemented with profiles reconstructed

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

through a neural network (Sauzède et al., 2017) was used to statistically measure the impact of these evolutions on model’s performance. Figure 2 shows the overall global positive effect of this new system on classical statistical metrics (Root Mean Square Error, standard deviation, correlation), especially in the euphotic layer (0-100 m). In addition, a set of process-oriented metrics were designed in Mignot et et al., (2021) to assess dedicated features of ecosystem dynamics. Namely: phytoplankton growth via photosynthesis, oxygen minimum zones and the carbonate system. These metrics have thus been used to compare the global forecasting system with the BGC Argo dataset, and highlight the global agreement between both components (not shown here). Focusing on the spatial distribution of Chlorophyll and nutrients, both the extension and amplitude of oligotrophic gyres are now better represented. In the Tropical band, Chlorophyll and nutrient distribution, as well as their interannual variability, are improved thanks to the relaxation toward climatologies. Also note that the seasonal cycle has been improved (not shown here but available in QUID report, Lamouroux et al., 2019).

Figure 2: Taylor diagram comparing the statistical performance of the last two biogeochemical systems, through surface oxygen, oxygen at 300 m depth, chlorophyll in the 0-100 m surface layer and nitrate in the 0-100 m surface layer. The reference dataset (yellow star) is based on BGC-Argo database complemented with reconstructions through a neural network (see Mignot et al., (2021), in review, for further details).

111

#57

1.4 Biogeochemistry timeseries

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Levels (LMTL) module of the SEAPODYM model (Lehodey et al, 2015) and have been developed during GREENUP service evolution CMEMS project, 3. In 2021, a major upgrade of the system was implemented with a 1/12° resolution product covering 1998 to 2020. Figure 3 illustrates this improvement with small scale features in the zooplankton fields in the north Atlantic. A better match to the horizontal resolution of the global physical system (GLO12v3, section 2.1) is observed and especially the mesoscale features, meanders and eddies as illustrated in the Gulf Stream area.

Since 2015, the catalogue of global biogeochemical multiyear products has been greatly enriched: 1. From a 19-years (1998-2016) span, the biogeochemical multi-year timeseries BIORYS (based on the NEMO-PISCES model, capturing the first trophic levels) now covers almost 30 years (1993-2020) with high-frequency daily outputs, 2. In 2019, a new primary production together with the zooplankton and the micronekton at 1/4° were produced and delivered. These essential ecosystem variables are based on the Lower and Mid-Trophic

Figure 3: Sample of a 1/12° daily north Atlantic Ocean simulation of the biomass concentration of zooplankton with SEAPODYM-LMTL.

1.5 Multi system ensemble reanalysis As reviewed by Storto (20196)(a), global ocean model reanalyseis have the capacity to capture the ocean variability and trends accurately, and are used as oceanic initial conditions by seasonal forecasting systems. However, biases and errors appear where and when observations are sparse. For instance, errors which appear in the southern oceans or on the continental shelves may propagate, and

biases tend to accumulate at depth. In this context, the Global Reanalysis Ensemble Product (GREP) (Storto et al, 2019) was developed, based on four reanalyses: - GLORYS2V4 (Garric et al., 2018), - GLOSEA5v13 (MacLachlan et al 2015), - C-GLORS (Storto et al, 2016), - ORAS5 (Zuo et al, 2019). Those are comparable to global NEMO ORCA025 configurations but differ by their data assimilation

112

#57

methods and model parametrization choices. The four members and their daily 3D mean and standard deviation were exploited to derive envelopes of estimates for several global ocean monitoring indicators (oOcean heat Ccontent (OHC) and Ssteric Ssea lLevel, nino3.4 indices, Ssea Iice eExtent, ocean volume and heat transports …) published in the Ocean State Report. The signal-to-noise ratio highlights the most robust spatial structures on the OHCcean Heat

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Content trend maps (https://marine.copernicus.eu/ access-data/ocean-monitoring-indicators/global-trendmap-ocean-heat-content). Figure 4 illustrates, for instance, uncertainty of the sea ice extent in the Arctic Ocean computed from 1993 to 2018. The trend and interannual variability are similar between the four estimates based on available global reanalysies. Uncertainty decreased over time thanks to improvement of assimilated data set.

Figure 4: Sea ice extent ocean monitoring indicator in the Arctic Ocean computed with the four global reanalysis available in the multi system ensemble product (GREP).

Ocean Currents (SMOC) product was developed and, in March 2020, 3D 6-hourly frequency products were added in the catalogue, - GLOBAL 1/4° biogeochemical forecasting system (GLO4v2); based on PISCES model assimilating Ocean Colour satellite observations; forced by GLO12v3 and providing 14-day weekly forecast; is available on 1/4° regular grid since July 2019, - GLOBAL 1/10° wave forecast (WAVEv2); based on MFWAM model assimilating SWH and directional wave spectra from Sentinel-1; forced by ECMWF operational analysis and forecast and GLO12v3 surface current and providing daily 5-day forecast; is available on 1/12° regular grid since April 2018. In December 2020, the forecast length was extended to 10-day.

2. STATUS AT THE END OF COPERNICUS1 During Copernicus 1 a lot of developments have been performed to improve and to optimise existing systems, to develop new products and new systems and to improve the consistency of existing systems in terms of resolution; time extension for reanalysis; or forecast length; available variables; model configuration; and parameterisations. Current near real time forecasting systems provide physical, waves, sea ice and biogeochemistry products with the following main characteristics and evolution during Copernicus 1: - GLOBAL 1/12° physical forecasting system (GLO12v3); based on NEMO model assimilating SLA, T/S profiles, SST, SIC; forced by ECMWF operational analysis and forecast and providing physical and sea ice 10-day daily forecast; is available since November 2016 on 1/12° regular grid. In April 2019, a new Surface and Merged

The global reanalysis systems are consistent with the near real time systems, using the same models and data assimilation methods and providing the same variables. The, main characteristics and evolution during Copernicus 1 are the following:

113

#57

- A GLOBAL 1/12° physical reanalysis (GLORYS12v1) is available since April 2018 and provides daily products starting in 1993 available on 1/12° regular grid. The multi system Global Reanalysis Ensemble Product (GREP) at lower resolution (1/4°) is covering the same period and is available since 2019. In 2020 all temporal extension of reanalysis timeseries were produced using the ERA5 atmospheric reanalysis instead of ERAinterim, - A GLOBAL 1/5° wave reanalysis (WAVERYSv1); based on the MFWAM model, assimilating reprocessed SWH observations and directional wave spectra from Sentinel-1; and forced by ECMWF ERA5 reanalyses. has been produced,. Daily products starting in 1993 are available on 1/4° regular grid since December 2019, - A GLOBAL 1/4° biogeochemical reanalysis (BIORYS4v4) is based on the PISCES model without DAdata assimilation and forced by physical simulation without data assimilationDA. Daily products starting in 1993 are available on 1/4° regular grid since July 2019 and extension of timeseries are provided using physical simulation forced by ERA5 atmospheric forcing since July 2020, - A GLOBAL 1/12° micronekton reanalysis (MICRORYSv2) is based on the Seapodym model without DA and is forced by GLORYS12v1 and Net Primary Production merged between satellite observation and BIORYS4v4. Weekly products starting in 1998 are available on 1/12° regular grid since May 2021. A previous version at lower resolution (1/4°) was developed and disseminated in 2019.

3. POST 2021 PERSPECTIVES

3.1 Update of the global reanalyses, analysis and forecasting physical systems Next versions of the global forecasting (GLO12v4) and reanalysis systems (GLORYS12v2) are under development, in order to improve, among others:, the representation of mesoscale activity, the mass/steric distribution (loss of mass and too much steric in GLO12v3 and GLORYS12v1), and equatorial dynamics which directly impact the biogeochemistry. To achieve these ambitious goals, focus iwas set toon developments in the ocean model, reanalyses and data assimilation method. Some steps in the developments for these versions of the systems are tested, other are ongoing. About models, we have implemented version 3.6 of NEMO and version 3 of LIM sea ice model are implemented. Consequently, the time-splitting mode is operated enabling

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

rapid barotropic waves in the system, high resolution spatial and temporal atmospheric forcing. About reanalyses, we have implemented the use of the interannual discharge of 13 major rivers based on GloFas data and CMEMS BRONCO service evolution project are used. Parametrization for ocean/atmosphere interaction (Renault et al., 2019) is updated as well as light penetration using a 5-bands formulation and vertical diffusion using a second order k-epsilon scheme (Reffray et al, 2015). About Regarding data assimilation, a 4D analysis with a new Mean Dynamic topography is now exploited and an updated data base of error covariances (computed from the GLORYS12v1 reanalysis) is leveraged. The temperature and salinity large scale bias correction method based on a 3Dvar analysis has been optimized. For assimilated observations, the main update concerned the SST. In the near real time system (GLO12v4), now the L3 Odyssea SST product is assimilated, and, in the reanalysis (GLORYS12V2), now the L4 OSTIA SST product is assimilated. A new Mean Dynamic Topography is also used to assimilate Sea Level Anomaly in the system. An ocean wave coupled approach as described in Law Chune and Aouf (2018) will be also investigated for reanalysis and forecast systems. Before launching the target system at 1/12° of resolution in 2022, several unitary tests based on development listed above have been performed in a “twin” global system at ¼°. Improvements have been quantified and will be documented with the new version of CMEMS products including impact on the biogeochemistry system. A significant development started during Copernicus 1 period is the improvement of atmospheric forcings used to force ocean real-time and reanalysis systems (based on meteorological forecasts and reanalysis produced by ECMWF). However, this approach has two main shortcomings. First, there is an inconsistency between atmospheric forcing and ocean surface conditions which can deteriorate ocean forecasts. Then, ocean feedback on atmospheric forcings is currently neglected, while it is now clearly established that ocean surface mesoscale features (such as eddies and fronts) significantly influence the atmosphere evolution. To avoid such issues, coupled ocean-atmosphere models represent a satisfying scientific solution despite having some limitations. For instance, the prohibitive numerical cost at high resolutions, and initialization issues related to the lack of coupled assimilation systems. Consequently, GLO MFC decided to develop an innovative solution consisting in the coupling of the ocean model to a reduced-complexity atmospheric boundary layer model named ABL1D (Lemarié et al, 2021). This model resolves only the necessary atmospheric processes to represent air-sea interactions accurately, while being driven by the meteorological forecast or reanalysis from ECMWF. Such a strategy offers significant benefits compared to forced ocean models and coupled models:

114

#57

- the lower atmosphere is adjusted coherently to ocean and sea-ice surface conditions produced by GLO MFC operational systems, - mesoscale coupled processes are represented at high resolution to improve atmospheric forcings, and consequently, ocean forecast and reanalysis, - t he numerical cost is insignificant compared to an atmospheric model. First evaluations of the ABL1D model have been conducted with multi-year coupled simulations. Results suggest that this coupled system is able to realistically simulate the ocean feedback, and also sea-ice effect on the atmospheric boundary layer. Ultimately, the significant effect of this coupling to the ocean kinetic energy (up to a depth of 1500 m) was demonstrated. These promising findings pave the way toward including the ABL1D model into the nextgeneration of operational systems developed in GLO MFC for Copernicus 2.

3.2 Long term perspective Dedicated developments already started for the next generation of forecasting systems with two main complementary purposes: - a resolution increase to fully resolve ocean meso scale in the global ocean, and -a quantification of forecast uncertainty based on ensemble forecast system. To reach the first goal a new global configuration has been developed to increase the spatial resolution for future Copernicus Marine Service global forecasting and reanalysis systems. This new configuration has a 1/36° horizontal resolution (2-3 km), which is three times finer than the present global system. First simulation tests were performed enabling quantification of HPC resources and optimisation (needed for an operational use). First runs

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

required 3.5 hours to compute a 7-day forecast using 30000 CPUs of the new HPC ATOS BULL Sequana XH2000 supercomputer available at Meteo France. Data assimilation in the global high resolution system at 1/36° will be also developed to assimilate higher resolution observation and especially SWOT sea level anomaly observations. Concerning the second goal, an Ensemble Prediction System (EPS), which are forecasts running several numerical models to estimate the system’s uncertainty, has been developed based on a coarser resolution model, at 1/4°, weekly producing a 30-day ensemble forecast with 50 members. Currently, the ensemble is generated using the 50 members of the ensemble atmospheric forecast from ECMWF combined with 5 ocean initial states including ±2-day lag. An assessment of the ensemble spread and forecast skill is still needed before a transition to an operational chain. Evolution of the EPS system, based on NEMO3.6 standard version and coupled with a module to simulate stochastic perturbations developed in Brankart (2015), is under development. A Local Ensemble Transformed Kalman Filter is used to calculate corrections to the ensemble of model trajectories. This method takes an ensemble of model forecasts and calculates an ensemble of analyzed model trajectories using a transformation matrix that is a function of: the ensemble dispersion, observation errors and the misfit between model and observations. Some recent experiments show that the ensemble approach can improve forecast skill by 7-20%, depending on the variable. To go further, ongoing R&D activities related to ensemble methods are the uptake of two developments: a multiscale analysis focusing on the large-scale processes (Tissier et al., 2019) and an ensemble modulation method increasing the subspace rank spanned by the ensemble (Brankart, 2019). Ensemble approach will be also used to assimilate sea ice observation and for the biogeochemistry system.

115

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS The authors acknowledge all the Global MFC team members who have contributed to products from the development phase to the operation including: J. Le Sommer, P. Brasseur, JM Brankart, P. Lehodey, B. Tranchant, F. Lyard, F. Lemarié, H. Giordani, the NEMO system team in charge of development, distribution and support of the shared NEMO reference; and Mercator Ocean International colleagues in charge of development, validation and operation activities as well as service desk and technical and cross cutting coordination of Copernicus Marine Service.

REFERENCES: 1. Aouf, L., Hauser, D., Chapron, B., Toffoli, A., Tourrain, C., & Peureux, C. (2021). New directional wave satellite observations: toward improved wave forecasts and climate description in Southern Ocean. Geophysical Research Letters, 48, e2020GL091187. https://doi. org/10.1029/2020GL091187 2. Ardhuin F, Rogers E, Babanin AV, Filipot JF, Magne R, Roland A, vander Westhuysen A, Queffeulou P, Lefevre JM, Aouf L, Collard F (2010) Semiempirical dissipation source functions for ocean waves. Part I: definition, calibration, and validation. J Phys Oceanogr 40:1917–1941. https://doi. org/10.1175/2010JPO4324.1 3. Bidlot, J.R. and Holt, M.W., 2006: Verification of Operational Global and Regional Wave Forecasting Systems against Measurements from Moored buoys, JCOMM Technical Report No. 30 4. Brankart, J.-M., Candille, G., Garnier, F., Calone, C., Melet, A., Bouttier, P.-A., Brasseur, P., and Verron, J.: A generic approach to explicit simulation of uncertainty in the NEMO ocean model, Geosci. Model Dev., 8, 1285–1297, https://doi.org/10.5194/gmd-81285-2015, 2015.

5. Brankart J-M (2019) Implicitly Localized MCMC Sampler to Cope With Non-local/Non-linear Data Constraints in Large-Size Inverse Problems. Front. Appl. Math. Stat. 5:58. doi: 10.3389/ fams.2019.00058 6. Garric G., L. Parent, E. Greiner, M. Drévillon, M. Hamon, JM Lellouche, C. Régnier, C. Desportes, O. Le Galloudec, C. Bricaud, Y. Drillet, F. Hernandez, C. Dubois and P-Y. Le Traon (2018), Performance and quality assessment of the global ocean eddy-permitting physical reanalysis GLORYS2V4 Operational Oceanography serving Sustainable Marine Development. Proceedings of the Eight EuroGOOS International Conference. 3-5 October 2017, Bergen, Norway. E. Buch, V. Fernandez, G. Nolan and D. Eparkhina (Eds.) EuroGOOS. Brussels, Belgium. 2018. ISBN 978-2-9601883-3-2. 516 pp.

7. Gasparin, F., Cravatte, S., Greiner, E., Perruche, C., Hamon, M., Van Gennip, S., & Lellouche, J. M. (2021). Excessive productivity and heat content in tropical Pacific analyses: Disentangling the effects of in situ and altimetry assimilation. Ocean Modelling, 160, 101768.

11. Lehodey, P., Conchon, A., Senina, I., Domokos, R., Calmettes, B., Jouanno, J., Hernandez, O., and Kloser, R. (2015) Optimization of a micronekton model with acoustic data. – ICES Journal of Marine Science, 72(5): 13991412. doi: 10.1093/icesjms/ fsu233.

8. Julien Lamouroux, Coralie Perruche, Alexandre Mignot, Julien Paul, Camille Szczypta (2019). QUALITY INFORMATION DOCUMENT For Global Biogeochemical Analysis and Forecast Product GLOBAL_ANALYSIS_FORECAST_ BIO_001_028. CMEMS report issue 1.0. http://marine. copernicus.eu/documents/QUID/ CMEMS-GLO-QUID-001-028.pdf

12. Lellouche, J.-M., Greiner, E., Le Galloudec, O., Garric, G., Regnier, C., Drevillon, M., Benkiran, M., Testut, C.-E., Bourdalle-Badie, R., Gasparin, F., Hernandez, O., Levier, B., Drillet, Y., Remy, E., and Le Traon, P.-Y.: Recent updates to the Copernicus Marine Service global ocean monitoring and forecasting real-time 1∕12° high-resolution system, Ocean Sci., 14, 1093– 1126, https://doi.org/10.5194/ os-14-1093-2018, 2018.

9. Law Chune, S., & Aouf, L. (2018). Wave effects in global ocean modeling: parametrizations vs. forcing from a wave model. Ocean Dynamics, 1-20. 10. Stéphane Law-Chune, Lotfi Aouf, Alice Dalphinet, Bruno Levier, Yann Drillet, et al., WAVERYS: a CMEMS global wave reanalysis during the altimetry period. Ocean Dynamics, Springer Verlag, 2021, (10.1007/s10236-02001433-w). (meteo-03131602)

116

13. Lellouche Jean-Michel, Greiner Eric, Bourdallé-Badie Romain, Garric Gilles, Melet Angélique, Drévillon Marie, Bricaud Clément, Hamon Mathieu, Le Galloudec Olivier, Regnier Charly, Candela Tony, Testut Charles-Emmanuel, Gasparin Florent, Giovanni Ruggiero, Mounir Benkiran, Drillet Yann, Le Traon PierreYves, 2021. The Copernicus global 1/12° oceanic and sea ice GLORYS12 reanalysis. Submitted to Frontiers.

#57

14. Lemarié, F., Samson, G., Redelsperger, J. L., Giordani, H., Brivoal, T., & Madec, G. (2021). A simplified atmospheric boundary layer model for an improved representation of air–sea interactions in eddying oceanic models: implementation and first evaluation in NEMO (4.0). Geoscientific Model Development, 14(1), 543-572. 15. MacLachlan C., A. Arribas, K.A. Peterson, A. Maidens, D. Fereday, A.A. Scaife, M. Gordon, M. Vellinga, A. Williams, R. E. Comer, J. Camp and P. Xavier, 2015. Description of GloSea5: the Met Office high resolution seasonal forecast system. Q. J. R. Met. Soc., DOI: 10.1002/ qj.2396.

16. Mignot, A., Claustre, H., Cossarini, G., D’Ortenzio, F., Gutknecht, E., Lamouroux, J., Lazzari, P., Perruche, C., Salon, S., Sauzède, R., Taillandier, V., and Terruzzi, A.: Defining BGC-Argo-based metrics of ocean health and biogeochemical functioning for the evaluation of global ocean models, Biogeosciences Discuss. [preprint], https:// doi.org/10.5194/bg-2021-2, in review, 2021. 17. Reffray, G., BourdalleBadie, R., and Calone, C.: Modelling turbulent vertical mixing sensitivity using a 1-D version of NEMO, Geosci. Model Dev., 8, 69–86, https://doi. org/10.5194/gmd-8-69-2015, 2015. 18. Renault, L., Lemarié, F., & Arsouze, T. (2019). On the implementation and consequences of the oceanic currents feedback in ocean– atmosphere coupled models. Ocean Modelling, 141, 101423.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

19. Sauzède, R., Bittig, H. C., Claustre, H., Pasqueron de Fommervault, O., Gattuso, J. P., Legendre, L., & Johnson, K. S. (2017). Estimates of water-column nutrient concentrations and carbonate system parameters in the global ocean: a novel approach based on neural networks. Frontiers in Marine Science, 4, 128. 20. Storto, A. and Masina, S.: C-GLORSv5: an improved multipurpose global ocean eddypermitting physical reanalysis, Earth Syst. Sci. Data, 8, 679–696, https://doi.org/10.5194/essd-8679-2016, 2016. 21. Storto A. et al, 2019. The added value of the multi-system spread information for ocean heat content and steric sea level investigations in the CMEMS GREP ensemble reanalysis product. July 2019. Climate Dynamics 53(1). DOI: 10.1007/ s00382-018-4585-5

117

22. Tissier, A.-S., Brankart, J.-M., Testut, C.-E., Ruggiero, G., Cosme, E., and Brasseur, P.: A multiscale ocean data assimilation approach combining spatial and spectral localisation, Ocean Sci., 15, 443– 457, https://doi.org/10.5194/ os-15-443-2019, 2019 23. Zuo, H., Balmaseda, M. A., Tietsche, S., Mogensen, K., & Mayer, M. (2019). The ECMWF operational ensemble reanalysis-analysis system for ocean and sea-ice : a description of the system and assessment. Ocean Science, (January), 1–44.

SYNTHESIS OF ACHIEVEMENTS FROM THE GLOBAL COUPLED MONITORING AND FORECASTING CENTRE HARRIS, C.

Met Office, FitzRoy Road, Exeter, EX1 3PB, UK

OVERVIEW

1. MAIN ACHIEVEMENTS

The Global Coupled Monitoring and Forecasting Centre (GLOCPL) provides a single product in the CMEMS catalogue which differs from the other near-real-time global physical products in that the marine forecasts are delivered from the NEMO ocean component of a coupled ocean-atmosphere system. Details of this system and how it is initialized have changed significantly during Copernicus 1 (see below) and a further upgrade is nearly ready. However, ocean resolution remains at ¼° and there are still neither biogeochemical nor wave components in the system. Datasets delivered to users every day include: daily mean temperature, salinity, sea surface height, currents, mixed layer depth, and sea ice concentration and thickness (provided for both analysis and 10 days of forecast). Since July 2017, datasets have been complemented by: hourly instantaneous sea surface height, sea surface temperature and surface currents. The analysis is updated the following day to make use of latearriving observations.

At the start of Copernicus 1, the GLO-CPL product was delivered from two separate systems. Forecasts were provided from the Met Office GloSea5 coupled seasonal forecast system (MacLachlan et al., 2015) while analyses came from the FOAM ocean-only system (Blockley et al., 2014) which is the same system used to initialise the ocean in GloSea5. These systems shared an almost identical ocean and sea ice science configuration (using the NEMO model coupled to the multi-thickness-category sea ice model CICE), with the FOAM global ocean configuration being forced (using CORE bulk formulae to specify the surface boundary condition) by Met Office global atmospheric Numerical Weather Prediction (NWP) fields. The NEMO global ocean configuration used the tripolar ORCA025 grid (with a 1/4° or 28 km horizontal grid spacing at the equator, reducing to 7 km at high southern latitudes, and ~10 km in the Arctic Ocean). The scientific configuration of the Met Office Unified Model used as the atmosphere component of the GloSea5 system was near identical to the NWP system providing the FOAM forcing fields, although the latter had a higher resolution (~17 km rather than ~50 km). The main disadvantage of this system was that only the ocean forecast was delivered from an interactively coupled ocean-atmosphere system. This meant there was a high overhead in keeping uncoupled and coupled systems scientifically consistent to reduce the likelihood of initialisation shocks between uncoupled analyses and coupled forecasts. Control over diagnostics provided from the forecasts was also more limited due to the use of the GloSea5 system.

The operational GLO-CPL system has been upgraded regularly during Copernicus 1 to make the best use of both in-situ and new satellite observations (particularly Sentinel-3) and ensure continued robustness. Following an upgrade introducing weakly coupled data assimilation to initialise the GLO-CPL forecasts, work during the later part of the period has focussed on a future transition to a combined ocean forecasting and weather prediction system. This will benefit from a higher resolution atmosphere, as well as providing potential opportunities for ensemble forecasting and, ultimately, more strongly coupled data assimilation.

The most significant upgrade during Copernicus 1 was in July 2017 with the introduction of a ‘weakly coupled’ data assimilation system (Lea et al., 2015; Guiavarc’h et al., 2019) to initialise the coupled forecasts. These are now produced from a new integrated operational system rather

118

#57

than relying on seasonal forecasts; this facilitated the extension of ocean forecasts to 10 days and the introduction of some higher frequency hourly diagnostics which were a strong user requirement and are now some of the most downloaded datasets. In this context ‘weakly coupled’ means the assimilation background states for both ocean and atmospheric components are from the coupled model. However, the data assimilation codes are run independently (without any opportunity for atmosphere observations to directly affect ocean increments on the same assimilation cycle) before increments are applied back into the coupled model with an incremental analysis update (IAU) step. As a consequence, the ocean data assimilation can be treated in a very similar way to the ocean-only FOAM system used previously. It continues to use NEMOVAR, a variational (3D-var) scheme developed specifically for NEMO and further tuned for the 1/4° resolution global model. Key features of NEMOVAR are multivariate relationships (specified through a linearized balance operator) and the use of an implicit diffusion operator to model background error correlations. The use of a weakly coupled data assimilation system necessitated a change from 6-hour to 24-hour assimilation windows in the ocean for consistency with the atmosphere. At present the ocean data assimilation has not been

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

specifically tuned to account for this, although there would be potential benefit from such work in future. Multiple ‘update cycles’ are used to mimic the previous behaviour where a ‘best analysis’ (assimilating as many ocean observations as possible) is made available a day later than the near-real-time analysis used to initialise the daily forecasts. The complete consistency between ocean analysis and forecast in the upgraded system benefits users, particularly those who are concerned with short lead-time forecasts. Significant upgrades in marine observations assimilated during Copernicus 1 were motivated both by newly available observations (with the potential to improve product quality and system robustness) and changes to the technical characteristics of observational products. In both cases careful testing is required to understand the technical and scientific implications prior to final assimilation. For the most part it was not possible to demonstrate a ‘step change’ in product quality at the point assimilation of these observations was initially activated. However, subsequent degradation or discontinuity of other assimilated products means that, without the Sentinel-3 satellite observations in particular, the GLO-CPL product quality would have gradually degraded.

Figure 1: ‘Class 4’ temperature profile statistics for the ‘best analysis’ against Argo for the current GLO-CPL system (labelled as CPLDA; also shown are the Met Office FOAM and Mercator Ocean PSY4 systems for comparison); bias and root-mean-square deviation (RMSD) are averaged over depth; note that prior to July 2017 the GLO-CPL analysis was provided from the FOAM system.

119

#57

In 2015, assimilated observations were: - s atellite SST data (AVHRR data supplied by the GHRSST project), - i n-situ SSTs from moored buoys, -d rifting buoys and ships (considered unbiased and used as a reference for satellite SST bias correction), - s ea level anomaly from Jason-2, Cryosat-2, and SARAL/AltiKa, - s ub-surface temperature and salinity profiles from Argo, underwater gliders, moored buoys, sensors carried by marine mammals and manual profiling methods, - s ea ice concentration (SSMIS data provided by OSI SAF as a daily gridded product), -T he main upgrades to observation satellite usage during Copernicus 1 have been the assimilation of: Jason-3 (2016) and Sentinel-3A & B (2017 & 2019) sea level anomaly observations, -A MSR2 (2016), Suomi-NPP & NOAA-20 VIIRS (2017 & 2019) and Sentinel-3A & B SLSTR (2019) sea surface temperature data. These continual upgrades to observation usage, in combination with model upgrades, have ensured product quality has either been maintained or improved over time. Figure 1 shows an example of the improved ‘Class 4’ statistics (Ryan et al., 2015) for temperature profiles against Argo data. Note that this specific comparison to PSY4 (the GLO high resolution system) for temperature profiles favours GLO-CPL

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

and FOAM, whilst PSY4 shows improved error statistics for other variables, such as sea level anomaly or currents. A major achievement during Copernicus 1 has been the operational robustness of the GLO-CPL system, particularly following the upgrade to weakly coupled data assimilation. Very good back-up and contingency procedures are in place and, as a result, no extended outages have occurred and most delays to delivery have only been 1-2 hours. The system benefits from being relatively self-contained with no dependence on external forcing data or boundary conditions. For the last three years, the focus of research and development of the GLO-CPL system has been on the next major upgrade. This is now expected just after the end of Copernicus 1 and is represented by the schematic in Figure 2. It will involve delivery of the GLO-CPL product from a coupled atmosphere-ocean system which is also being used for Numerical Weather Prediction. This involves many changes including upgrade of atmospheric resolution (to 10 km) and the inclusion of an ensemble at lower atmospheric resolution (20 km), both of which have the potential to deliver improvements in ocean products. The ocean component is also being upgraded to be consistent with the configuration now operational in the ocean-only FOAM system at the Met Office. Amongst other enhancements, this includes a new scientific configuration (‘GO6’, Storkey et al., 2018) on the extended ORCA025 grid and an improved variational bias correction scheme for SSTs (While and Martin, 2019).

Figure 2: Schematic showing some of the main changes developed for the upcoming upgrade of the GLO-CPL system (labelled as ‘CPLDA’). Pending final testing it is likely the atmospheric configuration will be GA8 (rather than GA7.1 as shown). The addition of the coupled ensemble provides great potential for future ocean developments.

120

#57

Although the system described above for implementation in 2022 includes an interactive ocean in the ensemble part, it does not involve the generation of a realistic ocean ensemble. To account for this, SST perturbations seen by the atmosphere are generated, and applied, in exactly the same way as when running atmosphere-only models. However, in preparation for a later upgrade, a global ocean and sea-ice ensemble system has been developed. This is based on the ocean-only FOAM system with ensemble members forced by different members of an atmospheric ensemble. Ensemble spread is also driven by perturbed observation locations and values (linking with work in the GENOA service evolution R&D project),

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

stochastic model perturbations (which will benefit from the NEMO code development in the SCRUM2 project) and an ensemble inflation scheme. Results from tests of hybrid ensemble/3DVar data assimilation (where the NEMOVAR code is used to combine the existing representation of background error covariances with a localised estimate of the daily-varying sample error covariances coming from the ensemble) have been very promising. Figures 3 and 4 show an example for sea level anomaly assimilation statistics: the hybrid DA significantly improves the deterministic model, and there are further benefits from using the ensemble mean instead of the unperturbed member.

Figure 3: Global sea level anomaly root-mean-square deviation (RMSD) compared with altimeter observations before their assimilation. The black line shows the error in the standard 3DVar unperturbed member (equivalent to the current FOAM system). The blue line shows the error in the ensemble mean using standard 3DVar. The red line shows the error in the hybrid DA unperturbed member and the green line shows the error in the ensemble mean using the hybrid DA.

121

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Global sea level anomaly root-mean-square deviation (RMSD) compared with altimeter observations before their assimilation. The map shows the percentage change in the RMSD of the innovations of the hybrid DA ensemble mean compared with the standard 3DVar (blue implies the hybrid DA has lower RMSD).

estimate forecast error covariance characteristics in ORCA12 and to improve HPC efficiency.

2. POST 2021 PERSPECTIVES The immediate priority beyond Copernicus 1 is operational implementation, expected by early 2022, of the upgrade to deliver ocean products from the coupled NWP system (currently in final testing). Several drivers exist for future development and upgrades after that. These are particularly around enhancing the representation of processes in the surface ocean as this will both directly improve ocean product quality, and have most benefit for improving NWP performance (and therefore also atmospheric forcing) in the coupled system. Given the relatively low eddypermitting 1/4° ocean resolution used at present there is a clear encouragement to increase it toward an eddyresolving resolution. This would enable: -m ore accurate feature representation for surface currents and sea surface temperatures, -m ore effective use of high-resolution satellite observations, - i mproved feedbacks between the oceanic and atmospheric boundary layers. An upgrade to 1/12° would allow the GLO-CPL system to benefit from work in the IMMERSE H2020 project to

Other user drivers include improved analyses and forecasts in Arctic regions, the ability to provide reliable probabilistic ocean products based on an ensemble of ocean configurations, and the provision of more accurate surface ocean predictions as a result of better representation of vertical mixing, tides and wave effects. Examples of likely developments to help address these requirements include assimilation of sea ice thickness data (within a new SI3 ice configuration replacing CICE as used currently), and implementation of the ocean ensemble generation described above within the GLO-CPL system. Other developments include an upgrade to the NEMO configuration to use a z~ type Arbitrary LagrangianEulerian vertical coordinate (to help reduce spurious numerical mixing associated with vertical motion; this would build on work within the RENUMERATE and IMMERSE projects). A wave component may also be added but this is not an immediate priority. The system will also be funded and used for global weather prediction. Hence, there will be a whole range of other drivers and constraints on the development and evolution of the GLO-CPL system to ensure the atmospheric performance also continues to improve due to model, observation, assimilation and ensemble method upgrades.

122

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS The author acknowledges all those who have, while working at the UK Met Office, contributed to the development of the GLO-CPL system and evaluation of product quality including D. Lea, M. Martin, I. Mirouze, C. Guiavarc’h, J. Roberts-Jones, M. Price, I. Ascione, J. Maksymczuk, A. Ryan and K. Smout-Day.

REFERENCES: Blockley, E.W., Martin, M.J., McLaren, A.J., Ryan, A.G., Waters, J., Lea, D.J., Mirouze, I., Peterson, K.A., Sellar, A. and Storkey, D. (2014) Recent development of the Met Office operational ocean forecasting system: an overview and assessment of the new global FOAM forecasts. Geoscientific Model Development, 7, 2613–2618. https://doi.org/10.5194/gmd-72613-2014

Guiavarc’h, C., RobertsJones, J., Harris, C., Lea, D. J., Ryan, A., and Ascione, I. (2019) Assessment of ocean analysis and forecast from an atmosphere–ocean coupled data assimilation operational system, Ocean Science, 15, 1307–1326. https://doi.org/10.5194/os-151307-2019 Lea, D. J., Mirouze, I., Martin, M. J., King, R. R., Hines, A., Walters, D., and Thurlow, M. (2015) Assessing a new data assimilation system based on the Met Office coupled atmosphere-land-ocean-sea ice model, Monthly Weather Review, 143, 4678–4694. https://doi. org/10.1175/MWR-D-15-0174.1

MacLachlan, C., Arribas, A., Peterson, K.A., Maidens, A., Fereday, D., Scaife, A.A., Gordon, M., Vellinga, M., Williams, A., Comer, R.E., Camp, J., Xavier, P. and Madac, G. (2015) Global Seasonal forecast system version 5 (GloSea5): a highresolution seasonal forecast system. Quarterly Journal of the Royal Meteorological Society, 141, 1072–1084. https:// doi.org/10.1002/qj.2396 Ryan, A.G., Regnier C., Divakaran, P., Spindler, T., Mehra, A., Smith, G.C., Davidson, F., Hernandez F., Maksymczuk, J., & Liu, Y. (2015) GODAE OceanView Class 4 forecast verification framework: global ocean intercomparison, Journal of Operational Oceanography, 8, 98-111. https://doi.org/10.1080/175587 6X.2015.1022330

123

Storkey, D., Blaker, A. T., Mathiot, P., Megann, A., Aksenov, Y., Blockley, E. W., Calvert, D., Graham, T., Hewitt, H. T., Hyder, P., Kuhlbrodt, T., Rae, J. G. L., and Sinha, B. (2018) UK Global Ocean GO6 and GO7: a traceable hierarchy of model resolutions. Geoscientific Model Development, 11, 3187– 3213. https://doi.org/10.5194/ gmd-11-3187-2018 While, J. and Martin, M.J. (2019) Variational bias correction of satellite seasurface temperature data incorporating observations of the bias. Quarterly Journal of the Royal Meteorological Society, 145, 2733– 2754. https://doi. org/10.1002/qj.3590

SYNTHESIS OF ACHIEVEMENTS FROM THE ARCTIC MARINE FORECASTING CENTER

BERTINO, L.1, ALI, A.2, CARRASCO, A.3, LIEN, V.S.4, MELSOM, A.3 Nansen Environmental and Remote Sensing Center, Bergen, Norway - 2MET Norway, Bergen, Norway - 3MET Norway, Oslo, Norway 4 Institute of Marine Research, Bergen, Norway

OVERVIEW During the 2015-2021 period, the Copernicus Marine Service has diversified its Arctic portfolio of modeling products. The addition of waves, tides and ocean carbon variables satisfies more adequately user needs in the industry, academia and public sectors. Many validation metrics have also been introduced, providing more intuitive forecast quality measures. The resolution of several products has increased, such as ice forecasts horizontal resolution thanks to a standalone sea ice model based on a new rheology. At the end of the Copernicus 1 period the products are using the model and assimilation systems TOPAZ, neXtSIM and WAM.

1. MAIN ACHIEVEMENTS 2015-2021 When Copernicus Services started in 2015, the Arctic MFC provided four forecasts and reanalyses products of physical and biogeochemical variables. These products were based on the TOPAZ system, which leverages Ensemble Kalman Filter (EnKF) data assimilation applied to satellite ocean observations (SLA, SST), sea ice observations (concentration and drift) and in situ T/S profiles (from Argo and Ice-Tethered Profilers) in a coupled physical-biogeochemical model. Using an advanced EnKF assimilation algorithm in operational settings is still today a unique achievement. The HYbrid vertical Coordinate Ocean Model (HYCOM) was the ocean model, coupled to the CICE sea ice model using an Elasto-Viscous-Plastic rheology and coupled online to

the NORWECOM biogeochemical model. All products had a resolution of 12.5 km or coarser - interpolated to a polar stereographic projection - and 28 hybrid z-isopycnic layers, interpolated to 12 “Levitus’’ vertical levels. There was no nesting between Global and Arctic systems

1.1 Waves A pan-Arctic operational wave forecast product (see domain on Figure 1) has been first setup using the WAM model code from the MyWave FP7 project. The code has been modified by MET Norway to allow wave propagation under sea ice (Sutherland et al., 2019). Sea ice concentration, ice thickness and surface currents are all extracted from the Arctic MFC physical forecast. In 2019, the model horizontal resolution increased from 8 km to 3 km and two forecasts were run daily with an horizon of 5 and 10 days respectively. A wave hindcast was later added to the CMEMS catalogue at 3 km resolution with an updated version of WAM. It included new physics, a mean wavenumber and mean frequency reformulation as well as a new method to detect freak waves. The code has been enhanced as well by correcting wave growth in very high winds and by allowing wave propagation under sea ice. A sub-grid scale parametrization of “obstructions’’ is used. On the surface, WAM is forced by hourly winds merging the ERA5 reanalysis and a local refinement (shown as the rectangle in Figure 1) by a 2.5 km non-hydrostatic convection-permitting model. The wave products are exploited for navigation purposes, support to offshore operations and downscaling to coastal wave models, among other uses. The 3 km resolution products provides high enough quality to fill the mandate of the Norwegian preparedness services (search and rescue, oil spill response) and superseded pre-existing national systems.

124

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: Domain of the Arctic wave model. Shading colours are significant wave height Hs in meters. In the rectangle winds hindcast come from the Harmony-Arome model with 2.5 km resolution.

1.2 Ocean physics High-frequency signals (tides and storm surges) were introduced in March 2020, with a pan-Arctic 3 km configuration of the 3-dimensional HYCOM-CICE model. At lateral boundaries - which are close to the wave model boundaries shown in Figure 1 - Global-HR MFC forecasts are combined with tidal heights and currents computed from the FES2014 tidal database, including 34 tidal constituents.

The model is intended to become the main workhorse for ocean physical and biogeochemical forecasts. It has therefore been set up with 50 hybrid z-isopycnic layers. Hence, in addition to surface tide forecasts, the model is capable of internal tides prediction. As for the wave model , the resolution of 3 km is also adequate for the Norwegian national mandate and has replaced a pre-existing national forecast system. It provides boundary conditions to coastal models around mainland Norway and Svalbard.

125

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

1.3 Sea ice rheology

standalone forecast mode for the Central Arctic including a nudging term to daily satellite sea ice concentrations.

The TOPAZ4 reanalysis has shown a lack of sensitivity of the rheological model. A new sea ice model based on the Brittle-Bingham-Maxwell rheology has thus been developed in a Lagrangian coordinate (the neXtSIM model, Rampal et al., 2016) to improve sea ice drift and other related sea ice properties. This model has been set up in

NeXtSIM-F forecasts show much more detailed sea ice features than TOPAZ4 (Figure 2, leads and landfast ice in particular are not visible on TOPAZ4) and their motions are more accurately forecasted, with drift distance errors cut from 8 to 4 km per day. Sea ice forecasts are used in navigation services.

Figure 2: Sea ice thickness on the 12th March 2021 from the TOPAZ4 system and the recently introduced neXtSIM-F forecast (right).

1.4 Biogeochemical modeling The biogeochemical model coupled to the ocean model has been updated twice during Copernicus 1.0. The first upgrade in April 2016 replaced NORWECOM with ECOSMO, where parameters were re-tuned to avoid an excessive amplitude of the Spring bloom. In a second upgrade in May 2021, several changes were brought to ECOSMO: -d oubling of both horizontal and vertical resolution (6 km and 50 hybrid layers), - s imple assimilation of satellite surface Chlorophyll data (Uitz et al., 2006), - i nclusion of the carbon cycle, - i nclusion of light transmission through sea ice, - i mprovements of the model inputs (rivers discharge from Arctic-HYPE model, atmospheric deposition of nutrients from EMEP model and lateral boundary conditions from Global MFC model PISCES), - t he FABM software now couples ECOSMO to HYCOM.

When compared to independent Chlorophyll profiles from BGC-Argo buoys in the Nordic Seas, the assimilation single-handedly reduces errors drastically (Figure 3). The primary production accuracy is an important prerequisite for carbon cycle simulation and thereby provides up to date information about the ocean carbon pump and ocean acidification. About data assimilation, the biogeochemical reanalysis adopted an Ensemble Kalman Smoother (EnKS) to assimilate both satellite surface Chlorophyll data and nutrient profiles. The EnKS optimizes biogeochemical model parameters in ECOSMO using data from posterior week and can correct the timing of the Spring bloom. The resulting reanalysis product is the first demonstration of an EnKS in CMEMS.

126

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Comparison of Chlorophyll profiles to BGC-Argo buoys (panel a). Note the inclusion of surface Chlorophyll from satellite in January 2017 and the logarithmic scale for concentrations.

1.5 Physical data assimilation The Arctic MFC started assimilating sea ice thickness products with the thin ice product from SMOS in both reanalysis and forecast products using the EnKF in 2017. The merged product from CryoSAT-2 and SMOS was then assimilated, first in reanalysis, then in near-real time in November 2020. The resulting improvement of sea ice thickness persists a few months through the summer when satellite products are unavailable. The physical reanalysis product was updated with the following: - doubling of HYCOM ocean model vertical resolution, - replacement of time-mean model by CNES/CLS Mean SSH Rio2018 reference to assimilate sea level anomalies, - inclusion of freshwater discharge related to the Greenland mass loss, - improvement of salinity profiles assimilation. Systematic assimilation of ESA CCI products throughout the whole reanalysis period, removing discontinuities in the previous physical reanalysis (Xie et al., 2017). The new reanalysis product should therefore be better suited for climate studies.

The validation of wave parameters uses satellite altimeter data (Bohlinger et al., 2019) for both forecast and multiyear products. Compared to the validation with wave buoys, results are now much more representative of the Arctic.

1.7 Ocean Monitoring Indicators The Nordic Seas is an area for key climatic processes in the North Atlantic. The ARC MFC has therefore established two sets of Ocean Monitoring Indicators that monitor North Atlantic - Arctic Ocean exchanges through the Nordic Seas. The first is the exchange of water across the straits that separate the two basins. Moorings are available there for validation. Then, the sea ice export from the central Arctic to the south was later included since it makes up an important part of sea ice budget in the Arctic. Ocean monitoring indicators thus make highly valuable data accessible to a large number of users interested in the Arctic, without the need to download discouraging amounts of data.

1.6 Enhanced validation Objective forecast evaluation metrics are provided monthly to Copernicus Marine Service for dissemination. In addition, products are monitored on a weekly basis by our team. Due to its relevance for operations in the Arctic, ice edge position is particularly scrutinized using two metrics: integrated ice edge error (IIEE) and fractions skill score (FSS). Melsom et al., (2019) have reviewed these metrics.

127

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. STATUS AT END OF COPERNICUS 1

between waves and the ocean model can also be improved using wave input terms into the ocean model (Ali et al., 2019).

The Arctic MFC now offers twice as many products as initially and include waves, tides and ocean carbon variables. New products have up to 4 times higher resolution compared to 2015, both horizontally and vertically and adhere to Copernicus Marine Service standard naming conventions. Products offered at the end of Copernicus 1.0 have improved performance, more targeted quality checks and easy access to important monitoring indicators, which make them better suited to user needs.

Ocean forecasts also need improved bathymetry data around Greenland and near-real-time forecasts of river discharge as from the Arctic-HYPE hydrological model. The ocean reanalysis should assimilate sea level anomalies from the SWOT mission, sea surface salinities from the SMOS mission as prepared in the ESA Arctic+ Salinity project. The standalone sea ice model should also include the assimilation of sea ice deformations from Sentinel-1 SAR ice drift. When available, ocean and sea ice data from the High-Priority Copernicus Missions CIMR and CRISTAL should be assimilated too.

3. POST-2021 PERSPECTIVES After having introduced a few independent products, it will be necessary to improve their mutual consistency. The first step should be to provide the physical forecast at higher horizontal and vertical resolution. A second step will be to synchronize the slow variability of the tidal model to the data assimilative ocean forecast model. The consistency

We also plan to distribute ensemble forecasts from the TOPAZ system, and improve their uncertainty estimates by matching them to ensemble predictions from the ECMWF. Monitoring the quality of sea ice forecasts can also be further enhanced (Palerme et al., 2019). Ocean biogeochemistry products would benefit from a longer (3 decades) multiyear timeseries. The ECOSMO model would be improved by using a more advanced sinking scheme developed in the SE project ZOOMBI and should include a sea ice biogeochemistry model.

128

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS Authors acknowledge all Arctic MFC team members who contributed to our products, their quality evaluation and monitoring, including A. Samuelsen, J. Xie, T. Wakamatsu, V. Ç. Yumruktepe, M. Müller, T. Williams, R. P. Raj, P. Bohlinger, A. Burud, M. Øiestad, H. Berge, M. Svanevik, J. A. Johannessen, J. Bruggeman, P. Rampal, Ø. Breivik, G. Sutherland and B. Hackett.

REFERENCES: Ali, A., Christensen, K. H., Breivik, Ø., Malila, M., Raj, R. P., Bertino, L., et al., (2019). A comparison of Langmuir turbulence parameterizations and key wave effects in a numerical model of the North Atlantic and Arctic oceans. Ocean Modelling, 137, 76–97, https://doi.org/10.1016/j. ocemod.2019.02.005 Bohlinger, P., Breivik, Ø., Economou, T. and Müller, M.: A novel approach to computing super observations for probabilistic wave model validation, Ocean Model., 139(May), 101404, doi:10.1016/j. ocemod.2019.101404, 2019.

Melsom A. Edge displacement scores, The Cryosphere, 15, 3785–3796, https://doi. org/10.5194/tc-15-3785-2021, 2021. Melsom A., Palerme C., Müller M. 2019. Validation metrics for ice edge position forecasts. Ocean Sci., 15, 615630. doi:10.5194/os-15-6152019 Palerme C., Müller M., Melsom A. 2019. An intercomparison of skill scores for evaluating the sea ice edge position in seasonal forecasts. Geophys. Res. Lett., 46, 4757-4763. doi:10.1029/2019GL082482

Williams, T., Korosov, A., Rampal, P., and Ólason, E.: Presentation and evaluation of the Arctic sea ice forecasting system neXtSIM-F, The Cryosphere, 15, 3207–3227, https://doi.org/10.5194/tc-153207-2021, 2021. Sutherland, G., Rabault, J., Christensen, K. H., & Jensen, A. (2019). A two layer model for wave dissipation in sea ice. Applied Ocean Research, 88, 111-118. DOI:10.1016/J. APOR.2019.03.023

129

Uitz, J., Claustre, H., Morel, A., Hooker, S.B., 2006. Vertical distribution of phytoplankton communities in open ocean: an assessment based on surface Chlorophyll. J. Geophys. Res. 111, C08005 (Doi:10.1029/2005JC003207). Xie, J., Bertino, L., Counillon, F., Lisæter, K. A. and Sakov, P.: Quality assessment of the TOPAZ4 reanalysis in the Arctic over the period 1991–2013, Ocean Sci., 13(1), 123–144, doi:10.5194/os-13-123-2017, 2017.

SYNTHESIS OF ACHIEVEMENTS FROM THE BALTIC SEA MONITORING AND FORECASTING CENTER

HUESS, V.1, JANDT-SCHEELKE, S.2, KÄRNÄ, T.3, LAGEMAA, P.4, LINDENTHAL, A.2, LORKOWSKI, I.2, MALJUTENKO, I.4, NORD, A.5, SCHWICHTENBERG, F.2, SHE, J.1, TUOMI, L.3. 1 Danish Meteorological Institute (DMI), Denmark - 2Bundesamt für Seeschifffahrt und Hydrographie (BSH), Germany - 3Finnish Meteorological Institute (FMI), Finland - 4Tallinn University of Technology (TalTech), Estonia - 5Swedish Meteorological and Hydrological Institute (SMHI), Sweden

substituted the HBM as the ocean circulation model. The implementation and setup of this new system, called NemoNordic 2.0, is described in more detail in Kärnä et al., 2021.

OVERVIEW The objective of the Copernicus Marine Service Baltic Monitoring and Forecasting Center (MFC) is to provide a state-of-the-art operational service with sea state, ocean physics and biogeochemical conditions of the Baltic Sea. This requires to operate up-to-date modelling systems with the best input and forcing data available. Starting Copernicus 1 contract in 2015, Baltic MFC products were based on different operational modelling systems provided by 5 institutes forming the Baltic MFC Consortium. Various ocean and biogeochemical model systems, as well as different forcing dataset were used. Our major achievement over the last six years, has been to successfully harmonize model systems and forcing dataset used in production. Also, model codes and forcing dataset had been updated. As a result, Baltic model products improved significantly to the benefits of Copernicus Marine Service users. During this six year period, the model development has been divided into two phases with respect to the ocean model code system. During the first part (2015-2017) the main development was focused on improving the HIROMB BOOS Model (HBM) ocean-ice system (Berg & Poulsen, 2012). Then, an Intercomparison study between the HBM and the NEMO (Nucleus for European Modelling of the Ocean; Madec et al., 2019) was performed. The outcome was a strategic decision in 2017: to phase out the HBM model system and move toward implementation of the more widely used community NEMO code as the Baltic MFC physical ocean model. During the second phase (2018-2021) all ocean model development focused on improving the implementation of the NEMO system for the Baltic Sea area. This was completed during year 2020, with the upgrade in the Baltic MFC product catalogue in 2020 where the NEMO model

The best available dataset for the Baltic Sea area is expected to be used to force the models. This includes information for river outflow and nutrient loadings, for which data from the latest updated E-HYPE hydrological model system (running at SMHI, Sweden) are employed. For the forecast product, atmospheric forcing data from the high resolution MetCoOp HARMONIE weather forecasts are used, with hourly frequency and 2.5 km resolution. For the multi-year products, data from the brand-new ERA5 global reanalysis (issued by Copernicus Climate Change Service at ECMWF) are used.

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 1.1 The Blue and White ocean Major developments made in the HBM system during the first years were on improving sea ice modelling, turbulence scheme and code modernization. For further details on the HBM achievements from 2015 to 2017 see description in Le Traon et al., (2017). The Baltic MFC consortium decided in 2017 to move toward an implementation of the NEMO system.. The NEMO system was introduced in the Baltic Service in the ocean reanalysis product. This setup was based on NEMO 3.6 and covered the Baltic Sea and North Sea area with a 2 nautical miles (~ 3.7 km) horizontal grid and 56 vertical levels. The domain has two open boundaries, one in the western English Channel and the other between Norway and Scotland. Efforts toward an operational forecast production started in 2018 with a new grid of 1 nautical mile (~ 1.85 km) horizontal resolution for the whole model area. During 2018 and 2019

130

#57

comprehensive tuning and validation was performed for this set up called Nemo-Nordic 1.0. New bathymetry data, including a new coastline was implemented; the boundary data was updated to use the Copernicus Marine Service NWS MFC near real time product. However, results showed that the sea level forecast in the transition water between the North Sea - Baltic Sea, the salinity inflow and sea ice forecast had lower quality than compared with the existing Copernicus Marine Service HBM based product, and an

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Intercomparison study was performed with the newly released NEMOv4.0 code version. Improved results were achieved by: - tuning the bathymetry data and bottom roughness, - changing the Galpering parameter in the turbulence model, - using more accurate numerics, - adding the Stokes drift from the wave model WAM.

Figure 1: Sea level validation statistics from the HBM (V201804) and the NEMO (V202012) systems based on observations from Baltic Sea coastal tide gauges for a two year validation period. The bias, centered root mean square difference, correlation and number of missing observations are shown (from left to right).

131

#57

Tests with different turbulent schemes and albedo values were performed to eliminate a cold surface temperature bias. Some numerical schemes (e.g., using higher order numerics) and parameterizations were optimized to improve inflow and sea ice. Thereafter, vertical resolution sensitivity on salt inflows was studied. This investigation resulted in keeping the 56 vertical layers for the time being, and an increase in resolution in the top layer. Validation results from the final Nemo-Nordic 2.0 setup are seen in Figure 1, 2 and 3 for sea level variations assessed by tide gauge observations, the Baltic total ice extent estimated by ice charts, and salinity observations from moorings, respectively. Comparison of results between the old HBM and new NEMO systems shows improvements in stratification and the distribution of salinity in favour of the Nemo-Nordic 2.0 setup. Additional validation results are available on the Copernicus Marine Service website. This new NemoNordic 2.0 based system (Kärnä et al., 2021) has provided the operational forecast product since December 2020, and will further be used for a new reanalysis production to be started during 2021, planned to be released end of 2021 to Copernicus Marine Service users.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The Baltic MFC wave analysis and forecast system is based on the WAM wave model (Komen et al., 1994). The first Baltic wave product, introduced into the CMEMS catalogue in 2017, used WAM cycle 4.5.4 and an upgrade to cycle 4.6.2 was done in 2019 to enable coupling with the physical forecast system. The model domain covers the Baltic Sea with 1 nautical mile horizontal resolution with an open boundary in the Skagerrak area. At the open boundary spectral-data from ECMWF’s deterministic wave forecast are used. During the ice season, the ice conditions are accounted for by excluding grid points in the calculations that have an ice concentration over 30%. Ice concentrations at each model grid point have been evaluated based on FMI’s ice charts (available in the Copernicus Marine Service catalogue) until December 2020. Subsequently, the wave production system was updated to use hourly ice concentration from the Baltic MFC NEMO based physical forecast. Using the ice concentration forecast enables accounting for changing ice conditions during the 5 days forecast (Tuomi et al., 2019). However, the accuracy of the ice forecast does not match the ice charts.

Figure 2: Computed total ice extent from the HBM (V201804) and the NEMO (V202012) systems based on ice chart observations for a 2 year period.

Northern Baltic Sea shorelines display an irregular structure and, in places, covered with islands and islets much smaller than the model grid size. To account for the effect this archipelago has on the attenuation of wave energy, a method to handle unresolved islands (e.g., Tolman, 2003) was implemented in the Baltic MFC WAM model code. This method reduces the wave energy

propagated from one grid cell to the next one, according to the shadowing effect caused by unresolved islands. The method improves the forecast quality and usability in coastal areas such as the Archipelago Sea, between the Baltic Proper and Gulf of Bothnia. For more information see description in Le Traon et al., (2017).

132

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Salinity from the HBM (left) and the NEMO (right) systems, plotted along with available observations from profile moorings in the Gotland Deep area.

In 2019, the wave production system was further improved with offline coupling to surface currents produced by the Baltic MFC physical forecast. Overall, coupling effects to the wave field were small. In certain situations though, the difference in the significant wave height with and without coupling to the currents was up to 60 cm (Kanarik et al., 2021). Also, in the Gulf of Finland, the refraction of swell, induced by currents, improved the peak period and the swell and wind sea energies.

1.2 The Green Ocean For simulations of the green ocean the ERGOM (Ecological Regional Ocean Model) model is used. ERGOM is a wellknown bio-geo-chemical model developed with focus on the Baltic Sea biochemical dynamics (Neumann, 2000). It describes the basic nitrogen and phosphorus cycle through 15 main state variables: - t hree different functional phytoplankton species, - two groups of zooplankton and detritus, - labile dissolved organic nitrogen, - t otal alkalinity (TA), - dissolved inorganic carbon (DIC), - ammonium, nitrate, phosphate, silicate and oxygen (see Figure 4).

Primary production, Chlorophyll and Secchi depth are calculated diagnostically. The sediment is not vertically resolved and consists of two nutrient state variables. Since 2015, several additions have been made into the Baltic MFC biogeochemical model system. For example, a new optical module has been added that also includes the labile dissolved organic nitrogen as a new state variable. The new module enables a more detailed calculation of turbidity and subsequent Secchi depth (Neumann et al., 2015). Furthermore, the net primary production was implemented as another diagnostic variable. Last but not least, a rudimentary iron circle has also been implemented. The most extensive addition to the Baltic MFC biogeochemical model system was made with the carbonate system in 2018. This was done following basically Kuznetsov and Neumann (2013) and Zeebe and WolfGladrow (2001). TA and DIC are two prognostic parameters used to calculate the biogeochemical processes influence on the carbonate system (see also Schwichtenberg et al., 2020). Respiration and primary production increase and decrease DIC. TA is mainly affected by river run-off and changes in nutrient concentrations. Both TA and DIC are required for the calculation of the diagnostic variables pH and pCO2. The carbonate system also includes exchange with annually increasing and seasonally variable atmospheric CO2 concentrations (see Figure 4).

133

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Overview of state variables and their interaction in ERGOM.

The coupling between ERGOM and NEMO is done via the NEMO TOP (Tracers in the Ocean Paradigm) component. The tracers transport, mixing and dilution are solved by the standard/native TRP (Passive Tracer Transport) modules, whereas users are free to define the bio-geo-chemical processes in the so-called «source-minus-sinks» module. This interface passes the column-wise concentration fields to the subroutines, which further evaluate the biogeo-chemical dynamics and update the concentrations. Additionally, this interface links ERGOM with the water temperature to estimate: - growth rate of temperature sensitive plankton species and oxygen solubility calculation, - c urrent speed for the sediment resuspension, - short wave radiation and salinity to calculate seawater optical parameters (Wan et al., 2013), - wind and temperature to estimate the gas-exchange through the surface. Validation with Chlorophyll and nutrient concentration observations showed that the new coupled NEMO-ERGOM system was able to reproduce observed dynamics with lower bias than the previous HBM-ERGOM system (Spruch et al., 2020). Preliminary results showed somewhat larger biases for oxygen and nutrients in the deeper layers. The implementation of a less diffusive 4th order flux corrected transport advection scheme in NEMO improved these inflow events with a positive effect on the ERGOM results.

1.3 Data assimilation The Parallel Data Assimilation Framework (PDAF) developed by the Alfred Wegener Institute (AWI) in Germany has been chosen for the data assimilation task (Nerger and Hiller, 2013). During the past six years, the PDAF system has been implemented and tested in close contact with AWI, for both the near real-time forecast production and the reanalysis products. The implementation and testing started with the PDAF-HBM interface using a Local Ensemble Square Root Transform Kalman Filter (LESTKF) to assimilate the SST L3 Copernicus Marine Service dataset covering the North Sea - Baltic Sea. It was shown that by assimilating SST, the system greatly reduced not only the SST bias and Root Mean Square differences but also had a positive impact on forecasting of the sea ice concentration and thickness. The PDAF-HBM system never ran in operational mode within the CMEMS as the decision toward the NEMO system was taken. Hence, since 2018, focus has been to develop a PDAFNEMO-ERGOM system based on the LESTKF filter for both the forecast and the reanalysis products. Since late 2020, the coupled system PDAF-NEMO-ERGOM produces the near real time forecast with a univariate SST assimilation scheme. Further developments have been performed and the next release of a new multi-year reanalysis product will include a multi-variate SST assimilation scheme, as well as a univariate scheme for assimilation of temperature and salinity profile observations. Assimilation of biogeochemical parameters such as oxygen and nutrient profiles is under development.

134

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. STATUS AT THE END OF COPERNICUS 1

3. POST 2021 PERSPECTIVES

By the end of the Copernicus 1 contract, the Baltic production system is based on: - t he ocean model code NEMOv4.0 (released 2019). The model configuration and set up has been heavily calibrated to the Baltic Sea (see the Nemo-Nordic v2.0 description in Kärnä et al., 2021), - t he Biogeochemical model code ERGOM, with both carbon and iron cycles included and online coupled with the NEMOv4.0, - t he wave sea state model code, WAMv4.6.2, offline coupled with NEMOv4.0.

The Baltic MFC consortium plans to continue the coupling activities, to enhance the number of processes and variables that are exchanged between models and to move toward a fully coupled system. Improving products in coastal areas will also be a field of study. These activities will be planned in interaction with the national agencies. In the Blue ocean system special focus will be on improving the important inflow events to the Baltic Sea with saline and oxygenated water. This will include implementation of new bathymetry data, increased vertical resolution in the computational grid, as well as introduction of 2-ways nesting grids in areas with narrow passage and complicated coastline. Focus will also be on surface layer dynamics by increasing the wave-current interaction.

For the data assimilation, the LESTKF scheme from the PDAF system is applied in both forecast and reanalysis production. The coupling between the Baltic model sub-system has evolved during the Copernicus 1 contract, from two separate standalone systems for the wave and the coupled 3D ocean-biogeochemical system, respectively, to a coupled and harmonized production system. This also included harmonizing the weather forcing dataset. The goal in 2021 is to have all six Baltic Sea model products in the Copernicus Marine Service data catalogue produced by this coupled system. The purpose is to ensure seamless products from a user point of view when investigating the Baltic Sea conditions for the whole product period from year 1993 and up to the latest forecast for the next 6 days.

Developments in the Green ocean will focus on the optical parameters, improve the forecast capability of algae blooms, hypoxia and eutrophication processes. Further development in data assimilation techniques would lower products biases and improve our quantitative understanding of the processes (see for instance Raudsepp et al., 2019 and Kõuts et al., 2021). Additionally new observation types and dataset will, when available, be tested and included in the data assimilation (e.g., future Sentinels and SWOT missions).

The Baltic MFC ocean forecast product is furthermore included in a multi-model-ensemble product approach together with other available forecast products for the Baltic Sea produced by the national institutes. See the dedicated website under the Baltic Operational Oceanographic System and Golbeck et al., 2015 for further details.

135

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGEMENTS The authors acknowledge all team members in the Baltic MFC Consortium from BSH, DMI, FMI, SMHI and TalTech for all achievements performed.

REFERENCES: Berg, P. and Poulsen, J. W.: Implementation details for HBM. Technical report, Danish Meteorological Institute. 2012. Golbeck, I., Li, X., Janssen, F., Brüning, T., Nielsen, J. W., Huess, V., Söderkvist, J., Büchmann, B., Siiriä, S.-M., Vähä-Piikkiö, O., Hackett, B., Kristensen, N. M., Engedahl, H., Blockley, E., Sellar, A., Lagemaa, P., Ozer, J., Legrand, S., Ljungemyr, P., Axell, L. Uncertainty estimation for operational ocean forecast products - a multi-model ensemble for the North Sea and the Baltic Sea. Ocean Dynamics (2015) 65:1603- 1631. DOI10.1007/s10236-015-0897-8 Kanarik, H., Tuomi, L., Björkqvist, J.-V., Kärnä, T., 2021. Improving Baltic Sea wave forecasts using modelled surface currents. Ocean Dynamics (forthcoming). Doi: 10.1007/s10236-021-01455-y. Komen GJ, Cavaleri L, Donelan M, HasselmannK, Hasselmann S, Janssen P (1994) Dynamicsand Modelling of Ocean Waves. Cambridge: Cam-bridge University Press, DOI https://doi.org/10.1017/ CBO9780511628955

Kõuts, M., Maljutenko, I., Liu, Y., Raudsepp, U., 2021. Nitrate, ammonium and phosphate pools in the Baltic Sea. In: Copernicus Marine Service Ocean State Report, Issue 4, Journal of Operational Oceanography, 12:sup1, accepted Kuznetsov, I. and Neumann, T., 2013. Simulation of carbon dynamics in the Baltic Sea with a 3D model. Journal of Marine Systems, 111, pp.167-174. Kärnä, T., Ljungemyr, P., Falahat, S., Ringgaard, I., Axell, L., Korabel, V., Murawski, J., Maljutenko, I., Lindenthal, A., Jandt-Scheelke, S., Verjovkina S., Lorkowski, I., Lagemaa, P., She, J., Tuomi, L., Nord A., Huess, V. 2021. Nemo-Nordic 2.0: Operational marine forecast model for the Baltic Sea. Geoscientific Model Development (submitted). Le Traon, P.Y. et al., 2017. The Copernicus Marine Environmental Monitoring Service: Main Scientific Achievements and Future Prospects. Special Issue Mercator Océan Journal #56. https://doi.org/10.25575/56. Madec G, Bourdallé-Badie R, Chanut J, et al. NEMO ocean engine. 2019. https://doi. org/10.5281/ZENODO.3878122.

Neumann,T. 2000. toward a 3D-ecosystem model of the Baltic Sea. J. Mar. Syst. 25, 405–419. https://doi.org/ https://doi.org/10.1016/S09247963(00)00030-0Neumann, Thomas, Herbert Siegel, and Monika Gerth. 2015. ‘A new radiation model for Baltic Sea ecosystem modelling’, Journal of Marine Systems, 152: 83-91. Nerger, L., Hiller, W. 2013. Software for Ensemble-based Data Assimilation Systems - Implementation Strategies and Scalability. Computers and Geosciences, 55, 110-118. doi:10.1016/j.cageo.2012.03.026 Raudsepp, U., Maljutenko, I., Kõuts, M., 2019. Cod reproductive volume potential in the Baltic Sea. In: Copernicus Marine Service Ocean State Report, Issue 3, Journal of Operational Oceanography, 12:sup1, s26–s30; DOI: 10.1080/ 1755876X.2019.1633075 Schwichtenberg, F., Pätsch, J., Böttcher, M.E., Thomas, H., Winde, V., Emeis, K.-C, 2020. ‘The impact of intertidal areas on the carbonate system of the southern North Sea’, Biogeosciences, 17: 4223-4245.

136

Spruch, L., Verjovkina, S., Jandt, S., Schwichtenberg, F., Huess, V., Lorkowski, I. and Lagemaa, P., 2020. Quality Information Document: Baltic Sea Biogeochemical Analysis and Forecasting Product. Link: https://resources.marine. copernicus.eu/documents/QUID/ CMEMS-BAL-QUID-003-007.pdf Tolman, H. 2003. Treatment of unresloved islands and ice in wind wave modesl. Ocean Modelling. 5. 2019-231. DOI: 10.1016/S1463-5003(02)00040-9. Tuomi L, Kanarik H, Björkqvist J-V, Marjamaa R, Vainio J, Hordoir R, Höglund A and Kahma KK. 2019. Impact of Ice Data Quality and Treatment on Wave Hindcast Statistics in Seasonally Ice-Covered Seas. Front. Earth Sci. 7:166. doi: 10.3389/feart.2019.00166 Wan, Z., Bi, H. and She, J., 2013. Comparison of two light attenuation parameterization focusing on timing of spring bloom and primary production in the Baltic Sea. Ecological modelling, 259, pp.40-49. Zeebe, Richard E, and Dieter Wolf-Gladrow. 2001. CO2 in seawater: equilibrium, kinetics, isotopes (Gulf Professional Publishing).

THE BLACK SEA MONITORING AND FORECASTING CENTER FOR THE COPERNICUS MARINE AND ITS EVOLUTION IN 2016-2021

CILIBERTI, S.1, GREGOIRE, M.2, STANEVA, J.3, PALAZOV, A.7, COPPINI, G.1, LECCI, R.1, BEHRENS, A.3, VANDENBULCKE, L.2, JANSEN, E.1, LIMA, L.4, AYDOGDU, A.4, MASINA, S.4, MATREATA, M.5, PENEVA, E.6, MARINOVA, V.7, Ocean Predictions and Applications Division, Centro Euro-Mediterraneo sui Cambiamenti Climatici, Italy - 2University of Liege, Liege, Belgium 3 Helmholtz-Zentrum Geesthacht Centre for Materials and Coastal Research, Geesthacht, Germany - 4Ocean Modeling and Data Assimilation Division, Centro Euro-Mediterraneo sui Cambiamenti Climatici, Italy - 5National Institute of Hydrology and Water Management, Bucarest, Romania 6 Sofia University “St. Kliment Ohridski”, Sofia, Bulgaria - 7Institute of Oceanology – Bulgarian Academy of Science, Varna, Bulgaria 1

INTRODUCTION The Black Sea Monitoring and Forecasting Centre (BSMFC) provides analysis, forecast and reanalysis for the blue and green ocean state in the Black Sea basin. Since 2016, BS-MFC is guaranteeing service standards, efficiency in operations and users support through its Local Service Desk and technical interfaces PU-DU. The BS-MFC catalogue, includes 6 main products with a number of 72 online datasets, operationally updated (Near Real Time ones are updated every day, while Multi Year ones are updated twice a year; from May 2021, they include interimmode production performed every month). Additionally, BS-MFC implements scientific evolution program for product catalogue upgrades and for the next generation of forecasting and reanalysis systems.

1. MAIN ACHIEVEMENTS FROM 2016 TO 2021 1.1 BS-MFC Operational High Level Architecture and Users BS-MFC was the last centre to enter in the Copernicus Marine Service framework in 2016: after a ramp-up phase of 6 months (Apr-Oct 2016), BS-MFC operational services were launched in Oct 2016, including BS-MFC Dissemination Unit

(DU) for products delivery, Production Units and Local Service Desk. In April 2018, BS-MFC technical interfaces were migrated to Central DU. A high level architecture as in Figure 1 has been implemented since the beginning of the service: it consists of three Production Units - Physics (BS-PHY) run by CMCC, Biogeochemistry (BS-BIO) run by ULiege and Waves (BS-WAV) run by HZG - which are responsible for the operational production of Near Real Time (NRT) and MultiYear (MY) products over the Black Sea basin (Figure 2). The PUs are directly connected to upstream data - observations and atmospheric forcing. A Local Service Desk (LSD, maintained by CMCC and IO-BAS) is continuously supporting users, through the Service Desk, and Production Units for the operational maintenance and monitoring of product availability. Evolution of technical interfaces, discussed with the Technical Working Group, is guaranteed by an Expert Technical Team, which cooperates with operational and LSD groups. PUs implement science-driven tasks for service evolution, in collaboration with USOF, that leads product quality activities, and NIHWM, that is responsible for the Danube River hydrological measurements and forecast. Each PU implements a Backup Unit, for production recovery in case of IT failure, and an Archiving Unit for superseded products, made available to users upon request. Production pipelines offer daily analyses, simulation and 10 days forecast, while, twice per year, they deliver updated timeseries of reanalysis. The operational products are validated using available satellite and in-situ data in a continuous and consistent manner, adopting standards inherited from GODAE/OceanPredict and MERSEA/MyOcean.

137

#57

It is worth mentioning that the good relation with the Black Sea INS TAC and Marine Research Infrastructure in Bulgaria led to an observing system upgrade: the installation of

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

mooring platforms and regular deployment of profiling floats to use in the future for assimilation and validation activities.

Central Information System

Science-Driven Evolution

CMCC, ULIEGE, HZG, USOF, NHWM

CMCC

ULIEGE

HZG

Physics Backup PU

SEA LEVEL TAC

Biogeochemistry Backup PU

SEA SURFACE TEMP. TAC

OCEAN COLOUR TAC

Waves Backup PU

IN SITU OCEAN TAC

Figure 1: BS-MFC high level architecture in the configuration operating since 2018.

Figure 2: Black Sea spatial domain and bathymetry (expressed in meters).

138

IN SITU WAVE TAC

IO-BAS& CMCC

Waves PU

BS Technical Group

Biogeochemistry PU

Physics PU

BS-MFC Service Desk

CMENS

#57

At the end of 2020, a number of 60 users have downloaded BS-MFC products with around 3 TB of downloaded data and about 300k downloads from the catalogue. The area of benefits includes business and private companies (45%), academia (20%) and public sector (32%) according to 2019 internal reporting. Applications that use BS-MFC products are mainly dedicated to Commercial sector and in particular the Blue Economy. Such encouraging statistics are also supported by a very robust service, that ensures a minimum number of incidents or failures thanks to a very strict service monitoring and support to Production Units. Timeliness of BS-MFC products is around 99.9%, with some expected degradation - but always above 95% - at the occurrence of a new release. In 2020, the number of incidents has dramatically decreased (from about 40 in 2019 up to 15) thanks to improved operational chains and reliable interfaces PU-DU.

1.2 BS-MFC Physics The nominal product for BS-PHY NRT [1] provides analysis and 10-days forecast every day, as hourly and daily means with nominal start of the forecast at 00:00UTC for the following list of variables: - 3D temperature, - salinity and currents, - D sea surface height, - mixed layer depth and bottom temperature. The BS-PHY NRT system (EAS3 version) is based on NEMO v3.4 hydrodynamical model, implemented on a spatial grid of 3 km resolution and 31 z-levels, online coupled to OceanVar [2.3] for the assimilation of near real time observations (in situ temperature and salinity profiles, sea level anomaly along track and sea surface temperature, provided by corresponding CMEMS TACs). It is forced by ECMWF IFS analysis and forecast atmospheric fields at 12.5 km horizontal resolution and 3-6 h temporal frequency. It implements a closed boundary condition at the Bosporus Strait. An operational dashboard for monitoring analysis accuracy is provided by a regional website developed by CMCC that publishes daily bulletin for BS-PHY NRT analysis and forecast along root mean square model-observations

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

misfits at weekly frequency for temperature, salinity and sea level anomaly. Considering the 2019 operational production, the BS-PHY NRT system is characterized by an error in temperature of about 1.8°C and of 0.25 PSU in salinity in the layer 10-20 m. Below 20 m, the temperature error spans from 1.6°C to less than 0.2°C along the water column, while salinity error is between 0.4 PSU in the halocline and less than 0.2 PSU in the deepest levels. The nominal product for BS-PHY M [4] provides reanalysis for the past reconstruction of the physical ocean state in the Black Sea starting from 1993. Ocean variables included in the BS-MFC MY are identical to NRT, but as monthly and daily means. The BS-PHY MY system (E3R1 version) is based on a new implementation of NEMO v3.6 hydrodynamic model, at the same NRT system resolution but with more accurate bathymetry. It is also online coupled to OceanVar for the assimilation of reprocessed observations (insitu temperature and salinity profiles from merged datasets: INS TAC and historical SeaDataNet observations, sea level anomaly along track provided by SL TAC). It is forced by ECMWF ERA5 reanalysis atmospheric fields at 30 km horizontal resolution and 1 h temporal frequency. To improve the representation of warmer and saltier waters coming from the Mediterranean Sea into the Black Sea, a damping to temperature and salinity profiles as provided in [5] at the Bosporus exit has been implemented. Sea surface temperature restoring with magnitude of retroaction of -200 W/m2/K has been implemented using gridded L4 SST satellite data from CMEMS SST TAC. In terms of accuracy, the new hydrodynamical model and data assimilation developments in E3R1 guarantees a reduction of the error in temperature (salinity) of about 50% (60%) with respect to the previous reanalysis system - version E2R2, operational from 2017 to 2019 - (e.g., ~0.6°C and ~0.4 PSU in 0-10 m; a maximum error in temperature up to ~0.7°C in average is provided in the 10-100m layer, while salinity error is extremely low - up to ~0.05-0.1PSU below 100 m). Considering the accuracy in sea level anomaly, the new reanalysis exhibits a significant improvement with timeaveraged root mean square deviation of 2.3 cm (versus 3.7 cm in the previous version) (Figure 3).

139

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Vertical profiles of root mean square difference and bias for a) temperature and b) salinity, by comparing the reanalysis E3R1 and the old one E2R2 against Argo data in the period Jan 1993 - Dec 2018.

1.3 BS-MFC Biogeochemistry The nominal product for BS-BIO NRT [6] provides analysis and 10-days forecast every day, as daily means, with nominal start of the forecast at 00:00UTC for the following list of variables: - Chlorophyll and phytoplankton, - dissolved oxygen, nitrate and phosphate, - surface pressure of carbon dioxide, - pH, - surface downward mass flux of carbon dioxide, - net primary production, - sea water alkalinity and concentration of dissolved inorganic carbon in seawater. The first version of the operational system - operational in the period 2016-2018 - was based on online coupled version GHER3D hydrodynamical model and the BiogeochemicAl Model for Hypoxic and Benthic Influenced areas (BAMHBI) [7,8,9] on a spatial domain of 5 km resolution and over 40 -levels. During the Phase 2, it has evolved toward a new system, based on NEMO v3.6 online coupled to BAMHBI, aligned with BS-PHY NRT system (e.g., same

grid, atmospheric forcing). Since Jul 2019, the BS-BIO NRT system solves and delivers variables describing the carbonate system (i.e., pCO2, pH, DIC, CO2 flux, alkalinity). Since Jun 2020, it is assimilating Chlorophyll satellite L3 data from CMEMS OC TAC (CHL L3 NRT product based on multi-satellite composites) via the Ocean Assimilation Kit OAK [10], improving the surface Chlorophyll product quality. Figure 4 compares the simulated and observed surface Chlorophyll a in typical Black Sea regions located in the shelf and deep sea and displayed in Figure 5. Table 1 gives the model bias for different regions. The comparison of observed and simulated Chlorophyll highlights that, in regions under the direct influence of river discharges (i.e., regions 4 and 5 under the influence of the Danube, Dnestr and Dnieper rivers), differences between model and observations persist even after data assimilation. The bias is positive, and up to 0.37 and 0.26 mg.m-3 respectively in regions 4 and 5. In these coastal regions, the dynamic of blooms, biogeochemical cycling and foodweb is expected to be under the dominant influence of rivers’ discharges. The lower quality of the Chlorophyll product compared to more offshore areas is explained by the use of monthly climatological data of inorganic and

140

#57

organic materials (e.g., nitrate, phosphate, silicate) to force the model. To remedy this situation we will test the use of NRT river discharges delivered by NIHWM. In Region 7 (transition between the North West Shelf (NWS) and the open sea), the model and observations are very close.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

state in the Black Sea. Since Jul 2020, a new timeseries of 27 years (from 1992 to 2019) with the upgraded NEMOBAHMBI modelling framework has been released to users. The system is forced by ECWMF ERA5 reanalysis atmospheric fields.

The nominal product for BS-BIO MY [11] provides simulation for the past reconstruction of the biogeochemical ocean

Figure 4: timeseries of satellite (in blue) and model (in red) Chlorophyll spatial averages computed for (top row) open sea sub-regions 1, 2 and 3, and (bottom row) NWS regions 5, 6 and 7. Sub-regions limits are shown in Figure 5.

Figure 5: Sub-Regions (based on [12]) used for computing error statistics for Chlorophyll. Regions 1-3 are the “open sea”, regions 4-5 is the North-Western Shelf and region 7 is located at the shelf break and is considered as a transition region between eutrophic and oligotrophic conditions.

141

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Region

EAN

0.012

-0.012

-0.011

0.19

0.15

0.014

Region

EAN

0.04

0.009

0.055

0.07

-0.07

Table 1: Bias log10 statistic for Chlorophyll obtained for BS-BIO NRT when considering the observation-model prediction pairs, for the different regions (1 to 11).

1.4 BS-MFC Waves Black Sea wave products were included in catalogue at the very beginning of the Copernicus Marine Service. Despite this short experience, the BS-MFC managed to provide state-of-the-art wave modelling solutions (analysis, forecast and reanalysis) as other MFCs since 2016. The nominal product for BS-WAV NRT [13] provides analysis and 10-days forecast every day, as hourly instantaneous, with nominal start of the forecast at 12:00UTC for the most relevant wave variables, such as: - s ignificant wave height, wave mean period and wave direction, - s tokes drifts, -w ind wave significant height, wind wave mean period and direction, -p rimary and secondary swell.

The model is solved on the same grid and uses the same atmospheric forcing as the BS-PHY NRT. Since December 2020, the modelling framework has been newly upgraded to the state-of-the art WAM Cycle 6.0, which includes new dissipation terms parameterisations of [14,15]. The model efficiency has been significantly improved through: - implementations of the new 2D decomposition of the model area for the MPI, - shallow water wave representations, - providing new parameters wave height and crest height [16]. The BS-WAV NRT system is now offline coupled with BSPHY NRT using hourly means surface currents and water depths. Figure 6 shows examples for validation of the BSWAV NRT and MY products against different satellite data.

142

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 6: Top: (left) timeseries of maximum significant wave heights derived from combined Jason satellite measurements and BS-WAM MYP; (right) Scatter plots showing Sentinel-3a satellite measurements versus BS-WAV NRT significant wave heights for Q4 2018. Bottom: Distribution of significant wave height of WAM with overlaid Jason-2 satellite track on 03 December 2016 06:00 UTC (left) and along-track measured and computed significant wave height (right).

The nominal product for BS-WAV MY [17] provides a reanalysis for the past reconstruction of the wave climate in the Black Sea starting from 1979 as hourly-instantaneous. The modelling framework is similar to NRT: it accounts for wave breaking and assimilation of significant wave height and wind speed as provided by AVISO satellite

2002-2013 (Jason-1)

products (ftp-access.aviso.altimetry.fr) using an Optimal Interpolation (OI) scheme. The system is forced by ECMWF ERA5 reanalysis atmospheric forcing at hourly frequency. Table 2 shows EAN for significant wave height over the period 1979-2019

2008-2016 (Jason-2)

2016-2019 (Jason-3)

2002-2019 (combined)

BIAS

RMSD

BIAS

RMSD

BIAS

RMSD

BIAS

RMSD

3.25

17.00

-0.62

23.6

-7.6

26.3

0.475

20.9

Table 2: Estimated accuracy numbers corresponding to the comparison between hourly model and measured significant wave height, 20022019. All values are centimetres.

143

#57

2. STATUS AT 2021: BS-MFC OFFER TOWARD COPERNICUS 2

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

3. POST 2021 PERSPECTIVES

BS-MFC service evolution programme provides state-ofthe-art modelling infrastructure for improving forecasting capabilities and climate studies in the Black Sea basin: being interconnected with an operational framework, it guarantees also users’ engagement and stakeholders’ requirements. As a result, the last phase of Copernicus 1 is dedicated to the provision of additional upgraded products, completing the offer to support Blue Growth, coastal communities and society. BS-PHY is going to contribute with a new operational analysis and forecasting system EAS4 system version - characterized by: - higher spatial resolution spatial grid (2.5 km and 121 vertical levels), - optimal interface between Mediterranean and Black Sea through the Marmara Sea: High resolution temperature, salinity, sea surface height and zonal and meridional velocity from the Unstructured Turkish Strait System (U-TSS, [18]) are provided to the BS-PHY NRT by edging the model in the Marmara Sea. This improves the representation of the Mediterranean Sea water inflow; - The Danube River interannual dataset from historical data provided by NIHWM to improve land forcing representation; - A novel implementation of the OceanVar, which uses a spatially varying set of 45 EOF to describe the covariance of sea surface height and temperature and salinity in the water column. The new system significantly improves both temperature and salinity along the water column with respect to EAS3. For BS-BIO, new optical products are delivered for the next entry into service. Since May 2021, BS-BIO contributes, with additional datasets, to complete the biogeochemical catalogue with transparency and turbidity - associated to the amount of photosynthetically active radiation and light attenuation coefficients. Together with PHY and BIO, BSWAV is going to start the operational update of the multiyear products in interim mode, with provision of monthly datasets, to support climate community and policy makers in the Black Sea region, enriching the product catalogue.

Continuous development of BS modelling framework for the next generation of NRT and MY systems accounts for a number of scientific challenges the BS-MFC is foreseeing to achieve in the next years. CMEMS has supported many of such scientific R&D priorities thanks to implementation of dedicated Service Evolution projects, that will be exploited by BS-MFC to improve product quality (Table 3). The plan includes a focus on coupling strategy for the representation of small scale processes in the Black Sea region such as: - a fully coupled physics and waves systems to improve essential variables forecasting, - an enhanced connection (several strategies of coupling online/offline, join data assimilation will be tested) between biogeochemistry with physics and waves will help in representing Suspended Particulate Matter dynamics and light penetration schemes, - an implementation of a coupled high resolution wave-atmosphere system for improving the air sea interaction representation, - an ensemble forecasting (using perturbation of atmospheric forcing, model parameters and initial conditions) to improve NRT skills and to support coastal applications will be developed for the 3 PUs. Data assimilation scheme will evolve toward high frequency correction and the ingestion of new data such the future Sentinel-6, CFOSAT, SWOT and SL L3 at 5 Hz, ocean colour spectral products that require dedicated Observing System Experiments (OSE) to assess their impact in the overall BSMFC NRT quality.

144

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

CMEMS SE project

BS-MFC uptake

SOSSTA

Using the dynamical-statistical observation operator for satellite SST observations that account for skin SST layer diurnal variability.

SCRUM2

To create the new ensemble members (exploiting the stochastic code added to NEMO) and afterwards to evaluate them by using ensemble consistency analysis tools.

BIOTIMOD, OPTIMA

Implementation of radiative transfer model in the BS-BIO model.

LAMBDA

Evaluation of river runoff data to improve BS-PHY systems.

LATEMAR

Proposal of new wave parameters such as expected value of maximum crest height and expected value of maximum crest through wave height (both in spatial and temporal domain) for BS-WAV system.

WAVEFLOW

Development of a new WAM code using validation methods based on comparison with spectral data, which had already been tested successfully for the Global Ocean, North Sea and the Baltic Sea.

CEASELESS

New developments of wave model, wave data assimilation and coupling between waves and ocean models. New developments of NEMO ocean model, in particular a) for dealing with the usage of the high resolution satellite dataset for ocean forecasting and validation, b) for adopting the new improved numerical NEMO kernel (for air-sea interaction, wave-currents coupling and biogeochemical interfaces), c) to exploit opportunities of new high performance computing technology for high resolution BS-MFC products.

IMMERSE

Using last stable version of the NEMO ocean model to: - test air-sea interaction parameterization, wave-currents coupling and biogeochemical interfaces, - exploit the high performance computing optimizations for the next generation of high resolution BS-MFC products, - test interfaces with high resolution observations for validating NEMO configuration for the Black Sea physical model.

FORCOAST

To develop specialized products for wild fishery and mariculture sectors in the Black Sea

Table 3: List of uptakes from CMEMS Service Evolution and H2020 projects by BS-MFC to improve the NRT and MY systems.

ACKNOWLEDGEMENT The Authors thank the BS-MFC Group composed by D. Azevedo, S. Creti’, L. Stefanizzi, S. Causio from the BS-PHY PU at CMCC, A. Capet from the BS-BIO PU at ULiege, M. Ricker, H. Günther and G. Gayer from the BS-WAV PU at HZG for their scientific developments and technical support; F. Palermo (CMCC) as Technical Expert of the BS Technical Group; N. Pinardi, M. Gunduz and M. Ilicak as Scientific Experts contributing to BS-PHY developments; N. Valcheva (IO-BAS) and P. Agostini (CMCC) for their support in the management.

145

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

REFERENCES: 1. Ciliberti, S. A., Peneva, E. L., Jansen, E., Martins, D., Cretí, S., Stefanizzi, L., Lecci, R., Palermo, F., Daryabor, F., Lima, L., Coppini, G., Masina, S., Pinardi, N., & Palazov, A. (2020). Black Sea Analysis and Forecast (CMEMS BSCurrents, EAS3 system) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https:// doi.org/10.25423/CMCC/ BLKSEA_ANALYSIS_FORECAST_ PHYS_007_001_EAS3. 2. Dobricic, S. and Pinardi, N. (2008). An oceanographic threedimensional variational data assimilation scheme. Ocean modelling, 22(3-4), 89-105, 2008. 3. Storto, A., Dobricic, S., Masina, S. and Di Pietro, P. (2011). Assimilating along-track altimetric observa-tions through local hydrostatic adjustment in a global ocean variational assimilation system. Monthly Weather Review, 139(3), 738754, 2011. 4. Lima, L., Aydogdu, A., Escudier, R., Masina, S., Ciliberti, S. A., Azevedo, D., Peneva, E. L., Causio, S., Cipollone, A., Clementi, E., Cretí, S., Stefanizzi, L., Lecci, R., Palermo, F., Coppini, G., Pinardi, N., & Palazov, A. (2020). Black Sea Physical Reanalysis (CMEMS BS-Currents) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https://doi. org/10.25423/CMCC/BLKSEA_ MULTIYEAR_PHY_007_004.

5. Aydoğdu, A., Pinardi, N., Özsoy, E., Danabasoglu, G., Gürses, Ö., and Karspeck, A. (2018). Circulation of the Turkish Straits System under interannual atmospheric forcing, Ocean Sci., 14, 999-1019, doi:10.5194/os-14-999-2018, 2018. 6. Grégoire, M., Vandenbulcke, L., & Capet, A. (2020). Black Sea Biogeochemical Analysis and Forecast (CMEMS Near-Real Time BLACKSEA Biogeochemistry) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https:// doi.org/10.25423/CMCC/ BLKSEA_ANALYSIS_FORECAST_ BIO_007_010_BAMHBI.

11. Grégoire, M., Vandenbulcke, L., & Capet, A. (2020). Black Sea Biogeochemical Reanalysis (CMEMS BSBiogeochemistry) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https://doi. org/10.25423/CMCC/BLKSEA_ REANALYSIS_BIO_007_005_ BAMHBI 12. Kopelevich, O. V.; Sheberstov, S.V.; Sahling, I.V.; Vazyulya, S.V.; Burenkov V.I. (2013). Bio-optical characteristics of the Russian Seas from satellite ocean color data of 1998-2012. Proceedings of the VII International Conference “Current problems in Optics of Natural Waters (ONW 2013)”, St.-Petersburg (Russia), September 10-14, 2013.

7. Grégoire, M.; Raick, C.; Soetaert, K. Numerical modeling of the deep Black Sea ecosystem functioning during the late 80’s (eutrophication phase). Progress in Oceanography 2008, 76(9), 286-333. 8. Grégoire, M.; Soetaert, K. Carbon, nitrogen, oxygen and sulfide budgets in the Black Sea: A biogeochemical model of the whole water column coupling the oxic and anoxic parts. Ecological Modelling 2010, 15. 9. Capet, A.; Meysman, F.J.R.; Akoumianaki, I.; Soetaert, K.; Grégoire, M. Integrating sediment biogeochemistry into 3D oceanic models: A study of benthic-pelagic coupling in the Black Sea. Ocean Modelling 2016, 101, 83-100.

10. Vandenbulcke, L.; Barth, A. A stochastic operational forecasting system of the Black Sea: Technique and validation. Ocean Modelling 2015, 93, 7-21.

13. Staneva, J., Behrens, A., Ricker, M., & Gayer, G. (2020). Black Sea Waves Analysis and Forecast (CMEMS BSWaves) (Version 2) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https:// doi.org/10.25423/CMCC/ BLKSEA_ANALYSISFORECAST_ WAV_007_003. 14. Ardhuin, F., Rogers, E., Babanin, A., Filipot, J.-F., Magne, R., Roland, A., Van Der Westhuysen, A., Queffeulou, P., Lefevre, J.-M., Aouf, L., Collard, F., 2010. Semiempirical dissipation source functions for ocean waves: Part I, definition, calibration and validation. J. Phys. Oceanogr. 40, 1917–1941.

146

15. ECMWF 2020, IFS Documentation CY47R1, Book Chapter, ECMWF, https://www.ecmwf.int/. 16. Benetazzo, A., Barbariol, F., Pezzutto, P., Staneva, J., Behrens, A., Davison, S., Bergamasco, F., Sclavo, M., Cavaleri, L. (2021). toward a unified framework for extreme sea waves from spectral models: Rationale and applications. Ocean Engineering, 219, 108263. 17. Staneva, J., Behrens, A., Ricker, M., & Gayer, G. (2020). Black Sea Waves Reanalysis (CMEMS BS-Waves) (Version 2) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https://doi. org/10.25423/CMCC/BLKSEA_ MULTIYEAR_WAV_007_006. 18. Ilicak, M., Federico, I., Barletta, I., Pinardi, N., Ciliberti, S.A., Clementi, E., Coppini, G., Lecci, R., and Mutlu, S. (2021). Evaluation of the new high resolution unstructured grid Marmara Sea model. EGU General Assembly 2021.

THE IBERIA-BISCAY-IRELAND MONITORING AND FORECASTING CENTER

SOTILLO, M.G1,6, AZNAR, R.2, GUTKNECHT, E.1, LEVIER, B.1, REFFRAY, G.1, LORENTE, P.2, BARRERA, E.4, DABROWSKI, T.5, AOUF, L.3 Mercator Ocean International - 2Nologin - 3Meteo-France - 4AEMET - 5Marine Institute - 6PdE

phase (from 2015 to 2021), as well as a short overview of the IBI-MFC service evolution roadmap (to be developed within the next Copernicus 2 phase).

1. THE IBI-MFC MISSION The Copernicus Marine Service IBI-MFC (Iberia-Biscay-Ireland Monitoring & Forecasting Centre) offers a comprehensive portfolio of regular and systematic regional information on the state of the ocean for the European Atlantic façade, supporting all kind of marine applications. Its mission is to provide operational regional short-term ocean forecasts and reanalysis products for the so called IBI area, covering the Blue (ocean physics and waves) & Green (biogeochemical) ocean components. To this aim, different high-resolution model applications are being run (using the NEMO, WAM and PISCES model codes for the IBI ocean circulation [PHY], wave state [WAV] and biogeochemical [BIO] model runs, respectively). IBI-MFC model products integrate observations through Data Assimilation (DA) schemes used in the generation of IBI analysis & reanalysis products. IBI-MFC near-real-time (NRT) forecasts (with horizontal resolution ranging from 2.5 to 5 km resolution) are run on daily to weekly basis, delivering forecast products over 10 days ahead. On the other hand, the IBI Multi-year (MY) reanalysis, for circulation and waves, and non-assimilative hindcast (biogeochemical) products cover from 1993 till present time. IBI-MFC products and services are science-based and reliable, and their quality is assessed (both in near-real-time and delayed mode). The IBI-MFC service is engaged in a continuous improvement and its R&D roadmap, fully aligned with the general Copernicus Marine Service scientific objectives, is dedicated to meet identified end-users needs. To address these objectives, the IBI-MFC is managed by a consortium of centres, coordinated by Mercator Ocean International, and including NoLogin, Météo-France, the Galician Supercomputing Centre (CESGA), the Spanish Met Office (AEMET), and the Irish Marine Institute (IMI). This paper reviews the present status of the Copernicus Marine Service IBI-MFC service, highlighting the main evolutions and achievements occurred along Copernicus 1

2. IBI-MFC ACHIEVEMENTS ALONG COPERNICUS 1 The IBI-MFC has extended its product offer along this first Copernicus phase (2015-2021), and it delivers today NRT short-term forecasts and MY products for physical ocean, wave state and biogeochemical parameters. To achieve this significant IBI product portfolio enhancement along Copernicus 1, three new production lines were developed to operationally generate new IBI BIO and WAV NRT forecasts, together with a new IBI-WAV-MY reanalysis product. Today’s IBI-MFC service, based on these 3 new operational suites in addition to 3 other already existing in 2015 (IBI-PHY-NRT forecasts, IBI MY reanalysis and non-assimilative hindcast for PHY and BIO; Sotillo et al., 2015), reached completeness and it is the baseline for the next Copernicus 2 service phase (to be started in 2021). Generally, the IBI-MFC is continuously improving model applications used as base for the IBI operational production lines to: - increase IBI product resolution, - enhance Data Assimilation, using any new or improved observational product available, - achieve higher system interactions through model coupling. For this purpose, a roadmap was planned at the start of Copernicus 1 and has been strictly followed throughout the project. A key success factor was the the essential and continuous work of the R&D teams (in ocean physics, waves and biogeochemistry)maintaining models at the state of the art, to improve them as soon as possible and to adapt them specifically A summary of main specific achievements and key service evolution milestones achieved for each of these six IBI-MFC production lines is provided below.

147

#57

1. The IBI-MFC ocean physical analysis/forecast/ reanalysis component: The IBI NRT PHY forecast system (Sotillo et al., 2020) is based on a 1/36º ( 2.5 km) NEMO model application continuously upgraded, since its first implementation during MyOcean era (Maraldi et al., 2013). The IBI model set-up was coded (at the start of Copernicus 1) in NEMO3.4 and it has then been upgraded to NEMO3.6. A more recent NEMO4.x release code will be implemented (within the H2020 IMMERSE Project [2021], where a future IBI product demonstrator at 1/108º resolution is under development). The atmospheric forcing has been progressively upgraded from ECMWF 1/8º 3-hourly data to ECMWF IFS 1/12º hourly data. Tidal forcing was also updated, including the loading tide effect. Freshwater contribution from land is considered in the model set-up, including river discharges from 33 main rivers and an extra coastal runoff contribution. The IBI-MFC is working on enhancing this river freshwater forcing; evaluating impacts of new river databases (such as the one generated in the framework of the Copernicus Marine Service Evolution [SE] Project Lambda) and testing the model sensitivity with the IBI model operational set up (Sotillo et al, 2021). In the last Copernicus Marine Service release (Dec 2020), an offline wave-current coupling (including effects of Stokes drift & surface stress modification by waves) was first implemented in the IBI daily forecast cycle runs. The IBI ocean circulation forecast system downscales the Copernicus Marine Service Global solution (today, the IBI-PHY-NRT system is nested into the GLO 1/12° daily updated physical solution). Further details of downscaling effectiveness performed from the Global solution through the IBI regional application are available in Lorente et al., (2019). The performance of different models, in terms of their eddy properties and three-dimensional composite structures was evaluated over the western Mediterranean Sea. See the Copernicus Marine Service SE Project MEDSUB (2019). It focused on assessing the mesoscale and sub-mesoscale information inclusion in different Copernicus Marine Servic solutions. Mason et al., (2019) pointed out how models that include data assimilation (GLO and MED, not IBI at that time) approximate more precisely eddy property distributions observed with contemporaneous altimetry

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

observations in the study area. One of the main IBI-MFC objectives for Copernicus 1 was to include a DA scheme in its physical component. An application of the MOI SAM2, relying on a reduced order Kalman filter and based on SEEK formulation (see Amo et al., 2020) was implemented. Since April 2018, the IBI-MFC delivers regional physical analysis. Before this data assimilation release, the IBI solution was spectrally nudged to the Copernicus Marine Service global system as a temporary solution to substitute the periodic sequential forecast re-initialization. Figure 1 illustrates the IBI performance improvement (gained in terms of SST simulation) achieved along this path to include data assimilation in the IBI-PHY-NRT system. With respect to the IBI PHY MY products, the first 1/12º reanalysis was run in 2015, covering the 2002-2011 period. This first reanalysis was forced with ECMWF ERA-interim and assimilated along-track SLA, maps of SST and in-situ profiles (mostly from ARGO floats). In the last 6 years, two major upgrades have been performed in this reanalysis system, resulting in two new IBI long runs (and the related complete IBI-PHY-MY product update in the Copernicus Marine Service catalogue in 2018 and in December 2020). Through these 2 IBI reanalyses releases, both the model and the data assimilation scheme were upgraded along with assimilated observations (i.e., including new altimetric and SST observational products). Likewise, the atmospheric forcing was updated, using the new ECMWF ERA5 in the last IBI reanalysis run. All these resulted in a continuous enhancement of IBI MY product quality (further details on the Copernicus Marine Service Quality information document; Levier et al., 2021). Note that the IBI reanalysis temporal coverage was extended back to 1993, to cover the whole altimetric era. Finally, to point out that some efforts were done to evaluate differences between the IBI NRT and MY products delivered through the catalogue. Thus, periodic intercomparison with IBI physical ocean products along the overlapping time periods have been performed (see as example the analysis provided by Aznar et al., 2017). The IBI-MFC aims to enhance its product consistency, in particular between the NRT and MY products, and to produce IBI reanalysis on a same model grid than IBI NRT forecast services.

148

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: SST differences between IBI-PHY-NRT solutions and satellite observations (CMEMS L3 satellite product 010_009a) for Summer (JAS) 2014. IBI data from (a) the 2015 system version (with periodic sequential forecast re-initialization from the CMEMS Global solution); (b) the 2016 system version (with spectral nudging to the Copernicus Marine Service Global solution) and (c) the 2018 system version (already with data assimilation scheme).

2. The IBI-MFC biogeochemical analysis/forecast/ hindcast component: A physical–biogeochemical coupled model system, based on a NEMO-PISCES 1∕36° model application, was developed to forecast ocean dynamics and marine ecosystems in the IBI area. The first IBI BIO NRT forecast system was set up in 2017, providing weekly forecast products. Then, different specific upgrades were applied in this IBI BIO forecast systems such as: -N EMO-PISCES model upgrade, - i mprovement of carbon cycle with surface spCO2 boundary reflecting anthropogenic inputs, - inclusion of additional nutrient inputs for rivers (EEA data), - s etup of a new formulation for permanent burial to the sediment, - i mprovement of initial and boundary conditions. The catalogue product list has also been expanded, including now the main biogeochemical variables (Chlorophyll, oxygen, iron, nitrate, ammonium, phosphate, silicate, net primary production, euphotic zone depth) and variables related to the carbon cycle (surface partial pressure of carbon dioxide, dissolved inorganic carbon and

pH). These upgrades have resulted in an enhancement of the IBI NRT BIO product quality (Figure 2). Gutknecht et al., (2019) presents more consistency and skill assessment information of the IBI PISCES biogeochemical model (comparing data from a 7-year qualification simulation to available satellite estimates as well as in situ observations from ICES, EMODnet and BGC-Argo). The IBI NRT BIO forecast system is now capable of reproducing the main biogeochemical and ecosystem features of the IBI area, successfully capturing the spatial distribution and seasonal cycles of oxygen, nutrients, Chlorophyll, and net primary production. The product provides a greater understanding of today’s state (and changes) in the marine biogeochemistry of European waters. Since the beginning of Copernicus 1, the IBI-MFC has delivered a BIO MY product derived from a regional high-resolution 1/12° non-assimilative biogeochemical hindcast. This IBI BIO hindcast model application is online coupled to the IBI MY physical reanalysis, and nested into the Global BIO MY product. In the last 6 years, this IBI BIO MY system was upgraded to be consistent with the IBI PHY MY system, extending their temporal coverage from 1993 till present time. The NEMO-PISCES model code was updated in parallel to the upgrades performed in the NRT BIO

149

#57

forecast model system (previously described). The present NEMO-PISCES 3.6 code includes a new formulation for permanent burial to the sediment). Nutrient inputs were also updated, using now the Global NEWS 2 (Mayorga et al., 2010) and additional nutrient at rivers (inputs from EEA data sources). The carbon cycle modelling was improved by including the anthropogenic forcing in the surface CO2

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

pressure. Also, suplementary variables related to carbon cycle (such as spCO2, pH and DIC) are now provided. Additionally, the IBI BIO MY solution has been enhanced, with all variables showing a general improvement in open ocean, but with very strong impacts on the simulation accuracy of coastal/shelf nutrients and chlorophyl. For more details, please refer to McGovern et al. (2020).

Figure 2: Improvements on IBI BIO Products. (a) and (b) Continental shelf nutrients (PO4) in NRT products (2018 and 2020 system releases) compared to ICES observational data. (c) Clorophyll-a RMSD between MY products (2015, 2018 and 2020 system releases) and satellite ocean colour product (Copernicus Marine Service products 009_090 and 009_091).

150

#57

3. The IBI-MFC wave forecast/analysis/reanalysis component: The IBI-MFC provides a short term (10-days) highresolution wave forecast product for the IBI area. The IBI wave system, based on the MF-WAM model, is today run twice-a-day on a 1/20º grid, forced with the ECMWF hourly wind data. The system assimilates significant wave height (SWH) altimeter data, and it is off-line coupled with surface currents provided by the IBI NRT ocean forecasts system. The IBI WAV NRT forecast product, in operations since December 2016, has been continuously upgraded with: -e volution of the model application (MFWAM model upgraded in March 2019), -e nhancement of horizontal resolution (from the original 1/10º to 1/20º, changed in December 2019) and extension of forecast horizon (up to 10 days ahead since December 2020), - i mprovements of bathymetry and atmospheric forcing (since December 2019 using hourly winds, instead of the original 3-hourly ones).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The first IBI WAV analysis product was delivered in July 2020, with applied the data assimilation of SWH altimetric observations. Finally, IBI-MFC has made progress on the wave-ocean coupling of their NRT PHY and WAV forecast products. Thus, the impact of surface currents on waves was activated in July 2020, whereas the wave contributions (Stokes drift and stress modification) in the IBI forecasted ocean currents was included later, in December 2020. The IBI wave forecasts are reliable and accurate, and the product quality, as well as the potential enhancements, are assessed by comparing IBI wave outputs (for significant wave height, mean period and wave direction) to observations, from both in-situ moorings and satellite remote sensed products (Toledano et al., 2021). Figure 3 shows an example of accuracy level of the IBI WAV NRT product when compared with the independent (nonassimilated) HY2A altimetric observations.

Figure 3: Top panels: Validation of IBI WAV NRT forecast product for SWH with the (non-assimilated) HY2A SWH Altimetry observations (Maps of scatter index and bias, together with a scatter plot are shown). Bottom panels: Validation of IBI WAV MY reanalysis product with in-situ observations: SWH RMSD at moored buoys; Taylor diagram showing comparison of the present IBI MY product (currently delivered through the Copernicus Marine Service catalogue) and the previous IBI wave reanalysis product released.

151

#57

Since the beginning of Copernicus 1, a wave MY product was delivered. This IBI-WAV-MY product was initially based on a 1/10º MF-WAM non-assimilative hindcast forced by the ECMWF ERA-Interim. The period initially covered was 1993-2016 and this first system didn’t include data assimilation nor any coupling with surface currents, and it was nested into the ECMWF global wave reanalysis. The IBI wave MY system was fully upgraded in April 2018 and a completely new product was delivered. This new wave product increases the horizontal resolution by a factor of 2, to 1/20º, and it is based on reanalysis (a DA scheme for altimetric SWH observations, analogous to IBI NRT WAV). The atmospheric forcing was also updated and became the new wave reanalysis driven by the ECMWF hourly ERA5 reanalysis. Likewise, the open boundary condition was updated, using wave spectra boundary conditions from the Copernicus Marine Service GLOBAL wave reanalysis solution. Furthermore, this new IBI-wave MY system has an off-line coupling with the IBI MY ocean circulation reanalysis to include impacts that surface currents have on the sea states. All these upgrades resulted in a significant enhancement of the IBI MY wave product quality. An example of this is shown in Figure 3, and further details can be seen in the Copernicus Marine Service product quality document of the product (San Martin et al., 2020).

3. THE IBI-MFC SERVICE TODAY As shown, the IBI-MFC provides an operational service in constant evolution. In the last years, the number of datasets offered by the IBI-MFC to the Copernicus Marine Service catalogue has steadily grown from 3 products (7 datasets) in 2016 to 6 IBI products (accounting for 20 datasets) in 2021. Thus, the current IBI service delivers 37 variables covering the Blue and Green ocean from physics (temperature, salinity, currents, sea level, etc.), waves (significant wave height, peak period, etc.) and biogeochemistry (Chlorophyll, oxygen, nutrients, etc.). These variables currently encompass a wide range of temporal frequencies (i.e., 15-minute data, together with hourly, daily, and monthly means). The core of the current IBI-MFC service lies in operational suites that embed models in charge of producing the different NRT forecast, analysis and MY reanalysis/hindcast

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

outputs. In particular, these operational suites manage: - the acquisition of the best upstream data available to force models and assimilate observations, - the running of models and post-processing procedures for generating the products, - the validation of modelling solutions based on comparisons against all available observational data, - the timely delivery to users of the final IBI products. To supervise the proper operation, the IBI-MFC relies on several monitoring tools to check operation status, in terms of upstream sources availability, HPC (High-Performance Computing) & storage resources, control of different suites’ phases and a correct product dissemination. On top of it, the IBI-MFC relies on a Local Service Desk responsible for: incidents management, programmed service outages, updates and changes, and user relations. The rigorous design, control, and management of the suites and all related incidents ensure a highly reliable, robust and fully monitored IBI-MFC service, with timeliness always higher than 94%. Together with the forecast, analysis, and reanalysis products accessible through the Copernicus Marine Service catalogue, the IBI-MFC also contributes to monitor the health of the ocean over past decades with the delivery of tailored ocean monitoring indicators (OMI). Several IBI OMIs have already been published in Copernicus Marine Service Ocean State Reports (OSRs): the IBI Coastal Upwelling Index (OSR#1, 2016), the Mediterranean Outflow Water OMI (OSR#2, 2018), OMIs on extreme variability of Sea Surface Temperature (SST) and Significant Wave Height (SWH) (OSR#3, 2019) and the variability of stormy wave events index on IBI (OSR#4, 2020). Most of these IBI OMIs issued in OSRs are delivered through the Copernicus Marine Service catalogue and operationally updated (based on the last reanalysis products extensions available). The availability of HPC resources is a key element to sustain the here-described IBI-MFC operational service and to ensure its future evolution. Finally, IBI-MFC provides a service always devoted to meet requirements of an ever-growing community of end-users (from 178 users in 2014 to more than 500 in 2020 with half of them operational regular users downloading IBI products at least twice a week). The increasing interest in using IBIMFC products is reflected by the 100 TB of IBI physics, biogeochemistry and waves data were downloaded in 2020 by users (mainly from academia, private companies and public organizations).

152

#57

4. THE IBI-MFC ROADMAP The IBI-MFC roadmap is aligned with Copernicus Marine Service scientific objectives, and along Copernicus 2, the following amelioration goals will be pursued: 1. Increase products resolution. 2. Upgrade models used in the NRT and MY production. 3. Enhance Data Assimilation with better use of any new improved observational product available and special interest on DA benefits on shelf and coastal areas. 4. Increase system interactions through model coupling (interactions between ocean, waves, atmosphere, land -mainly its hydrological part-, and biogeochemistry). 5. Explore new products based on probabilistic approaches. By means of ensembles (applied both to models and DA schemes) to provide end users with product’s uncertainties. Furthermore, the IBI service evolution roadmap responds, as main driver, to meet identified end-users needs. Thus,

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

some research and development actions, planned to evolve the post-2021 IBI service, will focus on: 1. Ensure service continuity and keep present NRT forecast up-to-date (mostly in daily basis with product availability before midday) bringing MY product updates closer to present time (a monthly update to be achieved by new “Interim” reanalysis streams). 2. Enhance homogeneity in services and products. In particular, increase analogy between NRT and MY IBI products (i.e., using same grid to deliver same variables at same temporal frequency). 3. Make IBI products more coastal-oriented. Many IBI-MFC end-users are coastal ones, very interested in on-shelf processes. Thus, the IBI-MFC put a good deal of effort into IBI products to get a better representation of ocean scales driving the coastal ocean (including mesoscale, sub-mesoscale features such as eddies and fronts and a more accurate reproduction of high frequency processes). Enhancement of the IBI model capacities on shelf and coastal environments will foster further IBI down-streaming by coastal stakeholders.

ACKNOWLEDGEMENTS The authors acknowledge all experts who participated along Copernicus 1 Service as IBI-MFC team members, contributing to the production, delivery, evaluation, evolution and dissemination of our IBI products, including: E. Álvarez-Fanjul, A. Amo, M. Benkiran, P. Bowyer, S. Cailleau, A. Dalphinet, C. Fernandez, L. García, J.M. García-Valdecasas, M. Ghantous, K. Guihou, I. Lopez, A. Pascual, P. Rey, R Renaud, A. Rodriguez and C. Toledano.

153

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

REFERENCES Amo A, G Reffray, M G. Sotillo, R Aznar, K Guihou. Product User Manual (PUM) for Atlantic -Iberian Biscay Irish- Ocean Physics Analysis and Forecasting Product: IBI_ANALYSIS_FORECAST_ PHYS_005_001 https:// resources.marine.copernicus. eu/documents/PUM/CMEMSIBI-PUM-005-001.pdf (last accessed on 29/03/2021). Aznar R, M.G. Sotillo, S. Cailleau, P. Lorente, B. Levier, A. Amo-Baladrón, G. Reffray, E.Álvarez-Fanjul. (2016) Strengths and weaknesses of the CMEMS forecasted and reanalyzed solutions for the Iberia–Biscay–Ireland (IBI) waters. Journal of Marine Systems 159 1–14. Gutknecht, E., Reffray, G., Mignot, A., Dabrowski, T., and Sotillo, M. G.: (2019) Modelling the marine ecosystem of Iberia–Biscay–Ireland (IBI) European waters for CMEMS operational applications, Ocean Sci., 15, 1489–1516, https://doi. org/10.5194/os-15-1489-2019 IMMERSE Project (Improving Ocean Models for the Copernicus Programme). EU H2020 Programme. (2021) https:// immerse-ocean.eu/#about (accessed on 15/03/2021). Lambda Project. CMEMS Service Evolution Project. http://www. cmems-lambda.eu/ (accessed on 15/03/2021). Levier B, P Lorente, G Reffray, M Sotillo. (2021) CMEMS Quality Information Document (QUID) for Atlantic -Iberian Biscay Irish- Ocean Physics multi-year reanalysis Product: IBI_MULTIYEAR_PHY_005_002 https://resources.marine. copernicus.eu/documents/ QUID/CMEMS-IBI-QUID-005-002. pdf (accessed on 15/03/2021).

McGovern J.V., Dabrowski T., Pereiro D., Gutknecht E., Lorente P., Reffray G., Sotillo M.G (2020) CMEMS Quality Information Document (QUID) for Atlantic IBI -Iberian Biscay Irish-Biogeochemical Multiyear ProductIBI_MULTIYEAR_ BGC_005_003http://marine. copernicus.eu/documents/QUID/ CMEMS-IBI-QUID-005-003.pdf.

Lorente, P., Sotillo, M., AmoBaladrón, A., Aznar, R., Levier, B., Sánchez-Garrido, J. C., Sammartino, S., de PascualCollar, Á., Reffray, G., Toledano, C., and Álvarez-Fanjul, E.: (2019) Skill assessment of global, regional, and coastal circulation forecast models: evaluating the benefits of dynamical downscaling in IBI (Iberia–Biscay–Ireland) surface waters, Ocean Sci., 15, 967–996, doi:10.5194/os-15-967-2019. Mason, E., Ruiz, S., BourdalleBadie, R., Reffray, G., GarcíaSotillo, M., and Pascual, A. (2019) New insight into 3-D mesoscale eddy properties from CMEMS operational models in the western Mediterranean, Ocean Sci., 15, 1111–1131, https://doi.org/10.5194/os-151111-2019 Maraldi, C., Chanut, J., Levier, B., Ayoub, N., De Mey, P., Reffray, G., Lyard, F., Cailleau, S., Drévillon, M., Fanjul, E. A., Sotillo, M. G., Marsaleix, P., and the Mercator Research and Development Team: NEMO on the shelf: assessment of the Iberia–Biscay–Ireland configuration, Ocean Sci., 9, 745– 771, https://doi.org/10.5194/ os-9-745-2013, 2013. MEDSUB Project. CMEMS Service Evolution Project. https://www.mercator-ocean. fr/en/portfolio/medsub-2/ (accessed on 15/03/2021). Mayorga, E., Seitzinger, S. P., Harrison, J. A., Dumont, E., Beusen, A. H. W., Bowman, A. F., Fekete, B. M., Kroeze, C., and Van Drecht, G.: Global Nutrient Export from WaterSheds 2 (NEWS 2): model development and implementation, J. Environ. Modell. Softw., 25, 837–853, 2010.

San Martín L G, E Barrera, C Toledano, A Dalphinet, M Ghantous, L Aouf, P Lorente, M de Alfonso, M G Sotillo (2020). Quality Information Document (QUID) for Atlantic -Iberian Biscay Irish- Wave Multi-year reanalysis Product: IBI_ANALYSIS_FORECAST_ WAV_005_006 https://resources. marine.copernicus.eu/ documents/QUID/CMEMS-IBIQUID-005-006.pdf (accessed on 15/03/2021). Sotillo MG, Cailleau S, Lorente P, Levier B, Aznar R, Reffray G, Amo-Baladrón A, AlvarezFanjul E (2015) The MyOcean IBI ocean forecast and reanalysis systems: operational products and roadmap to the future Copernicus service. J Oper Oceanogr. 8(1):63–79. https:// doi.org/10.1080/175587 6X.2015.1014663 Sotillo M G, B Levier, P Lorente, K Guihou, R Aznar, A Amo, L Aouf, M Ghantous (2020) CMEMS Quality Information Document (QUID) for Atlantic -Iberian Biscay Irish- Ocean Physics Analysis and Forecasting Product:IBI_ ANALYSISFORECAST_ PHYS_005_001 https:// resources.marine.copernicus. eu/documents/QUID/CMEMSIBI-QUID-005-001.pdf (accessed on 15/03/2021).

154

Sotillo MG, Campuzano F, Guihou K, Lorente P, Olmedo E, Matulka A, Santos F, AmoBaladrón MA, Novellino A. River Freshwater Contribution in Operational Ocean Models along the European Atlantic Façade: Impact of a New River Discharge Forcing Data on the CMEMS IBI Regional Model Solution. Journal of Marine Science and Engineering. 2021; 9(4):401. https://doi.org/10.3390/ jmse9040401 Toledano C, A Dalphinet, P Lorente, M de Alfonso, M Ghantous, L Aouf, M G Sotillo (2021) Quality Information Document (QUID) for Atlantic -Iberian Biscay IrishWave Analysis and Forecasting Product: IBI_ANALYSIS_ FORECAST_WAV_005_005 https://resources.marine. copernicus.eu/documents/ QUID/CMEMS-IBI-QUID-005-005. pdf (accessed on 15/03/2021).

THE MEDITERRANEAN MONITORING AND FORECASTING CENTER COPPINI, G.1, CLEMENTI, E.2, COSSARINI, G.3,

KORRES, G.4, AYDOGDU, A.2, DRUDI, M.1, ESCUDIER, R.2, LECCI, R.1, PISTOIA, J.2, RAVDAS, M.4, SALON, S.3, TERUZZI, A.3, ZACHARIOUDAKI, A.4 1 Ocean Predictions and Applications Division, Centro Euro-Mediterraneo sui Cambiamenti Climatici - CMCC, Italy Ocean Modeling and Data Assimilation Division, Centro Euro-Mediterraneo sui Cambiamenti Climatici - CMCC, Italy 3 Istituto Nazionale di Oceanografia e di Geofisica Sperimentale - OGS, Italy - 4Hellenic Centre for Marine Research - HCMR, Greece 2

OVERVIEW

1. MAIN ACHIEVEMENTS 2015 - 2021

The Copernicus Marine Environment Monitoring Service (CMEMS) Med-MFC (Mediterranean Monitoring and Forecasting Center) is a consortium led by CMCC and gathers three European research institutes (CMCC, OGS and HCMR). The implemented service adopts a state-ofthe-art modelling development to operationally provide Near Real Time (NRT) and Reanalysis (RAN) products for the Mediterranean Sea dynamics, from currents to waves, and biogeochemistry. The user-driven nature of this organization fosters the commitment to ensure benefits to major users and to key target applications such as: - oil spill emergency management, - industrial and private sector applications, MSFD implementation, and - climate trend monitoring.

At the launch of Copernicus 1 in 2015, the Med-MFC consortium was delivering Mediterranean physical and biogeochemical NRT and Reanalysis products. In 2017, the wave component was also integrated. Figure 1 shows a schematic summary of the NRT systems’ evolutions since 2015, which mainly consist of the continuous improvement of numerical models and data assimilation systems. Details on evolutions are provided in the following subsections for each system.

The consortium ensures a cutting-edge system by yearly updates of the service catalogue.

155

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: Med-MFC NRT systems evolutions since 2015.

1.1 Med-MFC Physical Systems In 2015 the Med-MFC physical component was delivering both NRT and RAN products. The Mediterranean analysis and forecast system, MedFS, (Pinardi and Coppini, 2010; Tonani et al., 2014) was composed of a two-way coupled hydrodynamic-wave modelling system (Clementi et al., 2017a). It was based on NEMO v3.4 and WaveWatchIII models implemented over the whole Mediterranean Sea with a horizontal resolution of 1/16˚ (ca. 6 km) and 72 vertical levels. The Atlantic lateral open boundary conditions were provided by nesting into the Global Ocean daily analyses and forecasts fields. ECMWF (European Centre for Medium-Range Weather Forecasts) 1/8˚ surface atmospheric analyses and forecast fields (sea level pressure, cloud cover, humidity, air temperature, winds) were used with a temporal frequency of 6 hours (3 hours for the first 3 days of forecast). Meanwhile climatological monthly precipitations were exploited to constrain the Mediterranean Sea long-term water budget. Surface nonsolar heat fluxes were corrected through a whole day relaxation to the Sea Surface Temperature (SST) satellite Level 4 Copernicus Marine Service product. The land river runoff was imposed as monthly climatologies for seven major rivers around the Mediterranean Sea (Ebro, Rhone, Po, Vjosë, Seman, Buna-Bojana, Nile). The Dardanelles strait inflow was also parameterized as a river input. The OceanVar (3D variational) data assimilation system (Dobricic and Pinardi, 2008) was used to assimilate Sea Level Anomaly (SLA) along tracks data from altimeters

(Sea Level Thematic Assembly Center/SL TAC) and in situ vertical profiles of temperature and salinity (In-Situ Thematic Assembly Center/INS TAC) by means of a daily assimilation cycle. The modelling system has continuously evolved since 2015 (see Figure 1) to meet the increasing demand for: - higher model accuracy, - more complete representation of dynamical processes, - finer spatial scales, - improved capability in storm surge forecasting including tidal forcing, - improved representation of boundary conditions. Following those objectives, model evolutions in the last 6 years included: - increase of system resolution to 1/24° (ca 3.5 km) and 141 vertical levels (Clementi et al., 2017b), - update of the hydrodynamic model to NEMO v3.6, - increase of the river inputs (from 7 to 39), - implementation of lateral open boundary conditions in the Dardanelles strait (Delrosso, 2020), - use of higher ECMWF atmospheric forcing (1/10°), - inclusion of tides. Furthermore, a continuous upgrade of the data assimilation component together with the assimilation of new observations such as SLA Sentinel-3A/B data and an improved SST nudging were achieved. Those model upgrades led to a quality increase of the analysis and forecast product, which is continuously monitored.

156

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 2: Timeseries of temperature Root Mean Square Difference (RMSD) between model and observations in a 10 to 30 meters layer for model analysis (solid line) and 3rd day of forecast (dashed line). Red lines: 1/16° model; blue lines: 1/24° model resolution. Dashed areas represent the number of observations.

An example is shown in Figure 2 presenting a timeseries of temperature Root Mean Square Difference (RMSD) between model and in situ observations averaged between 10-30 meters. The skill in this layer is dominated by a seasonal cycle characterized by a larger error in the summer period due to the temperature stratification and a shallow thermocline. Both temperature model analysis and the 3rd day forecast show a reducing error thanks to system improvements. The Physical Reanalysis system upgrades since 2015 were integrated in the new reanalysis timeseries delivered in 2020 (Escudier et al., 2020). The main differences between the reanalysis produced since the beginning of the Copernicus Marine Service (Simoncelli et al., 2019) and the new one consist of: - increase of system resolution to 1/24° (ca 3.5 km) and 141 vertical levels and improved bathymetry, - update of the hydrodynamic model to NEMO v3.6, - increase of the river inputs (from 7 to 39), - daily lateral open boundary conditions in the Atlantic derived from a global reanalysis (instead of monthly climatologies), - use of ERA5 (instead of ERA-Interim) atmospheric forcing, - update of the assimilation scheme and assimilation of a larger number of observations. The Physical Reanalysis presents an overall increased skill and a reduced RMSD for all variables and at all depths when compared to observations. Also, the representation of the mixed layer depth and deep convection events in areas of water mass formation is improved compared to state-of-the-art climatology and literature (more details in the product QUID).

1.2 Med-MFC Biogeochemical Systems The Biogeochemical analysis and Forecast modelling system (MedBFM) features the coupled transport-biogeochemical model (OGSTM-BFM, Lazzari et al., 2010) and the 3DVarBio assimilation scheme (Teruzzi et al., 2014). The MedBFM is off-line coupled with MedFS over the whole Mediterranean Sea. The OGSTM transport model resolves advection, vertical diffusion and sinking terms of biogeochemical variables. The BFM model describes biogeochemical cycles of 4 chemical compounds (carbon, nitrogen, phosphorus and silicon) through the dissolved inorganic, living organic and non-living organic compartments. Organic compartments include four phytoplankton groups (diatoms, flagellates, picophytoplankton and dinoflagellates), four heterotrophic zooplankton groups (carnivorous and omnivorous mesozooplankton, heterotrophic nanoflagellates and microzooplankton), one heterotrophic bacteria and four non-living compartments (labile, semilabile and refractory dissolved matter and particulate matter). The 3DVarBio is a variational scheme that decomposes the background error covariance matrix using a sequence of operators (Dobricic and Pinardi, 2008). The covariance operators account separately for vertical, horizontal and biogeochemical covariance (Teruzzi et al., 2014). In 2015, the system assimilated surface Chlorophyll from satellite sensors for the open-sea area.

157

#57

Since 2015, the biogeochemical component was continuously improved (Figure 1). The BFM was improved by: - revising nutrient formulations of phytoplankton (Lazzari et al., 2016), - including the carbonate system (Cossarini et al., 2015), - upgrading the optical component with a new Kd dataset and the day-night cycle in photosynthesis formulation (Salon et al., 2019; Feudale et al., 2021). The OGSTM transport model improvements included a new parallel implementation and the parameterization of z* level formulation (Salon et al., 2019). Regarding the 3DVarBio data assimilation scheme, developments consisted in: - assimilation of coastal data (Teruzzi et al., 2018), - scheme parallelization, - upgrade for multiplatform satellite and BGC-Argo profiles assimilation (Cossarini et al., 2019).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The off-line coupling between the physical component and the MedBFM guarantees the full alignment in terms of bathymetry, horizontal and vertical resolution (i.e., 1/24° and 141 vertical levels since 2017 version), the consistency of open boundaries (i.e., Atlantic and Dardanelles) and land forcing (i.e, 39 rivers since 2017 version). As a result of continuous developments, the quality of BIO products has significantly improved over the MyOcean and Copernicus 1 period. Indeed, timeseries of the RMSD of the first forecast day (Figure 3) shows the decrease of the error of Chlorophyll due to some relevant system developments. Other mentioned improvements, not visible on Figure 3, are fully accounted in the thorough validation framework implemented on the regional validation website (Salon et al., 2019) and the latest version of the NRT Copernicus Marine Service documentation (Feudale et al., 2021).

Figure 3: Timeseries of the RMSD of the first day of forecast of Chlorophyll computed after the weekly analysis run (blue solid line). Statistics have been computed pre-operationally in 2011-2012 and operationally since January 2013. Timings of changes in horizontal resolution (green) and in biogeochemical model and data assimilation (red) are indicated in the upper part of the chart.

During the 6 years of the Copernicus Marine Service program, two biogeochemical reanalyses have been produced integrating improvements and developments from the NRT Mediterranean Sea model systems. From the first (2015) to the latest (2021) one, the major improvements of the setup include: - the increase of resolution from 1/16° to 1/24° and from 71 to 125 vertical levels,

- the update of the atmospheric (e.g., ERA5) and land (e.g., nutrient loads from Perseus FP7 project) forcing and boundary (e.g., WOA2018 for the Atlantic boundary), - the update of initial conditions (e.g., Emodnet 2018). The 2021 biogeochemical reanalysis provides CMEMS with a 1999-2020 timeseries of 12 validated variables (NO3, PO4, NH4, Chl, PhytoC, ZooC, npp, oxygen, Alk, DIC, pH, pCO2, airsea CO2 flux).

158

#57

1.3 Med-MFC Waves Systems The evolution of the Med-MFC NRT wave component (Med-waves) is shown in Figure 1. First released in 2017, it provided (short term) wave forecasts and simulations for the Mediterranean Sea on a daily basis (i.e., 5 days). It had a 1/24° horizontal resolution, 32 frequency bins and 24 equally spaced directional bins for the discretization of the wave spectrum solved by the wave model (Ravdas et al., 2018). The Med-waves set-up included a coarse grid domain with a resolution of 1/6° covering the North Atlantic Ocean. It was developed and implemented using a stateof-the-art third-generation wave model, WAM Cycle 4.5.4. In March 2018, the system was upgraded by incorporating a data assimilation component to leverage available alongtrack Significant Wave Height (SWH) satellite observations gathered from Sentinel-3A and Jason-3 (Sea LevelTAC). The assimilation module was based on an optimal interpolation of the total SWH retrieved by altimeters (Lionello et al., 1992). In 2019, the wave model upgraded to Cycle 4.6.2 and the forecast was extended up to 10 days. Additionally, modifications were introduced to the wave age parameter (ZALP = 0.011) whitecapping dissipation coefficients, while a limitation at high frequencies matching the latest ECMWF wave forecasting system Cy43r1 (ECMWF, 2016) was applied in order to reduce the wave steepness at very high wind speeds. The same year, new along track inter-calibrated SWH observations from Cryosat-2 and Saral/Altika satellite missions were added to the data assimilation module while Sentinel-3b observations were added in 2020.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The quality of the Med-Waves analysis and forecast system over the period April 2017 – April 2021 (release in 2021) was continuously assessed over yearly reference periods (2014 reference period until the version released in Feb 2018, 2016 until the release of April 2019 and Jul 2019Jul 2020 for the latest versions of the system). It enabled the comparison between in situ observations from moored wave buoys (INS-TAC) and satellite altimeter observations (merged altimeter SWH database setup at CERSAT IFREMER and Wave-TAC). The overall quality of the system then showed annual scatter indexes (SI) equal to 24% for in situ and 15% for satellites SWH observations. The quality evolution of SWH and Mean Wave Period (MWP) components is shown in Figure 4 separately for in situ and satellite observations. Since April 2017, a continuous improvement of the SWH satellite observations’ quality is visible. When comparing Med-waves SWH with in situ observations a drop in normalized RMSD is evident in April 2018 while the quality remains almost unchanged afterwards. As for the MWP, the normalized RMSD with respect to buoy observations shows a decreasing trend from October 2017 release until 2021. Until 2020, the multi-year wave product consisted of a 13-year wave hindcast (2006 – 2019) produced using the 2017 Med wave NRT system, forced by 6-hours operational analysis 10 m ECMWF winds and daily surface currents from Copernicus Marine Service physical reanalysis. Since 2021, the multi-year wave product is a 27-year wave reanalysis for the period 1993 - 2019 (described in the following section).

159

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Metrics of simulation/analysis/first-guess NRT Med-waves SWH (left) and MWP (right) computed over a reference period of 1-year (indicated by the black triangles) and for the full Mediterranean domain, across system versions (release dates shown in the abscissa).

2. STATUS AT THE END OF COPERNICUS 1 At the end of Copernicus 1, the Med-MFC provides upgraded NRT and Reanalysis products for each component. All systems share the same grid resolution (1/24°), bathymetry and forcing fields (atmospheric and river). In particular, the NRT physical system includes tidal waves (Clementi et al., 2020). Tidal forcing is applied along the lateral boundaries in the Atlantic Ocean and the tidal potential is calculated across the domain for the 8 major tidal constituents of the Mediterranean Sea. Moreover, the data assimilation scheme is updated to account for the tidal signal in the SLA altimeter tracks. A new variable, the de-tided sea level, is also delivered to prevent possible inconsistencies in downstream applications not accounting for tides. The validation of the new system shows an overall good skill of all considered tidal constituents when compared to tide gauges, literature values and barotropic tidal model solutions. The physical Multi-Year product is also enlarged by including an INTERIM dataset produced by forcing the reanalysis system with ECMWF ERA5 and ERA5-T atmospheric fields and assimilating NRT observations (Nigam et al., 2020). This new dataset covers the period from the last day of reanalysis up to 1 month before present and is extended on a monthly basis.

validation framework that includes historical (WOA 2018, Emodnet 2018, Socat 2019), operational in situ (e.g., BGCArgo) and satellite (OceanColour-TAC) datasets. Latest updates are: - multiplatform (Satellite and BGC-Argo) and multivariate (nitrate and Chlorophyll) data assimilation, - the latest BFMv5.2 biogeochemical version (code updates), - new Atlantic open boundary conditions, - dependencies, such as the Kd (Ocean Colour-TAC) for bio-optics coupling. These advancements are stepping stones for the next Copernicus 2 phase of developments of the biogeochemical system. By the way, a new biogeochemical reanalysis was delivered in May 2021 (see Table 2). The Med-waves NRT system, run twice per day, operates hourly surface MedFS currents and sea level and wave refraction. Meanwhile CFOSAT and H2B SWH observations are additionally ingested into the assimilation module to increase the quality of the Mediterranean wave analyses. In May 2021, a new 27-year (1993 - 2019) wave reanalysis was delivered. It is produced using ERA5 10 m hourly winds, reanalysis surface currents and the assimilation of along-track SWH observations from all available satellite missions over the period. Intercomparison with previous multi-year timeseries (hindcast over the period 2006 – 2019) using wave buoy observations shows that the two sets have very similar skill during their overlapping period.

The biogeochemical NRT system provides 14 validated products as a result of an impressive improvement of the

160

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

products in coastal areas require better physical coastal dynamics, integration of high frequency land forcing data and specific skill performance assessment.

3. POST 2021 PERSPECTIVES Med-MFC post 2021 perspectives include several major system upgrades for all 3 components. In particular the physical system future improvements will account for: - a continuous physical model development toward a fully coupled physics-wave system will be performed, -a different vertical mixing scheme will be used, and - both the NEMO and WWIII models will be updated to the latest available versions. The system will Include a better representation of rivers in coastal areas using daily river inflow (where available) and river discharge. An ensemble ocean forecasting will be implemented by perturbing the atmospheric forcing and initial conditions. Moreover, the data assimilation scheme and the assimilation of gliders will be improved. Additionally, new and higher resolution observations will be assimilated. A new physical reanalysis will be delivered including tides, updated boundary conditions in the Atlantic and Dardanelles straits and higher resolution atmospheric forcing. Future upgrades of the BFM system will target the accuracy of phytoplankton and zooplankton compartments to provide indicators of plankton community diversity, and energy and matter pathways across the ecosystem. Parameterization of silicon and oxygen cycles and integration of optics and biogeochemistry are other important drivers of the future developments of the BFM models. Future demands for high quality biogeochemical

On a longer-term perspective the inclusion of benthicpelagic interactions, land-ocean coupling, high-resolution modelling, integration of biogeochemistry with pollutant modelling and dedicated data assimilation for coastal areas are further drivers to improve reliability and capability of biogeochemical operational products. Regarding assimilation, new developments will focus on adding new in situ and satellite data and evolving toward hybrid or ensemble 3Dvar methods and joint physicalbiogeochemical schemes. One of the main drivers for the future evolution of the MedWaves system is to increase the accuracy of the wave analyses and 5-10 days forecasts while providing additional products to end-users. This can be achieved by: - improving the wind forcing used to drive the WAM model, - assimilating additional satellite (wave spectrum from SAR) or in situ wave measurements, - improving the WAM physics according to recent developments of relevant projects (i.e., WAVEFLOW). Extreme waves forecasting is another path of future evolution of the system that can be achieved by implementing recent developments of LATEMAR CMEMS service evolution project (Benetazzo et al., 2020). Finally, the clear need to define and to define and provide uncertainty for current wave predictions will be achieved through the development and operational implementation of a Wave Ensemble Prediction System.

161

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

ACKNOWLEDGMENTS We acknowledge all Med-MFC team members contributing to systems’ evolutions: - CMCC (Euro-Mediterranean Center on Climate Change): Giovanni Coppini, Emanuela Clementi, Ali Aydogdou, Romain Escudier, Jenny Pistoia, Mohamed Omar, Anna Chiara Goglio, Tanuja Nigam, Rita Lecci, Massimiliano Drudi, Alessandro Grandi, Antonio Mariani, Vladyslav Lyubartsev, Sergio Cretí, Francesco Palermo, Paola Agostini, Simona Masina, Nadia Pinardi. - OGS (National Institute of Oceanography and Applied Geophysics): Gianpiero Cossarini, Stefano Salon, Anna Teruzzi, Giorgio Bolzon, Gianluca Coidessa, Valeria Di Biagio, Laura Feudale, Paolo Lazzari, Carolina Amadio, Alberto Brosic. - HCMR (Hellenic Centre for Marine Research): Gerasimos Korres, Michalis Ravdas, Anna Zacharioudaki, Dimitra Denaxa, Maria Sotiropoulou. - INGV (National Institute of Geophysics and Volcanology): Damiano Delrosso, Claudia Fratianni, Gelsomina Mattia, Antonio Guarnieri, Simona Simoncelli, Pierluigi Di Pietro, Paolo Oliveri, Stefano Marino.

REFERENCES Benetazzo, A., Barbariol, F., Pezzutto, P., Bertotti, L., Cavaleri, L., Davison, S., and Sclavo, M., (2020). Toward a unified framework for maximum wave computation from numerical models: outcomes from the LATEMAR project., EGU General Assembly 2020, EGU2020-13683, https:// doi.org/10.5194/egusphereegu2020-13683 Clementi, E., Oddo, P., Drudi, M., Pinardi, N., Korres, G., Grandi A., (2017a). Coupling hydrodynamic and wave models: first step and sensitivity experiments in the Mediterranean Sea. Ocean Dynamics. doi: https://doi. org/10.1007/s10236-017-1087-7.

Clementi, E., Pistoia, J., Delrosso, D., Mattia, G., Fratianni, C., Storto, A., Ciliberti, S., Lemieux, B., Fenu, E., Simoncelli, S., Drudi, M., Grandi, A., Padeletti, D., Di Pietro, P., Pinardi, N., (2017b). A 1/24 degree resolution Mediterranean analysis and forecast modelling system for the Copernicus Marine Environment Monitoring Service. Extended abstract 8th EuroGOOS Conference, Bergen. Clementi, E., Aydogdu, A., Goglio, A. C., Pistoia, J., Escudier, R., Drudi, M., Grandi, A., Mariani, A., Lyubartsev, V., Lecci, R., Cretí, S., Coppini, G., Masina, S., Pinardi, N., (2021). Mediterranean Sea Analysis and Forecast (CMEMS MEDCurrents, EAS6 system) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https:// doi.org/10.25423/CMCC/ MEDSEA_ANALYSISFORECAST_ PHY_006_013_EAS6.

Cossarini, G., Lazzari, P., Solidoro, C., (2015). Spatiotemporal variability of alkalinity in the Mediterranean Sea. Biogeosciences, 12(6), 1647-1658. Cossarini, G., Mariotti, L., Feudale, L., Mignot, A., Salon, S., Taillandier, V., Teruzzi,, A., D’Ortenzio F., (2019). Toward operational 3D-Var assimilation of Chlorophyll BiogeochemicalArgo float data into a biogeochemical model of the Mediterranean Sea. Ocean Modelling, 133, pp. 112-128. DOI: 10.1016/j.ocemod.2018.11.005. Delrosso, D., (2020). Numerical modelling and analysis of riverine influences in the Mediterranean Sea, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Geofisica, 32 Ciclo. DOI 10.6092/unibo/ amsdottorato/9392.

162

Dobricic, S. and Pinardi, N., (2008). An oceanographic three-dimensional variational data ssimilation scheme. Ocean Modelling, 22 (3-4) 89-105. - ECMWF: ifs.documentation CY43R1. Part VII: ECMWF Wave - Model documentation 2016, [online] Available from: https:// www.ecmwf.int/sites/default/ files/elibrary/2016/17120-partvii-ecmwf-wave-model.pdf Escudier, R., Clementi, E., Omar, M., Cipollone, A., Pistoia, J., Aydogdu, A., Drudi, M., Grandi, A., Lyu-bartsev, V., Lecci, R., Cretí, S., Masina, S., Coppini, G., Pinardi, N., (2020). Mediterranean Sea Physi-cal Reanalysis (CMEMS MEDCurrents) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service. https://doi.org/10.25423/ CMCC/MEDSEA_MULTIYEAR_ PHY_006_004_E3R1.

#57

Feudale, L., Bolzon, G., Lazzari, P., Salon, S., Teruzzi, A., Di Biagio, V., Coidessa, G., & Cossarini, G. (2021). Mediterranean Sea Biogeochemical Analysis and Forecast (CMEMS MEDBiogeochemistry, MedBFM3 system) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service . https://doi.org/10.25423/CMCC/ MEDSEA_ANALYSISFORECAST_ BGC_006_014_MEDBFM3 Lazzari, P., Teruzzi, A., Salon, S., Campagna, S., Calonaci, C., Colella, S., Tonani, M., Crise, A., (2010). Pre-operational short-term forecasts for the Mediterranean Sea biogeochemistry. Ocean Science, 6, 25-39. Lazzari, P., Solidoro, C., Salon, S., Bolzon, G., (2016). Spatial variability of phosphate and nitrate in the Mediterranean Sea: a modelling approach. Deep Sea Research I, 108, 39-52. Lionello, P., Günther, H., Janssen, PAEM, (1992). Assimilation of altimeter data in a global third-generation wave model. Journal of Geophysical Research: Oceans 97 (C9), 14453-1447.

Nigam, T., Escudier, R., Pistoia, J., Aydogdu, A., Omar, M., Clementi, E., Cipollone, A., Drudi, M., Grandi, A., Mariani, A., Lyubartsev, V., Lecci, R., Cretí, S., Masina, S., Coppini, G., Pinardi, N., (2021). Mediterranean Sea Physical Reanalysis INTERIM (CMEMS MED-Currents, E3R1i system) (Version 1) [Data set]. Coper-nicus Monitoring Environment Marine Service (CMEMS). https://doi. org/10.25423/CMCC/MEDSEA_ MULTIYEAR_PHY_006_004_E3R1I. Pinardi, N. and Coppini, G., (2010). Operational oceanography in the Mediterranean Sea: the second stage of development. Ocean Sci., 6, 263–267, https://doi. org/10.5194/os-6-263-2010. Ravdas, M., Zacharioudaki, A., Korres, G., (2018). Implementation and validation of a new operational wave forecasting system of the Mediterranean Monitoring and Forecasting Centre in the framework of the Copernicus Marine Environment Monitoring Service, Nat. Hazards Earth Syst. Sci., 18(10), 2675–2695, doi:10.5194/ nhess-18-2675-2018.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Simoncelli, S., Fratianni, C., Pinardi, N., Grandi, A., Drudi, M., Oddo, P., Dobricic, S., (2019). Medi-terranean Sea Physical Reanalysis (CMEMS MED-Physics) (Version 1) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https:// doi.org/10.25423/MEDSEA_ REANALYSIS_PHYS_006_004. Salon, S., Cossarini, G., Bolzon, G., Feudale, L., Lazzari, P., Teruzzi, A., Solidoro, C., Crise, A., (2019). Marine Ecosystem forecasts: skill performance of the CMEMS Mediterranean Sea model system. Ocean Science, 15, 997-1022. Storto, A., Masina, S., Navarra, A., (2015). Evaluation of the CMCC eddy-permitting global ocean phys-ical reanalysis system (C-GLORS, 1982-2012) and its assimilation components. Quarterly Journal of the Royal Meteorological Society, 142, 738–758, doi:10.1002/qj.2673.

Teruzzi, A., Dobricic, S., Solidoro, C., Cossarini, G., (2014). A 3D variational assimilation scheme in coupled transport biogeochemical models: Forecast of Mediterranean biogeochemical properties, Journal of Geophysical Research, doi:10.1002/2013JC009277. Teruzzi, A., Bolzon, G., Salon, S., Lazzari, P., Solidoro, C., Cossarini, G., (2018). Assimilation of coastal and open sea biogeochemical data to improve phytoplankton simulation in the Mediterranean Sea. Ocean Modelling, 132, 46-60. Teruzzi, A., Bolzon, G., Cossarini, G., Lazzari, P., Salon, S., Crise, A., & Solidoro, C., (2019). Mediterranean Sea Biogeochemical Reanalysis (CMEMS MED-Biogeochemistry) [Data set]. Copernicus Monitoring Environment Marine Service (CMEMS). https://doi. org/10.25423/MEDSEA_ REANALYSIS_BIO_006_008 Tonani, M., Teruzzi, A., Korres, G., Pinardi, N., Crise, A., Adani, M., Oddo, P., Dobricic, S., Fratianni, C., Drudi, M., Salon, S., Grandi, A., Girardi, G., Lyubartsev, V., Marino, S., (2014). The Mediterranean Monitoring and Forecasting Centre, a component of the MyOcean system. Proceedings of the 6th Int. Conference on EuroGOOS 4-6 October 2011, Sopot, Poland. Edited by H. Dahlin, N.C. Fleming and S. E. Petersson. First published 2014. Eurogoos Publication no. 30. ISBN 978-91-974828-9-9.

163

NORTH WEST-EUROPEAN SHELF MONITORING AND FORECASTING CENTRE TONANI, M.¹, MAKSYMCZUK, J.1, GOLBECK, I.2, LORKOWSKI, I.2, LI, X.2, KAY, S.1,3, KING, R.R.1, PEQUIGNET A.-C.1, POLTON, J.A.4, SAULTER, A.1, SKAKALA, J.3, WAKELIN, S.L.4, ARTIOLI, Y.3, ASCIONE, I.¹, BRUCIAFERRI, D.¹, MCCONNELL, N.¹, SYKES, P.¹, RENSHAW, R.¹, SIDDORN, J.1,4 Met Office, Exeter, United Kingdom - 2Federal Maritime and Hydrographic Agency, Hamburg, Germany 3 Plymouth Marine Laboratory, United Kingdom - 4National Oceanography Centre, United Kingdom

has been developed and applied to inter-compare the accuracy of forecasting products at different resolutions.

OVERVIEW The North West-European Shelf (NWS) Monitoring and Forecasting Centre (MFC) delivers forecast and reanalysis products for the European shelf seas. This region is characterized by shallow seas tidally dominated. The model domain extends into the deep part of the Atlantic to properly resolve the across shelf exchanges of Atlantic water. The model eastern boundary covers the Kattegat/ Skagerrak straits for getting the Baltic water inflow, characterized by low salinity water. The NWS-MFC provides ocean, wave, and biogeochemical forecast and reanalysis products. All the components of the systems have been improved during the last five years, considering the user’s feedbacks and requirements. Wave products did not exist at the beginning of the project and have been added for both forecast and reanalysis. The resolution of the model was increased, from 7 to 1.5 km, for resolving the mesoscale and improving the resolution of the coastline. The wave component, added in 2017, had been recently coupled with the ocean, improving the representation of the ocean momentum budget equation. The accuracy of the initial conditions of all the components has been increased, thanks to the improvement of the data assimilation. More observations and new variables are now assimilated in both the physical and the biogeochemical models. It is of paramount importance to be able to understand the impact of these evolutions on the quality of the products delivered to users. Hence, all products are regularly assessed with classical verification metrics. New verification methods suited to increasingly complex and high-resolution models are being introduced. A new method based on spatial neighbourhood

1. MAIN ACHIEVEMENTS FROM 2015 TO 2021 To provide a state-of-the-art ocean forecasting system, the various components are regularly upgraded, and new components added. A significant effort on R&D activities has guaranteed a continuous pull-through of improvements. Not all the R&D developments have yet been implemented in the forecast and reanalysis systems but will be included in future evolutions of the systems. An overview of the major evolution of the forecast (named analysis_forecast in Copernicus Marine Service’s catalogue) and reanalysis (or multi-year) model systems is described in Figure 1 and Figure 2. Data assimilation has been improved, giving better initial conditions for both the physical (PHY) and the biogeochemical (BGC) components. At the beginning of Copernicus 1, the NWS systems assimilated only SeaSurface Temperature (SST) from in-situ and satellite observations, using a 3D-VAR scheme NEMOVAR (Waters et al., 2015). Vertical profiles of temperature and salinity and Sea Level Anomaly have been included in the observations assimilated in the PHY forecast system since March 2016 (King et al., 2018). Assimilating SLA in a regional, shallow and tidally dominated model is challenging. The first implementation assimilated SLA observations only where the model bathymetry was deeper than 700 m.

164

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: Major evolution of the NWS analysis and forecast components during Copernicus 1 in chronological order.

This restriction was overcome from more recent work by developing the capability to assimilate SLA also in the shelf areas, which increased the accuracy of the model SLA and of the water column physical properties. This is an important achievement that will be implemented in operations in the next cycle of developments and in the next generation of reanalysis. R&D work has also studied the impact of future high-resolution observations on data assimilation. A series of Observing System Simulation Experiments (OSSEs) has been carried out to assess the

potential impact of assimilating wide-swath altimeter SLA data from SWOT. The results of this set of experiments showed a clear positive impact off-shelf, but a smaller impact on-shelf. Further work will be required to assimilate SWOT observations with the expected large correlated errors. At the beginning of Copernicus, none of the NWS systems were assimilating Chlorophyll or other biogeochemical variables. The assimilation of satellite total Chlorophyll was introduced for the first time in the reanalysis timeseries

165

#57

in November 2018 (see Figure 2) and only recently in the forecasting system (Figure 1). The impact has been of highest importance for preventing an over-estimation of Chlorophyll during the spring bloom (see Figure 3). The Copernicus Marine Service Evolution (SE) project TOSCA developed the Phytoplankton Functional Type (PFT) data assimilation in the North West-European shelf region (Skakala et al., 2018). This evolution has been pulled through into the NWS reanalysis system to produce the new timeseries released in December 2020. This development has not only improved the impact of the data assimilation

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

on some phytoplankton types but has extended the biogeochemical variables available in the catalogue to include model PFT Chlorophyll fields. More recently, a multi-platform biogeochemical data assimilation study has tested the impact of combining ocean colour and glider measurements, along with temperature and salinity from multiple sources, in a pre-operational configuration of the system (Skakala et al., 2021). The interplay between the assimilated multi-platform physics and biogeochemistry has been strengthened by introducing feedback from biogeochemistry to physics showing some early promising

Figure 2: Major evolution of the NWS analysis and forecast components during Copernicus 1 in chronological order.

166

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Model validation against satellite observations. The major difference between reanalysis available at the beginning of Copernicus1 (V2 – green) and the new timeseries released in November 2018 (V4 -blue) is the addition of total Chl data assimilation. The satellite observations are in red.

results in improving the bloom timing and overall biogeochemistry. This work positions the MFC ready to use biogeochemical data from gliders when it becomes available; physics data from gliders is already assimilated. Biogeochemical Argo data will also be used in the off-shelf part of the region, but waters on the shelf are too shallow for Argo floats to operate. A recent study on feature-based verification methods has improved our capability to assess the spatial and temporal characteristics of the biogeochemical forecasts (Mittermaier et al., 2021) and will be used for the validation of future evolution of the system. Model components of the PHY (NEMO, O’Dea et al., 2017; Graham, 2018a) and BGC (ERSEM, Butenschön et al., 2016) have been updated in the last 6 years. Keeping the operational version of the models as close as possible to the most recent version contributes to reducing the gap between R&D and operational production, shortening and simplifying the process for pulling through the R&D developments into operations. Updating ERSEM to a more recent version enabled the addition of a carbonate module for the provision of pH and pCO2 products. These products are delivered for both reanalysis and forecast (see Figure 1 and Figure 2). Wave products were introduced for the first time in April 2017, and since then have quickly evolved. The products are generated using the phase averaged third-generation spectral wave model WAVEWATCH III (presently at version 4.18; Tolman et al., 2014). The first product had the resolution of 0.67 x 0.111 degrees (~7 km), the same as the physical and

biogeochemical systems at that time. In November 2018, the resolution increased to 0.0167 degrees (~1.5 km), doubling the number of the users (Figure 4, bottom-left panel). The WAV reanalysis product released in July 2020 (see Figure 2) at a resolution of 1.5 km was the first NWS reanalysis product at high resolution. Users’ feedback was very positive, and this product gained a considerable number of users soon after its addition to the catalogue (see Figure 4, red line). The PHY model (NEMO) was updated to v3.6 in the first part of Copernicus 1 for the forecast and the reanalysis. Several improvements have been made. Most significantly the horizontal resolution was refined to 1.5 km, enabling improved representation of tidal residual mean flow. This resolution allows simulating internal tides (Guihou et al., 2018), better representing fine-scale structures in velocity, temperature and salinity (Graham et al., 2018, Tonani et al., 2019) and simulates enhanced shelf exchange fluxes. This configuration was implemented operationally with data assimilation in November 2018 (Tonani et al., 2019). This was a step change in the capability of the NWS-MFC to forecast small mesoscale structures and to increase accuracy close to coastal areas. Figure 4 shows an abrupt increase in the number of users of the PHY forecast products as soon as the 1.5 km product was introduced, proving that the kilometric scale is needed and valued by users. The increased data volume has been a challenge from the beginning, with issues impacting users’ download and the delivery from the production centre to the catalogue. This is reflected in the latency of users to switch from the low- to the high-resolution product (Figure 4).

167

#57

The need to understand and validate the impact of the increased resolution has required an update of validation and verification techniques. High-resolution observations like gliders and high frequency (HF) radar have been used for assessing this new product. The NWS-MFC has pulled through the neighbourhood techniques developed by the Copernicus Marine Service SE Phase 2 HiVE project (Crocker et al., 2020) to assess the impact of the resolution on forecast fields. This methodology has also been used very recently to compare NWS products with other Copernicus Marine Service’s models, for regions where they overlap. In December 2020, the PHY and WAV forecast models have been coupled, thanks to the work done during CMEMS SE Phase 1 projects Nemo2wave and OWAIRS (Lewis et al., 2019). The PHY-WAV coupled system is coupled at hourly frequency and has three wave-currents interactions: -m odification of surface stress by wave growth and dissipation, - Stokes-Coriolis forcing, - wave-height-dependent ocean surface roughness. A study investigated the impact of the coupling on the sub-surface dynamics during storm events, assessing the accuracy of the currents against drifters (Bruciaferri et al., 2021). It demonstrated that the coupling improves the accuracy of the surface dynamics, with larger improvements on shelf (8%) than in the deep part of the NWS

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

domain (4%). Figure 5 shows an example of Lagrangian particles forced by the coupled and uncoupled version of the model, compared with drifter trajectories. The choice to focus on the horizontal current validation was user driven, since the most downloaded forecast variable is the hourly currents at 1.5 km resolution. Validation of ocean currents is challenging due to lack of high-resolution observations in this area. This R&D work was undertaken using drifters and HF Radar and will lead the way for future validation and verification processes for ocean currents. The tidal variation in the shoreline is appreciable in some regions with a model resolution of 1.5 km. A significant part of the coastal shelf area is very shallow and tidally dominated. NEMO version 3.6 does not allow grid cells to wet and dry so a minimum depth (10 m) was imposed for wet cells in shelf areas, where the tidal range is several metres. This choice decreases the reliability of the model in very shallow coastal areas. The processes of wetting and drying have been implemented on NEMO version 4 and tested in the NWS configuration (O’Dea et al., 2020). Including these processes in the PHY model will improve the tidal signal and the quality of the product in coastal area. NWS systems will be updated to NEMO version 4 with wetting and drying in a future version, removing the imposition of a minimum depth of 10 m. For the coupled PHY-WAV system this will also involve running WAVEWATCH III with time-varying water levels provided from the PHY model.

Figure 4: Users statistics for groups of NWS products. The number of distinct users for each class of product are plotted.

168

#57

The lateral open boundaries in the Atlantic Ocean and the Baltic Sea play a key role for shelf seas dynamics and the distribution of biogeochemical variables. R&D activities are assessing the impact of nesting the biogeochemical system into the Global Copernicus Marine Service’s biogeochemical model, instead of using relaxation toward World Ocean Atlas climatological values. In particular, changing the boundary condition will allow a better representation within reanalysis products of current biogeochemical trends in the Atlantic, especially relevant for processes like ocean acidification. In addition to updating biogeochemical variables currently used at the open ocean boundary, the nesting enables the inclusion of realistic boundary conditions for phytoplankton biomass and Chlorophyll for each phytoplankton functional group, calculated from total phytoplankton biomass and Chlorophyll, and SST. Studies have also been conducted to improve the nesting at the Baltic boundary which influences the salinity in the North Sea and Norwegian Coastal Current. Improvements were introduced in the latest version of the reanalysis, with the use of baroclinic boundary conditions instead of only barotropic conditions. Work is ongoing for improving the Baltic boundaries also in the forecast system. The river forcing plays an important role on the accuracy of the products, not limited to the immediate coastal areas. The data set for the river runoff and nutrient discharge has been improved during the NERC/DEFRA funded Shelf Sea Biogeochemistry project. The total annual discharge and mean nutrient concentration in the period 1991 to 2018, compared to the old climatology, shows significant interannual variability as well as a trend in phosphorus concentration, both of which might affect the product quality. For this reason, the decision was made to move from using the climatological forcing dataset to using the full dataset, which can more precisely represent the dynamic behaviour of the coastal areas of the domain. This change was implemented in both the latest version of the reanalysis and the forecasting system. In an ongoing contract exploring hydrological river forcing for the NWS

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

MFC, model data from Copernicus Marine Service runoff products (LAMBDA Project and European Flood Awareness System) were compared to the full and climatological runoff datasets, and their impacts on NWS models will be assessed. Multi Model Ensemble products have been developed for the forecast and the Multi Year products, following the work of Golbeck et al., (2015). The multi-model ensemble of forecast products (MME FC) is a valuable tool for users, revealing the temporal and spatial distribution of uncertainties of ocean forecast products, and it may yield the best estimate for some physical parameters. Currently, there are eight hourly forecast products contributing to the MME FC in the NWS for the following parameters: temperature, salinity, currents and transports (figures on NOOS webpage). A so-called warning system supports the near-real time evaluation of product quality by identifying forecast products drifting away from the ensemble. The MME FC is under continuous development and supports the activity comparison in overlapping regions: intercomparisons with neighbouring MFCs in overlapping areas have already been implemented in the MME, e.g., comparison of salt transport in the Skagerrak / Kattegat region (NWS and BAL MFC products). Like the MME FC, the MME of multiyear products (MME MYP) serves as a valuable tool for users, revealing the temporal and spatial distribution of uncertainties between the ensemble members. Currently there are three products providing monthly temperature and salinity fields covering the period 1993-2018 (CMEMS NWS MFC, GLO MFC, IBI MFC). This product has been available as Copernicus Marine Service internal products to all other production centrews since July 2020. The future development of the NWS system will include ensemble products. A preliminary ensemble version of the ocean-physics configuration at 7 km resolution, has been developed. This prototype ensemble system is meant to be a baseline from which to develop ensemble generation methods for the shelf-seas forecasting systems, and is certainly not in an operational-ready state yet.

169

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 5: Trajectories simulation during storm Jake (left) and storm Gertrude (right). The black line represents the drifter trajectory, the trajectory computed using the ocean currents and the Stokes drift from a non-coupled ocean and wave models are in blue, from a coupled ocean-wave is red.

2. STATUS AT THE END OF COPERNICUS 1 The NWS-MFC produces forecast and reanalysis products, delivered to the Copernicus Marine Service catalogue. Multi Model Ensemble Multi Year Products, based on NWS, Iberia-Biscay-Irish and Global reanalysis overlapping areas are also made available as internal product for all Copernicus Marine Service partners. The forecast and reanalysis for the North West-European Shelf are produced using systems based on different configurations and implementations of the ocean circulation with tides (NEMO), wave (WAVEWATCH-III) and biogeochemical (ERSEM) models. The physical and biogeochemical systems have a 3D-Var data assimilation scheme based on NEMOVAR. The wave products and the high-resolution physical forecast system have been developed and implemented during Copernicus1, and were not available in May 2015, when the service started. All the components have been subject to several updates during the last 6 years, as described in the previous section. Forecast systems are run once a day and have a forecast lead time of 6-day. The forecast is initialised by an analysis

produced by assimilating physical and biogeochemical variables. Models have 1.5 km resolution for physics and wave, and 7 km for biogeochemistry. The frequency of the products is of 15 minutes, hourly or daily means depending on the system and variables. The reanalysis is based on a 7 km coupled physicsbiogeochemistry model assimilative system and covers the period from 1993 until one year before present. The wave reanalysis timeseries has a longer time coverage, starting from 1980 and has a resolution of 1.5 km, like the forecast product. The product has a temporal resolution of 3-hour. The reanalysis timeseries are extended biannually and monthly (lower quality product) to minimize the gap from the end of the timeseries to present. The physical variables provided by the NWS-MFC forecast and reanalysis products are temperature, salinity, mixed layer depth, currents, sea surface height. The biogeochemical fields are Chlorophyll, nitrate, phosphate, oxygen, phytoplankton biomass, net primary production, light attenuation coefficient, pH and the partial pressure of CO2. The wave products consist of wave parameters integrated from the two-dimensional (frequency, direction) wave spectrum. More details are available on the Copernicus Marine Web page.

170

#57

All the products are monitored and constantly validated to guarantee the quality standard declared in the Copernicus Marine Service product Quality Information Document (QuID). The Copernicus verification has been constantly improved, with the latest addition of the SST class 2 statistics. The North West Shelf relies also on the collaboration with the North West European Shelf Operational Oceanographic System (NOOS) for intercomparison with all the observations and coastal models available in the area. NWS-MFC has contributed during these years to the development and maintenance of the NOOS Multi Model Ensemble Forecast intercomparison and alert system.

3. POST 2021 PERSPECTIVES The ambition of the NWS is : - i mproving the skills of its systems, - r efining the processes represented by models, - i mproving the data assimilation, - i ncreasing the complexity of NWS systems adding coupling effects.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Ensemble systems will allow the extension of the forecast lead time and to estimation of the uncertainty. New vertical grids, at higher resolution, will be trialled to better represent processes in the shelf and deep areas of the model domain. The horizontal resolution will be increased, where needed, in the coastal areas of the geographical domain. The quality and accuracy of products will be ensured through scientific developments to the models, the data assimilation schemes and the use of observations in the forecast and reanalysis systems. Users will benefit by having probabilistic information through the development of ensemble forecast systems. An improved design of the scheduling system used to manage model runs and data processing will guarantee efficient and robust production of a system with increased complexity (more components and dependencies). This includes the capability for rapid pull-through of R&D developments into operational product generation. Reducing the gap between R&D and operational implementation is a key factor for a fast development of the Copernicus Marine Service.

ACKNOWLEDGMENTS We would like to thank and acknowledge: Enda O’Dea (Met Office) and Nico Bruneau (NOC) for their contribution to the Shelf NEMO configurations; Stefano Ciavatta (PML) and David Ford (Met Office) for the work on the biogeochemical data assimilation; Helen Powley (PML), for providing the rivers dataset and testing new boundaries for the biogeochemical component; Rob McEwan (Met Office) for his contribution to the validation of the biogeochemical reanalysis; Ray Mahdon (Met Office) for the first reanalysis timeseries production; Juan Castillo (Met Office) for his work on the coupled version of the NEMO-WAVEWATCH-III model; and Nieves Valiente (Met Office) for the validation of the wave products.

171

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

REFERENCES: Bruciaferri, D., Tonani, M., Lewis, H., Siddorn, J., Saulter, A., Castillo, J.M., Garcia Valiente, N., Conley, D., Sykes, P., Ascione, I., McConnell, N. The impact of ocean-wave coupling on the upper ocean circulation during storm events, JGR oceans, vol26, issue6, 1-33, 2021 https:// doi.org/10.1029/2021JC017343, 2021. Butenschön, M., Clark, J., Aldridge, J. N., Allen, J. I., Artioli, Y., Blackford, J., Bruggeman, J., Cazenave, P., Ciavatta, S., Kay, S., Lessin, G., van Leeuwen, S., van der Molen, J., de Mora, L., Polimene, L., Sailley, S., Stephens, N., and Torres, R.: ERSEM 15.06: a generic model for marine biogeochemistry and the ecosystem dynamics of the lower trophic levels, Geosci. Model Dev., 9, 1293–1339, https://doi.org/10.5194/gmd-91293-2016, 2016. Crocker, R., Maksymczuk, J., Mittermaier, M., Tonani, M., and Pequignet, C.: An approach to the verification of high-resolution ocean models using spatial methods, Ocean Sci., 16, 831–845, https://doi. org/10.5194/os-16-831-2020, 2020. Golbeck, I., Li, X., Janssen, F., Brüning, T., Nielsen, J-W., Huess, V., Söderkvist, J., Büchmann, B., Siiriä, S-M., Vähä-Piikkiö, O., Hackett, B., Kristensen, N.M., Engedahl, H., Blockley, E., Sellar, A., Lagemaa, P., Ozer, J., Legrand, S., Ljungemyr, P., Axell, L., Uncertainty estimation for operational ocean forecast products—a multi-model ensemble for the North Sea and the Baltic Sea, Ocean Dynamics, 10.1007/s10236-015-0897-8, 65, 12, (1603-1631), (2015).

Graham, J. A., O’Dea, E., Holt, J., Polton, J., Hewitt, H. T., Furner, R., Guihou, K., Brereton, A., Arnold, A., Wakelin, S., Castillo Sanchez, J. M., and Mayorga Adame, C. G.: AMM15: a new high-resolution NEMO configuration for operational simulation of the European north-west shelf, Geosci. Model Dev., 11, 681-696, https://doi. org/10.5194/gmd-11-681-2018, 2018. Guihou, K., Polton, J., Harle, J., Wakelin, S., O’Dea, E., & Holt, J. (2018). Kilometric Scale Modeling of the North West European Shelf Seas: Exploring the Spatial and Temporal Variability of Internal Tides. Journal of Geophysical Research: Oceans, 123(1), 688–707. https://doi. org/10.1002/2017JC012960 King, R., While, J., Martin, M.J., Lea, D.J., Lemieux-Dudon, B, Waters, J., O’Dea, E.: Improving the initialisation of the Met Office operational shelf-seas model. Ocean Model. 130, 1-14, 2018. https://doi.org/10.1016/j. ocemod.2018.07.004 Lewis, H.W., Castillo Sanchez, J.M., Siddorn, J., King, R. R., Tonani, M., Saulter, A., Sykes, P., Pequignet, A.C., Weedon, G.P., Palmer, T., Staneva, J., and Bricheno L. 2018. Can wave coupling improve operational regional ocean forecasts for the North-West European Shelf? Ocean Science Discussion, https://doi.org/10.5194/os2018-148, in review, 2018.

Mittermaier, M., North, R., Maksymczuk, J., Pequignet, C., and Ford, D.: Using featurebased verification methods to explore the spatial and temporal characteristics of forecasts of the 2019 Chlorophyll-a bloom season over the European North-West Shelf, Ocean Sci. Discuss. [preprint], https://doi. org/10.5194/os-2020-100, in review, 2020. O’Dea E, Furner R, Wakelin S, Siddorn J, While J, Sykes P, King RR, Holt J, Hewitt H The CO5 configuration of the 7 km Atlantic Margin Model: largescale biases and sensitivity to forcing, physics options and vertical resolution GMD, 10, 2947-2969, 2017. https://doi. org/10.5194/gmd-10-2947-2017 O’Dea E., Michael J. Bell, Andrew Coward, Jason Holt, Implementation and assessment of a flux limiter based wetting and drying scheme in NEMO, Ocean Modelling, Volume 155, 2020, 101708, ISSN 1463-5003, https://doi.org/10.1016/j. ocemod.2020.101708. Skákala, J., Ford, D. A., Brewin, R. J. W., McEwan, R., Kay, S., Taylor, B. H., et al., (2018). The assimilation of phytoplankton functional types for operational forecasting in the northwest European Shelf. Journal of Geophysical Research: Oceans, 123, 5230–5247. https://doi. org/10.1029/2018JC014153 Skákala, J., Ford, D., Bruggeman, J., Hull, T., Kaiser, J., King, R. R., et al., (2021). Toward a multi-platform assimilative system for North Sea biogeochemistry. Journal of Geophysical Research: Oceans, 126, e2020JC016649. https:// doi.org/10.1029/2020JC016649

172

Tolman, H.L., 2014: User manual and system documentation of WAVEWATCH III® version 4.18. NOAA / NWS / NCEP / MMAB Technical Note 316, 282 pp + Appendices.http:// polar.ncep.noaa.gov/waves/ wavewatch/manual.v4.18.pdf Tonani, M., Sykes, P., King, R. R., McConnell, N., Péquignet, A.-C., O’Dea, E., Graham, J. A., Polton, J., and Siddorn, J.: The impact of a new high-resolution ocean model on the Met Office North-West European Shelf forecasting system, Ocean Sci., 15, 1133–1158, ps://doi. org/10.5194/os-15-1133-2019, 2019. Waters J, Lea DJ, Martin MJ, Mirouze I, Weaver A, While J. 2015. Implementing a variational data assimilation system in an operational 1/4 degree global ocean model. Quarterly Journal of the Royal Meteorological Society. 141(687):333–349. https://rmets.onlinelibrary.wiley. com/doi/abs/10.1002/qj.2388.

COPERNICUS MARINE SERVICE CENTRAL INFORMATION SYSTEM ACHIEVEMENTS GASCIARINO, G.1, TAN, T.A.2 Mercator Ocean International; 2SCALIAN

delivers two web portals (to allow new users to register, discover, view and download all available data) and a metadata catalogue for machine-to-machine interfaces.

INTRODUCTION

Being at the core of the service, the CIS main missions are to provide very high availability components and be flexible enough to adapt to user’s needs and changes in the products offer. The CIS was first implemented during the various MyOcean precursor services and had strong track records concerning its availability. Since, it has not been significantly updated during MyOcean, the main challenges for the Copernicus Marine Service were to make this historical system evolve toward state-of-the-art technologies and systems, and make it fulfill emergent needs.

The Copernicus Marine Service is an integrated system of systems. From a user perspective, it all starts at the Production Units (PU) level. They are producing, then sending data to the Dissemination Units (DU). There, registered users can download those data directly from their preferred interfaces. Unknowingly, they are using the Central Authentication Services (CAS) of an additional logical block: the Central Information System (CIS) (Figure 1). In addition to those authentication services, the CIS also

Web Portals Product Quality

Editorial

Forum

Help Center

Entry points

MyOcean Viewer

Pretty View

Monitoring Tools

Data Portal

CIS

Metadata Catalogue

LDAP

User Management Tools

CAS

Users

Authentication services

Figure 1: Simplified general architecture of the Copernicus Marine System.

173

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Programming Interfaces (APIs, red line in Figure 2) and the web user interfaces (Viewer + Joomla! in Figure 2).

1. CENTRAL INFORMATION SYSTEM OVERVIEW The historical CIS was built with as much on-the-shelf, open-source software as possible, that proved their quality over time. It has been a very effective way to deploy the foundation of a secure, maintained, efficient and robust system. Apache Solr, CAS, LDAP (Lightweight Directory Access Protocol), Geoserver, Geonetwork and Joomla! are still popular solutions today but they require tailoring some parts of the system to completely fulfill Copernicus Marine Service’s needs. Two areas demand a particular attention: the technical interfaces with DU Application

Machine-to-machine interfaces do not evolve quickly but they require code adaptation to match their performance constraints. For instance, if the DU API cannot provide more than 1 GB of data per request, the web interface should not let users request more than 1 GB of data. When the Copernicus Marine Service Phase 2 DUs came into operation, it provided a big improvement in performance at various levels and it provided new ways of delivering data. On the other hand, the web User Interface (UI) should be updated more often. Indeed, as industry standards are changing rapidly, users’ first impression may be to stumble upon an old website. Initially, this web development activity was not considered as critical as it is today and has changed significantly during the past 6 years.

Figure 2: Simplified general architecture of the CIS in 2015.

174

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- DUs availability. As users are interacting with the Data Portal, they experience indirectly the outages at DU level. For instance, DU 4 (Figure 5) knew an overload episode due to the release of a new product and was not able to recover to Marine Service standards for a few months.

2. MAIN ACHIEVEMENTS FROM 2015 TO 2021 To put things into perspective, some general figures showing how much the service has grown from 2015 to 2021 are provided: - subscribers are people registered to our service. On the other hand, users are subscribers that are actively using the service (downloading data). In 2015, Copernicus Marine Service had approximately 4 000 subscribers. It now has 30 000 (Figure 3), - evolution of downloaded data volume (Figure 4) skyrocketed starting from 2018. This shows that with DUs performance enhancement, usage of the service has changed. Therefore, it had to be reflected in the Data Portal,

High availability and scalability The first main mission of the CIS has been accomplished: the availability of the services has been kept high during all the period and has been improving with time. The 97% availability target has always been fulfilled for the authentication services, and fulfilled with only 2 exceptions for the 2 other services. Behind the scenes, redundancy and automation have been key to having a reactive system that prevents most of the downtime with minimum human intervention.

Figure 3: Evolution of users over time since MyOcean 1.

Figure 4: Evolution of downloaded data volume over time since MyOcean 2.

175

#57

Figure 5: Evolution of global DU availabilities over time since 2015.

Figure 6: Availability of CIS systems (in %).

176

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

#57

3. A CONSTANTLY EVOLVING SYSTEM 3.1 Web evolutions To follow user and service needs, all components have evolved several times during the 2015-2021 period. Here are some of the main evolutions done to streamline the user experience:

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

- 25 major catalogue updates, - Data Portal registration form has changed several times. The registration and reset password workflows have been changed from manual to automatic, - Data Portal search page has been changed several times, also to adapt to the ever-growing list and kinds of products (from 161 in 2015 to 326 in 2021), - Data Portal is directly connected via chat and/or mail with the Service Desk, - Data Portal has changed its user interface, - Improved metadata description, - One full rewrite of the Data Portal and the Viewer.

Figure 7: Evolution of the website. From top left to bottom right: 2015, 2018, 2020, 2021.

177

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

3.2 Service Management

3.3 Central Dissemination Unit

The number of changes and incidents to cope with has also drastically increased starting from 2018. Tried and tested procedures in place proved to be effective and did not affect man power requirements. This can be explained by 2 factors: -4 0% due to the increase in number of contractors (more products, bigger offer) and separation between PU and DU (the interfaces issues are now more visible), -6 0% due to the growing user base that increased the stress on the system, and new requirements that emerged from that.

Finally, the last major change in the system was the fact that all DUs have been regrouped into one central DU in 2018. All Production Units and the CIS had to rewire their connections and revisit their procedures.

Figure 8: Number of changes and incidents affecting the CIS per month.

Figure 9: Evolution of PUs and DUs. Left: 2015, Right: 2021.

178

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

As described in the previous chapter, the CIS proved to be a robust system. This aspect of the system must be kept for the future of the Copernicus Marine Service.

are now outdated (if still supported). In 2019, it was been decided that the technical debt was too high to be managed and that a full rewrite of the Data Portal and the Viewer must take place. In the same spirit, the authentication services could benefit from some additions that are now more widely used than the one used in the Copernicus Marine Service (OpenID or OAuth instead of CAS).

The main challenge has been to implement changes over the course of time. All those evolutions came at a price: the Data Portal and the Viewer were built using 2008 technologies that

Globally, the whole CIS architecture is still the same in 2021 (Figure 10), but many pieces have been removed or absorbed by modern technology capabilities.

4. STATUS AT THE END OF COPERNICUS 1

Figure 10: Simplified general architecture of the CIS in 2021.

ACKNOWLEDGEMENTS

We want to thank the whole Copernicus Marine Service CIS Team for the effort and professionalism showed during the whole lifetime of the Service that allowed the Service Provisioning during all those years, and people who does not collaborate with the CIS anymore: Rémi De Dianous1, Kate Dondenaz1, Sophie Besnard1, Tarek Habib1, Marie-Hélène De Launay1, Bruno Pirrotta1, Morgan Billon1, Siegfried Castanet1, Cathy Schgounn1, Sebastien Lebossé1, Pascal Mambert1, Benjamin Lacaze1, Cyril Bazin1, Tony Jolibois1, Sylvain Marty1, Jérôme Doumerc1, Françoise Mertz1, Sam Guenoun1, Julien Fontanel1, Karen Cordier1, Monique Gasc1, Caroline Maheu1, Joan Sala2, Alex Lopez2, Isabel Polo2, Laia Romero2, Thi-Anh Tan3, Benjamin Guerin3, Jeremy Debras3, Rémi Lebrouster3, Thomas Lebeau3, Anne Delamarche4, Michèle Fabardines4, Gabriel Gasciarino4, Jean-Lou Habemont5, Jérome Garcia5, Catherine Satra Le Bris6, Julien Meillon6, Erwann Quimbert6, Mickael Treguer6, Thomas Loubrieu6 1

CLS, 2Altamira Information; 3Scalian; 4Mercator Ocean International; 5ATOS; 6IFREMER

179

COPERNICUS MARINE SERVICE DISSEMINATION UNIT

FORNERIS, V.1, CESARINI, C.1, BONOFIGLIO, L.2, COLOMBO, F.2, NOVELLINO, A.2 CNR, Italy; 2ETT SpA, Italy

INTRODUCTION Dissemination Units (DU) are important components of the Copernicus Marine Environment Monitoring Service (CMEMS), being in charge of the dissemination of products elaborated by Production Centres (PC; MFCs and TACs), making data available to users by means of several interfaces such as: File Transfer Protocol (FTP), Web Map Service (WMS), Subsetter (MOTU) and Direct Get File (DGF). During Copernicus Marine Service Phase 1, there were multiple DUs, each part from different PCs (Figure 1). While providing more flexibility toward producer’s data (DU adapted for providing specific data), the overall Service was unbalanced as some DUs proved to be less stable than

others. Moreover, having multiple/different access points, above all for operative/automatic/m2m (machine-tomachine) procedures, increased the overall complexity for users accessing different portions of Copernicus Marine Service Catalogue. At the beginning of Phase 2, Copernicus Marine Service improved its service by changing this distributed approach in favour of a consolidated approach. The new DUs (one per Near Real Time/Analysis data, one per Multiyear data) main missions were (and are still) to: - set up & maintain infrastructures for integration and operations, - collect products from Production Units (PUs) & disseminate them to users, and - ensure & monitor operations related to the collection & dissemination.

Figure 1: Copernicus Marine Service phase1 DU(s) architecture.

180

#57

To be noted, the introduction of a new centralized DU has been one of the major technical challenges for both the new DU’s team and producers, as it implied changes in consolidated procedures and the introduction of new components such as (but not limited to) the Delivery Buffer Service (DBS).

1. COPERNICUS MARINE SERVICE PHASE 2 DU - OVERVIEW In order to meet the high level of service required by new requirements, in 2018 a new Consortium has been formed, led by the Consiglio Nazionale delle Ricerche (CNR - prime Contact Point) and composed by: - CNR (with the support of Collecte Localisation Satellites SA (CLS), Serco Italia SpA, and Laser Romae s.r.l), - ETT S.p.A, -A CRI ST (with the support of adwaisEO), - I nstitut Français de Recherche pour l’Exploitation de la Mer (IFREMER; with the support of Altran).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

This Consortium has a strong and long experience in leading International Project and Services, IT&ICT solutions in operative environment, software development for both producing, archiving and disseminating Earth-Observation data, including Copernicus Services. The Copernicus Marine Service DU provides an IT/ dissemination expert centre service to the organisation and provides operational links between the other Service’s elements and users. In fact, the DU collects products from production units and disseminates them to users, monitoring operations for both near real time and forecast (NRT) and multi-year (MY) products. This organization pursues the final goal of filling the gap between science and technology. The DU provides the Copernicus service with continuity, and consolidates contact points, resources and access points. It provides flexible, robust and scalable DU systems, improving the operational level of the system and setting up Key Performance Indicators’ (KPIs) dashboards. It resulted in a new DU Architecture, as shown in Figure 2. This article reports the results achieved during the period 2018-2021.

Figure 2: CMEMS phase2 DU(s) initial design/architecture.

181

#57

2. MAIN ACHIEVEMENTS & EVOLUTIONS (2018-2021)

The Consortium adopted a modular Service-Oriented Approach (SOA) in order to match requirements with evolutions. Due to the tight operative schedule, the complexity of the system and the lack of formal standards for some important items involved (e.g., data), the initial setup could neither take all needed details into account nor focus on components automation. In order to mitigate these weaknesses, systems are organised into components with well-defined interfaces. Since the very first release (Q1 2018), and at each following one (at least 3 per year), the DU team adopted continuous review and evolution cycle strategies to cope with the constant increase of product number, data quality and user transactions. Indeed, these changes induced some challenges in terms of compatibility or resource allocation. The DU evolution cycle fills the gap between stakeholder’s needs and expectations and DU’s operation. The framework behind the DU evolution cycle is represented by the following workflow: planning, development and testing, integration, maintenance, operative assessment and analysis, requirements analysis (see Figure 3).

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Based on this framework, in order to improve the Service on different aspects, the DU team has been constantly reviewing and updating the architecture design and internal procedures, focusing on: - infrastructure, - software, - collaboration with CMEMS Top Level, production centres, software developers, - consortium composition.

2.1 Infrastructure The DU team has been continually reviewing and updating the architecture design. The Consortium invested time and resources in improving the backend (connectivity, VMs numbers and resources, disk type, etc) and redistributing (re-balancing) products amongst virtual machines (VMs). To be noted, the cloud provider is an external, independent component; the DU team collaborated with them in order to identify and improve possible low-level bottlenecks. Moreover, the Copernicus Marine Service DU improved the Development Systems, used for debugging activities and facilitating the integration of new products and datasets, added a “PU Play Ground” System (PU’s testbed for DBS access and new Releases upload process) and several different Monitoring Systems and dashboards according to MOi feedback and internal needs.

Figure 3: CMEMS phase2 DU(s) evolution cycle.

182

#57

The Delivery Buffer Service (DBS), the most “delicate” PUDU interface, was revisited multiple times, mainly during the first 2 years, according to MOi and PU feedback for enhancing its functionalities (e.g., allowing PUs to move and delete files and folders, improving error detection, and m2m approach, etc), its robustness and performance. Effort was also directed to add and improve a disaster recovery (DR; backup) system, deployed in two steps: - introduction of the DR System in async mode (nightly sync); active from end of 2018 to end 2020, - DR improvements (sync mode): data is backed up as soon as ingested by the DU; active since early 2021. As result, the total number of systems increased from the initial 58 to the actual 93 VMs, for a total (MY+NRT) of 636 Cores and 2.1 Pb RAM (roughly +75% resources).

2.2 Software Apart from the standard/classic FTP interface, Copernicus Marine Service provides advanced Interfaces for accessing data via standard OGC protocols. These interfaces are provided by specific software, namely THREDDS (e.g., WMS) and MOTU (aka Subsetter; it provides access to the DGF interface too). These software are actually Commercial

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

off-the-shelf (COTS) and the DU team has no (THREDDS) or very limited (MOTU) control over them. Copernicus Marine Service DU has however constantly upgraded these software for bug fixing, or for introducing new features/ enhancements, particularly on MOTU component (five releases implemented since the beginning). Currently, running versions are: THREDDSv4.6.14 and MOTUv3.12. New software/Interfaces have been studied and proposed. The only one accepted, due to priority constraints, was ERDDAP Data Server (for INSITU data only) in substitution of Oceanotron due to its end of life/maintenance. Moreover, the DU team continuously adapted and improved software configuration according to users, PCs and Mercator Ocean international (MOi) feedback/needs, leading to, e.g., the enabling of CORS, and SSL over HTTP, improvement of WMS experience, introduction of new data type (e.g., shapefiles). Under software category fall also the monitoring systems. The DU team spent lot of time introducing and improving monitoring systems for internal purposes (e.g., Systems availability, Interface accessibility, etc), and for Service improvements. Concerning the latter, two of these systems are particularly important within the Service (Copernicus Marine Service internal tools).

Figure 4: SySMA.

183

#57

Centralized Product delivery Timeliness tool (SySMA) : the DU set up several monitoring tools on the DBS in order to monitor and record delivery achievements since the first implementation: SYSMA generates a long-term data warehouse for parameters monitored by each DU service. Moreover, it monitors different applications of the DU service, by means of specific parameters, and reports the information using data stored in real-time and long-term databases and presenting data via clear dashboards. The DU team continuously revisited and improved dashboards according to MOi feedback/needs.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Centralized Transaction Monitoring System: in 2021, DU team developed a new Centralized Monitoring System for collecting and presenting via dashboard statistics about data access from the users (transactions). It provides an easier method to query and extract statistics about data downloaded via different interfaces, and from different backend. Its original scope may be enhanced further, by providing detailed info about how users access data (e.g., which variables) and possibly helping the PUs in reorganizing and optimize the data itself.

Figure 5: Transaction Monitoring

2.3 Collaboration with Copernicus Marine Service Top Level, Production Centres, software developers The lack of strong standardizations, particularly on data production, posed several challenges at the beginning of the Centralized Service. In order to improve the Service Provisioning as requested, over time the DU team increased collaboration with other Copernicus Marine Service Entities, with particular reference to: - top Level: DU team collaborated with Copernicus Marine Service Top Level in the definition of common (internal: DU-MOi, DU-PU) procedure and approaches (e.g., NRT Double Dissemination, MY timeseries extension, etc), - production Centres: DU team promoted and coordinated a Technical Working Group (TGW) involving DU, MOi and PC representatives aiming to share best practices in DU-PU specific procedures,

data production (for increasing the compatibility with advanced software) and generally providing strong feedback on possible standardizations (such as possible future dataset nomenclature), - software (MOTU) Developers: even if software development was not part of the Agreement, the DU team increased the collaboration with MOTU Developers (CLS) in order to increase its expertise in the specific bug tracking and provide feedback for possible enhancements.

2.4 Consortium composition DU team reviewed constantly also its Consortium composition: sub contracts have been reinforced, staff has grown with new roles added and redefined according to the needs, for both operational or development purposes.

184

#57

3. STATUS AT THE END OF COPERNICUS 1 Despite the technical challenges and the high Service Level Agreements, the DU achieved significant results during the reporting period. The Virtual Infrastructure Availability has been always very stable (see Figure 6). On the other hand,

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

external interface availability has been the real challenge, as depending on the integration between Software and the data itself, and the actual usage. DU Team spent most of the time and evolutions (planned and unplanned) on improving this aspect, via different methods as described in previous sections (improvements on backend, software, collaboration with PCs, etc) resulting in a constant improvement of the robustness and general availability of the Service, as shown in Figure 7 (NRT) and Figure 8 (MY).

Figure 6: CMEMS DU Virtual Infrastructure availability.

Figure 7: CMEMS DU External Interfaces availability - NRT.

Figure 8: CMEMS DU External Interfaces availability - MY.

185

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

4.2 Software

4. POST 2021 PERSPECTIVES It is not easy for the DU Consortium to precisely define evolutions beyond 2021, as the Statement of Work for the Dissemination Unit provides usually very strict requirements and precise guidelines on (e.g.,) software and technical solutions to be adopted. The experience gained during Copernicus Marine Service Phase 2 however, helped in identifying possible improvements that can be grouped in four main activities: backend technologies and architecture, software, service management and introduction of stronger formalization on data production.

4.1 Backend technologies and architecture At the time of the Phase 2 proposal, no information was given about the cloud provider, making impossible to design an architecture tailored on specific features and leading to a generic design. This design proved to be flexible enough to fulfil the needs of the Service, but at the same time showed some limitations that must be overcome considering the next evolutions of the Service in terms of data provided and users served. In particular, the IO layer is the most important bottleneck and efforts should be directed to improving this aspect. Actual “block storage” approach should be abandoned in favour of “object storage” solutions, which grant better scalability, faster data retrieval possibly an optimization in terms of resources. Moreover, compatibly with the software needed for advanced interfaces and cloud provider features, more automated design should be implemented (e.g., backend auto scalability).

Two main software provides access to advanced interfaces: THREDDS and MOTU. Experience showed their specific limitations (e.g., THREDDS is not scalable natively) and it is advised to evaluate different solutions.

4.3 Service management The Service management is the core activity of the DU Consortium. With the quick increment of the number of backend and transactions, the DU needs more automatic solutions for deploying and connecting resources upon requests/needs (auto scalability), improving auto recovery procedures (in case of faults) and increasing the centralization of services for a more congruent Service management.

4.4 Data production formalization A weakness identified since the beginning of the Contract, that affected DU design and functionalities, is the lack of strong formalization on data production. While not part of the DU mission, it is strongly suggested to carry on the work undertaken by the TWG in order to define a stricter Copernicus Marine Service Data Specification (CDS), taking into consideration software solutions that will be adopted. It would help the DU to increase automatization (e.g., on data validation before the ingestion) but also producers above all during the design of new products and users, providing a more congruent catalogue.

ACKNOWLEDGMENTS The “Dissemination Unit” has been a long, interesting but challenging journey. Authors want to thank the whole CMEMS DU Team for the effort and professionalism showed during the whole lifetime of the Service that allowed the service provisioning even during the COVID-19 pandemic crisis, including administrative colleagues, and people who does not collaborate with DU anymore: Rosalia Santoleri1, Flavio La Padula1, Emma Dacunzo1, Cristina Tronconi1, Edoardo Seno1, Massimiliano Zanghi2, Marco Alba2, Paolo D’Angelo2, Michela Busi2, Giuseppe Manzella2, Alessia Scipioni2, Antoine Azemar3, Gilbert Barrot3, Stephane Clouaire3, Helene Le Fur3, David Ros3, Edouard Hubner3, Antoine Troullier3, Thierry Carval4, Cécile Salaun4, Jérôme Detoc4, Corentin Guyot4, Yann-Etienne Prigent4, Antoine Queric4, Alessandra Paciucci5, Andrea Tesseri5, Simone Garofalo5, Stanislaw Rogoza5, Tullio Crisafulli5, Barbara Borgia5, Leonardo Calamai5, Simone Integlia5, Marco Mastrecchia5, Giovanni Corato6, Dario Torregrossa6, Frederic Twahirwa6, Chrystele Besson6, Gaston Briot6, Damir Kero6, Lukasz Grandys6, Syed Mehdi6, Andres Salgado6, Tarek Habib7, Sylvain Marty7, Jerome Doumerc7, Gwenaël Renard8, Mazurier Alain8, Fabio D’Andria9, Emiliano Fedeli9 CNR; 2ETT SpA; 3ACRI ST; 4IFREMER; 5SERCO Italia SpA; 6AdwäisEO; 7CLS; 8Altran; 9Laser Romae s.r.l.

186

MARINE SERVICE MONITORING AND USER FEEDBACK

DELAMARCHE, A., DERVAL, C., GIORDAN, C., OBATON, D., CROSNIER, L. Mercator Ocean international (MOi), France

OVERVIEW

1. PRODUCT AND SERVICE MONITORING

The Copernicus Marine Service must deliver and maintain a competitive, state-of-the-art and operational European service responding to all users. It is monitored through key performance indicators (KPIs), quarterly and annually reported. Such KPIs assess its reliability against operational commitments, service level agreement (timeliness, robustness, etc.). The service monitoring activity encompasses many KPIs to steer the service and its uptake, and for example provides with figures about the product portfolio evolution, the evolution of the number of subscribers and their detailed characteristics, and the monitoring of the service availability and product timeliness.

The service monitoring activity encompasses many KPIs to steer the service, its portfolio, its operationality and its uptake. For example, it provides with figures about the product portfolio evolution, the evolution of the number of subscribers and their detailed characteristics, and the monitoring of the service availability and product timeliness.

The Copernicus Marine Service also needs to maintain a permanent dialogue with users while collecting their requirements and supporting them in the use of the service through user support and training services (Giordan et al., this issue). The present paper showcases elements of the Service monitoring and user requirement activities, and provides prospects for the next 2021-2027 Copernicus phase.

1.1. Product Management Since 2015, Mercator Ocean international (MOi) has managed system and service evolution through a formal development process (specification, design, acceptance, entry into service) with a review process at the end of each development phase. This has led to regular catalogue releases: 25 catalogue releases took place from 2015 to 2021. Between 2015-2020, the catalogue has been enlarged from 134 to 180 ocean products (Figure 1, left panel), and the number of active users downloading these ocean products every year has also been increasing (from 2 000 to 9 000 active users from 2015 to 2020 respectively) (Figure 1, right panel):

Figure 1: Evolution of products and active users.

187

#57

Figure 1 (left panel) shows that Mercator Ocean has been carefully and closely managing its product portfolio and each product life cycle. There are 3 catalogue updates per year on average, introducing new products, improving existing ones and removing old ones. Along the years, the service has, for example, been enriched with new wave products, new satellite (Sentinel-3, Jason3) data products, Ocean monitoring Indicators (OMI), Ocean State Report (OSR), catalogue homogenization (CF standard names usage), use cases, demonstrations, trainings and roadmap.

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Product management has also allowed to carefully track all product change impacting users along with product metadata updates. The service has been making sure that all changes are carefully communicated to users. Updates and improvements of the catalogue have been always announced in advance on the Copernicus Marine Service Web Portal “User Notification Service”, in the dedicated section (Improvement, Figure 2).

Figure 2: Improvement announcement.

1.2. Service Monitoring KPIs Timeliness and Availability are measured and monitored for each product in the catalogue. Timeliness indicator measures the difference (delay) between the effective product delivery time and the target delivery time.

Between 2015-2020, these 2 KPIs have been stable and stayed high: 98% for timeliness, 99% for availability. The usage of products has also been monitored and showed that the number of users downloading products has increased over 2015-2020 (Figure 1) along with the volume of data downloaded (from 167 TB in 2015 to 3000 TB in 2020) (Figure 3).

Availability indicator (factor of reliability) provides real time information about the presence and/or accessibility of a product.

188

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Evolution of the volume downloaded.

2.1 Process

2. USER REQUIREMENT AND ANALYSIS PROCESS Over the period 2015-2021, MOi has strengthened the feedback loop to manage user expectations: it has collected feedback from all channels (champion user advisory groups, service desk, training, events…) and organized surveys and market specific workshops to understand and collect feedback. MOi has built and maintained an operational User Requirements Database, gathering all feedback. Such feedback has fed the specification and design reviews. Data providers have adjusted the fitness-for-purpose of their products according to user feedback centralized and provided by Mercator Ocean.

User feedback and user satisfaction are measured, monitored, analyzed and injected back into the service through the implementation of new or updated products and services to better fit users’ demand. User satisfaction has been high and steady all along the years (4.7/5). User feedback is managed following this methodology: - collect, - record, - analyse, - identify new user requirements, - improve.

189

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: User Feedback infographic.

Whichever the entry point of the feedback collected (Service Desk, User Workshops, Questionnaires, Meetings, Champion User Advisory Group), the feedback is recorded within a table called “record of user feedback”. The Champion User Advisory Group (CUAG) is one of the channels to collect feedback and has advised MOi on user requirements and service evolution solutions implemented to satisfy user needs. The table “record of user feedback” for example contains:

190

- columns about user information including detailed feedback and date of feedback/request, - after analysis, this table is updated with further information about how this item will be considered in the next service evolution: which category this feedback belongs to? (Figure 5); the status of the feedback (open or close with explanation).

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 5: User Feedback thematic.

Every feedback item collected over the period is considered, aggregated and inter-compared and give rise to requirements with different weights.

- provide an OPeNDAP (a protocol to get data remotely) access: this service will be available at the end of 2021, - allow the use of various softwares (e.g., matlab, GIS): This will be possible thanks to OPeNDAP access.

Higher weight is given to: - s ame request from several users, - request from a regular user (user that downloads products at least three times in different weeks within the period of one month), - r equest with common sense.

2.2. Main user requirements The main requirements identified over the years are: - provide higher spatial and temporal resolution for products: this requirement is continuously considered. Resolution is a key priority and it has globally increased over the years following scientific and technical state-of-the-art capabilities, - add wave products: Near real time and reanalysis model wave products for each ocean basin have been continuously added since 2017. Satellite and in situ wave products are also distributed since 2019, - add model wind and atmospheric products: this requirement cannot be taken into account (yet) as it depends on the data policy from Numerical Weather Prediction centres, - ease the choice of, and access to, products for nonexperienced users: several updates were made, but there are still some improvements to be done. It is foreseen to provide an advanced catalogue search and catalogues by thematic applications (as Marine Strategy Framework Directive, Sea Ice services), - download more easily very large datasets: this requirement is not yet fully taken care of and is part of an ongoing effort,

3. CONCLUSIONS AND FUTURE PROSPECTS During the 2015-2021 Copernicus period, service monitoring and user requirement activities have allowed to: - carefully monitor the Copernicus Marine Service, and - steer its evolution and user uptake. Within the next 2021-2027 phase, those two activities will be reinforced and relationships with policy and major accounts will be strengthened to transform them into key strategic accounts. MOi will manage and contribute to: - develop a dedicated offer catalogue for each of the policy and major accounts, - ensure that the products and services developed and operated are fully aligned with the activities and plans of the European Union. Moreover, an independent operational User Requirements Database will be built and maintained and an external component will be set up to evaluate independently the efficiency and overall performance of the service. Performance will be assessed mainly toward users’ information needs and high-level requirements and the fitness-for-purpose of the Copernicus Marine Service with a specific emphasis on the needs of EU Policies.

191

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

COPERNICUS MARINE USER SUPPORT AND LEARNING SERVICES GIORDAN, C., MESSAL, F., BAZIN, D., LEGROS, V. Mercator Ocean international (MOi), France

1.1 Training activities

OVERVIEW User learning and user support services both strengthen user uptake: objectives are to train, answer questions, facilitate user experience, share knowledge and collect requirements. On the one hand, training activities foster products’ use and feedback collection to improve the service. On the other hand, the user support desk is the point of contact for all questions and comments from users, and its objective is to optimise their experience throughout the Copernicus Marine Service. This paper describes what has been done during the period [20152021] for those 2 activities and provides figures qualifying them.

1. USER LEARNING SERVICE The Copernicus Marine Service is committed to assist and to support new and potential users by offering training activities and providing access to learning materials. These 2 main pillars of the learning services encourage product use and demonstrate the full potential of the Copernicus Marine Service (data access, difference between products, use case presentation, etc.). In addition to activities dedicated to the Copernicus Marine Service, the training team and experts were also involved in the organization of several training workshops and demos for WEkEO, the Copernicus DIAS platform, where the Copernicus Marine Service data is also distributed.

As mentioned before, training activities foster product use and feedback collection to improve the service. The design of all training activities has been guided by considering the EU member States’ needs and the willingness to collaborate with local experts and the Copernicus Stakeholders. Training workshops are designed to train existing and new Marine Service users. The target audience is mainly beginners and new users. Participants learn about our products and services, and their possible applications across a wide range of subjects during plenary and practical training sessions. Participants are also able to share their experiences as well as express needs and requirements for future products to be included in the Copernicus Marine catalogue. These events consist of a mixture of plenary expert talks, round table discussions, and hands-on practical exercises to explore the Copernicus Marine product catalogue. During the period 2015-2021, the Copernicus Marine Service has organized several training workshops or webinars each year. During the period 2015-2019, a total of 15 Members States have benefitted from local training workshops. This series of face-to-face workshops (Figure 1, left) have gathered an average of 300 trainees per year. From 2019 onward, the Copernicus Marine Service has intensified its training activities and organized 6 regional training workshops per year (corresponding to the 6 different European basins) with the help of local marketing and scientific partners. These workshops were face-toface format with a limited number of participants. In 2019, the average satisfaction rating of the participants was 4,2/5.

192

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 1: : D. Bazin, from the Service Desk, during a training session (left); evolution of the number of participants during the training workshops (right).

In 2020, due to the context imposed by the COVID pandemic, workshops were offered in a full online version, with live webinars and a period for practical exercises. The audience has drastically increased in 2020 (x5) with the online format rather than face-to-face (Figure 1, right). Without limitation of registrations, these series were attended by over 1450 participants from more than 50

different countries (Figure 2), and 92% of the attendees surveyed were either satisfied or very satisfied. The amount of training material produced for these online workshops is quite large: for 2020 only, along with scientific experts from the different TACs & MFCs, the service desk and technical partners have produced more than 70 Jupyter Notebooks and 80 tutorial videos (Figure 3).

Figure 2: Number of participants and spatial repartition of the 2020 training workshops.

193

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 3: Recording of the 2020 training workshop for the Baltic Sea - Tutorial videos.

All these videos are hosted on the Copernicus Marine Service Youtube channel which has over 77000 view (Figure 4).

1.2 Learning material Tutorial videos and Jupyter Notebooks are the main material developed for the User Learning activities, which are shared on the website e-learning material. Year after year, the video library has expanded. Users now have access to: - tutorials providing basic information about how get started with the service, - r ecordings from diverse training events or webinars, - t utorials related to specific training material like GIS tool or Jupyter Notebooks.

A Jupyter Notebook is an open-source web application for experts to create practical exercises and share codes. It can contain narrative text, equations and images. Beginner users are very interested in this kind of material as they can find useful pieces of code to develop their own programming codes and understand how to use products. Since 2020, a JupyterLab (the required environment to run Jupyter Notebooks) is operational and its access is only granted to participants of training workshops.

194

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 4: Example of a Jupyter Notebook and the associated tutorial video.

1.3 WEkEO The training activities for WEkEO started in 2020. The successful online format (live webinar + training material, Figure 5) which have been tested during the Copernicus

Marine Service workshops was applied for WEkEO. Demo sessions and 4 long webinars (respectively dedicated to marine, atmosphere, climate and land data) were organised gathering over 800 participants in 2020.

Figure 5: WEkEO training timeline inspired by the Copernicus Marine Service training timeline.

195

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Service Desk also provides an internal link between users and production centres, scientists and technical experts. The Service Desk is also deeply involved in the training activity described above and participates in all such events.

2. USER SUPPORT 2.1 Context and missions

2.2 Users and user behaviour The Service Desk is the point of contact for all questions and comments for users of the Copernicus Marine Service. Its objective is to optimise their experience and its missions are described in Figure 6. Various means are available to initiate or conduct these exchanges: - a Chat, - an e-mail address, - online forms, - telephone, - video-conferencing. The Service Desk is also responsible for informing users of operational issues on products and services, such as incidents, maintenance and improvements. Moreover, the

Since 2015, the number of registered users has steadily increased, reaching over 32,500 in 2021. Currently, 600 new users per month are signing up for the Copernicus Marine Service. The number of active users (i.e., users who have downloaded data at least once) follows the same pattern (Figure 7). These Users are spread across all continents, in over 120 countries (Figure 7). Moreover, the volume of data downloaded per year has increased over the years, such as the number of downloads per year (Figure 8).

Figure 6: Service Desk missions.

196

#57

Figure 7: Left : Evolution in number of users and subscribers. Right : Geographical distribution.

Figure 8: Downloads and Volume - Monthly means.

197

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

#57

2.3 User Satisfaction After each exchange with the Service Desk, a short form is shared with the user. This form offers the user the opportunity to rate their satisfaction by giving a score out of 5, and opens a free text area for them to share their feedback and needs (Delamarche et al., this issue). Since the implementation of this process in August 2016, and with more than 1650 responses as of now, the average rating is

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

4,7/5. In case of a poor assessment, or a comment in the free text area, the Service Desk will get back to the user for more information. The graph below (Figure 9) shows the detail of the evaluations, from 1/5 (Extremely unsatisfied) to 5/5 (Extremely satisfied): The Service Desk is also evaluated in annual questionnaires, among other items. All this feedback, including satisfaction inquiries and questionnaires, are collected and analysed afterwards by a dedicated team.

Figure 9: Satisfaction Inquiry results.

198

#57

2.4 Tools and services 2.4.1 Chat: A Chat is available on each page of the website, facilitating exchanges with users. This solution was requested by users and was implemented at the end of 2020. The evaluation of user satisfaction is also done by this means, when a conversation is closed. 2.4.2 User Corner: Since January 2021, the User Corner has been expanded. It now includes 13 sections presenting services and useful links to different pages of the website: - Getting Started: useful links and information for new users, - FAQ: general questions and answers about the Copernicus Marine Service, - User Notification Service: be informed on operational issues on products and services, - Help Center: collections of technical articles to master Copernicus Marine Service’s data, - Product Roadmap: timeline of upcoming service developments and improvements, - User Learning Services: e-Learning resources, - Product Quality: product quality information updates, - Collaborative forum: join the Copernicus Marine Community, - Login/Register: create a free account, - Get Inspired: use cases across the 12 Blue Markets, - Contact Us: contact the User Support, - Service Commitments and Licence: range and level of services that the Copernicus Marine Service will supply to the user, - Product Catalogue: printable catalogue in PDF format. 2.4.3 Getting Started page: The Getting Started page was created at the end of 2018 to guide new Users through their first steps on the Copernicus Service. After a summary of the offer, several useful links are presented to find out everything to know to get operational. 2.4.4 Help Center: With the arrival of the Chat, a new Help

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Center has been made available, replacing the technical FAQs that had existed since 2015, and which now deal with more general questions. This intuitive interface is divided into different themes, known as collections, and brings together more than 100 articles facilitating the use and handling of Copernicus Marine Service data. Below are some examples of the collections available: - getting started with Copernicus Marine Service, - Copernicus Marine Database - Pan European Platform, and - Copernicus Marine Data Visualisation and Processing. At the end of each article, the user can evaluate it, and the Service Desk can then make it more fit to users’ expectations. 2.4.5 User Notification Service : The User Notification Service section of the User Corner gathers all operational issues impacting the Copernicus Marine Service products or services. Issues are displayed as notifications, and sorted under 4 categories (Figure 10): General, Incidents, Maintenance and Improvements: A customisable RSS feed is available to build it according to specific criteria (products, regions, settings, etc.). A dedicated Twitter account is also available to follow all notifications.

3. CONCLUSION AND NEXT STEPS The user learning and user support services have evolved continuously throughout the period 2015-2021, and are highly rated by the users (respectively 4.2 and 4.7/5). They are key to foster the user uptake. The next Copernicus phase (2021-2027) will allow the strengthening of these 2 activities.

Figure 10: User Notification Service.

199

COPERNICUS MARINE SERVICE USER ENGAGEMENT AND MARKET DEVELOPMENT ABADIE, V., THOMAS-COURCOUX, C., LEGROS, V., LABROUSSE, C., QUADE, G., BASTIDE, L., CROSNIER, L. Mercator Ocean international (MOi), France

OVERVIEW The objective of the user engagement and market development activities is to foster uptake of the Copernicus Marine Service portfolio, develop market intelligence and seek new opportunities for data use in new communities. Such activities encompass the definition of the Blue Markets to be targeted: -e xplaining the Copernicus Marine offer to new communities, - showcasing the use of data through use cases and the User Uptake program (Durand et al., this issue), - l aunching marketing campaigns, - organizing or participating to events showcasing the Copernicus Marine Service, - c onnecting with new partners and new communities. This paper provides with an overview of the user engagement and market development activities than have been developed during the 2015-2021 phase and provide with perspectives of what will be continued in the next 2021-2027 phase.

1. DEFINITION OF THE BLUE MARKETS The Copernicus Marine Service supports Blue Markets by providing data, information, and services across the three pillars of the United Nations Sustainable Development: Environment, Society and Economy.

Mercator Ocean International (MOi) has first identified and defined the Blue Markets and then created web pages providing data, services, and information for each of the 10 identified Blue Markets: - Polar Environment Monitoring, - Policies & Ocean Governance & Mitigation, - Climate & Adaptation, - Natural Resources & Energy, - Ocean Health, - Coastal Services, - Education, Public Health & Recreation, - Marine Food, -T rade & Marine Navigation, - Extremes, Hazards & Safety. Contents (factsheet, videos, infographics, etc.) of each dedicated page explain in a didactic way, how Copernicus Marine data can be used and which variables are relevant for the specific Blue Market. Numerous inspiring examples (use cases) are provided depicting how other organisations have already benefited from our marine data. In 2020, MOi has updated the 10 blue markets to 12 (Figure 1) to cover all user needs and markets, and for a better segmentation of the Copernicus Marine offer targeting different users. The two new markets are: - Marine Conservation & Biodiversity: the Copernicus Marine Service provides key data to monitor marine biodiversity and to protect Marine Protected Areas, preserving at-risk ecosystems, - Science & Innovation: this new market targets scientists and businesses illustrating how Copernicus Marine data can innovate in their practices and build new applications and services for intermediate and final users.

200

Environment 4

Society

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Polar Environment Monitoring

Climate & Adaptation

Ocean Health

Marine Conservation & Biodiversity

Science & Innovation

Policies & Ocean Governance & Mitigation

Copernicus Marine Service Segments and markets

Economy

The Copernicus Marine service supports all sectors of the Blue Economy

Environment

Coastal Services

Trade & Marine Navigation

Natural Resources & Energy

Education, Public Health & Recreation Extremes, Hazards & Safety

Marine Food

Society

Figure 1: The 12 Copernicus Marine Blue Markets. 1

2. EXPLAINING THE OFFER MOi has also developed editorial and digital contents to explain the Copernicus Marine Service offer (Figure 2) to all Blue Markets, from citizens to data scientist to researchers and private companies, engineers or project

Polar Environment Monitoring Climate

Economy

Science & Innovation Policies

Marine Food Coastal

Services & Ocean & Adaptation 6 holders. A video showcasing the full Copernicus Marine Governance & Mitigation offer has been developed (see here), along with dedicated Trade Ocean 3 blue/white/green videos detailing the offer: 11 & Marine Health Education, Navigation 7 - the Blue (physical) ocean: Public Healthtemperature, salinity, & Recreation Marine currents, and waves (see videos here), Natural 4 Conservation Resources 12 - the White (sea ice) ocean: (see & Biodiversity Extremes,sea ice conditions & Energy 8 Hazards videos here), & Safety - the Green (biogeochemical) ocean: nutrients, living species, sea water quality and transparency (see videos here).

Figure 2: Explaining the Copernicus Marine offer: the Blue White Green data products.

201

#57

3. SHOWCASING THE USE OF DATA: USE CASES MOi has developed numerous inspiring examples (use cases) of how users benefit from the Copernicus Marine data. About 200 use cases or “success stories” (Figure 3) have been published on our web portal, highlighting the end-to end use of Copernicus Marine data and highlighting the Copernicus Marine Service users and their innovative

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

applications. Such use cases demonstrate public or commercial end-to-end downstream applications, but also the implementation of policies by public and private actors. They are the proof of a vibrant downstream sector. Such use cases are also compiled into use case books, by countries, ocean basins or Blue Markets. An online “submit your use case” section is also available to our users willing to submit their own use case. A dedicated User Uptake program (Durand et al., this issue) has also allowed to promote dedicated demonstration of the end-to end use of Copernicus Marine data.

Figure 3: About 200 use cases or success story have been published on the web portal, highlighting the end-to end use of CMEMS data.

202

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Copernicus Marine Service

4. MARKETING CAMPAIGNS: THE EXAMPLE OF THE #CMEMSFORH2020 CAMPAIGN

increase the blue market penetration. MOi has developed a mix of paid, earned and owned communication channels with campaigns to showcase and explain the offer and its evolution to all audiences and all market sectors.

MOi has also deployed marketing campaigns to increase presence and visibility of the Copernicus Marine Service on all digital supports (social media and web portal), and to

During the 2015-2021 period, Mercator Ocean has conducted several marketing campaigns in order to promote and raise awareness among the blue markets about the Copernicus Marine data. For example, in 2020, a dedicated campaign was conducted targeting H2020 projects holders (Figure 4).

Copernicus Marine Service

Figure 4: Visual identity of the #CMEMSforH2020 campaign.

H2020 is a European Union program to build a society and a world-leading economy based on knowledge and innovation across the whole Union, while contributing to sustainable development. This covers activities ranging from research to taking new products to the market with a new focus on innovation-related activities, such as piloting, demonstration and support for public procurement and market uptake. The “Blue Economy” is of course included, and the Copernicus Marine Service has a key role to play in providing users across the globe with free marine data. We created the #CMEMSforH2020 campaign driven by the Copernicus Marine Service to: - stimulate marine data uptake among H2020 project holders, - provide better understanding of the usefulness of Copernicus Marine data for scientific disciplines, the business sector and research institutes.

This campaign showcases H2020 projects using Copernicus Marine data with significant economic, societal and environmental impacts. To engage all the different European Commission directorates and organisations involved in the H2020 programme, a communication kit was built to increase the visibility and the impact of the #CMEMSforH2020 campaign. This toolkit comprised a webpage, a news article, two videos (a short version and a long version with user testimonies), an email template and social media posts on Twitter and Linkedin. In only one week, the dedicated #CMEMSforH2020 hashtag generated: - 685,6 k digital impressions across Twitter and Linkedin, - 60 tweets, - 45 unique contributors.

203

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 5: Snapshot of the General Assembly digital platform.

5. ORGANISING EVENTS: THE EXAMPLE OF THE COPERNICUS MARINE GENERAL ASSEMBLY During the 2015-2021 period, Mercator Ocean International has organized or participated in many events, thematic workshops or training workshops (Giordan et al., this issue). The COVID-19 pandemic forced us to switch all face-to-face to online events, hence losing the warm and friendly social atmosphere but opening new opportunities with online digital platforms dedicated to event organization.

For example, MOi has successfully organized the first online Copernicus Marine Service General Assembly in January 2021 (Figure 5). It gathered all Copernicus Marine Service partners and teams, including: - Monitoring and Forecasting Centres, - Thematic Assembly Centres, - Central Information System, - User Uptake and Service Evolution projects. The General Assembly served to review the achievements of the Copernicus Marine Service over the period 2015 – 2021 and build future plans. It was a successful event gathering: - 11,2 k visits on the homepage of the virtual platform, - 635 unique participants, from more than 77 countries, and with a total of 16 hours of broadcast, - 5000 visits on virtual booths.

204

#57

- a side event was held during the OEE annual meeting in 2018, - a webinar entitled “What can the Copernicus Marine Service do for ocean energy?” was organised with 80 participants in 2020.

6. DEVELOPING PARTNERSHIPS MOi has also developed Market Development activities to seek new opportunities for data use in new communities. It has organized partnerships with relevant market trade associations such as Ocean Energy Europe (OEE) and the European Aquaculture Technology and innovation Platform (EATIP). OEE is the largest network of ocean energy professionals in the world (Figure 6). Its mission is to create a strong environment for the development of ocean energy, improve access to funding and enhance business opportunities for its members. Due to its large network of ocean energy professionals, part of a sustainable blue growth industry, OEE is a legitimate structure to engage in partnership with MOi in the natural resources and energy sector. The objectives of the partnership between MOi and OEE are to promote each other, to collect market requirements through feedback gathered from OEE members and to support OEE to better understand the ocean energy sector and facilitate the sharing of expertise. After 2.5 years of collaboration: - more than 40% of OEE members are now using Copernicus Marine products, - s everal use cases on the Marine Renewable Energy sector were submitted,

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

The Copernicus Marine Service also seeks to reach actors and stakeholders in the Aquaculture sector, which is expanding and represents a strategic sector for many countries. A partnership between the European Aquaculture Technology and innovation Platform (EATiP) and MOi began in 2018 and two joint events (Figure 7) were held in 2019 and 2020 (more info here). These conferences explored application possibilities of the available open-access satellite, model-based and in situ marine data for the aquaculture sector. This was done by sharing experiences and challenges that representatives from the aquaculture sector face managing fish farms. It was followed by a demonstration of how the Copernicus Marine Service might contribute to tackle these and possible future challenges. These events generated a widespread interest and there were: - 80 participants for each event, - 45% of the attendants from the private sector, - 40% worked in R & D organisations or universities, - 34% of the respondents had no or little experience with the use of the Copernicus Marine Service data.

Figure 6: Ocean Energy Europe infographic.

205

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 7: Two events were organized together with EATIP.

7. CONCLUSION AND PERSPECTIVES User engagement and market development activities during the 2015-2021 period fostered uptake of Copernicus Marine service portfolio, developed market intelligence, and opened new opportunities for data use in new communities.

Such activities will be duly continued in the next 2021-2027 phase. Moreover, the EU Space Programme Agency (EUSPA) will foster synergies amongst the EU space programmes (Galileo, EGNOS, Copernicus) promoting and supporting the downstream/applications market development and user uptake sectors. MOi will closely interact with EUSPA on activities related to Copernicus Marine Service user uptake and user engagement, in particular, with respect to the downstream private sector. In addition, for each Blue Market there will be development of more tools in order to facilitate user experience and simplify access to data.

206

THE COPERNICUS MARINE USER UPTAKE PROGRAMME

DURAND, E., CAILLEAU, S., CHABOT, G., DERVAL, C., DE NUCÉ, A., OBATON, D. Mercator Ocean International, France.

the successful elements of the service relationship, between Copernicus Marine and its users, more specifically intermediate users who are themselves service providers.

OVERVIEW: MOTIVATION AND STRATEGY To encourage use and uptake of Copernicus Marine products, Mercator Ocean International (MOI) has set up a Copernicus Marine User Uptake programme driven by two challenges: - demonstrating the use of the Copernicus Marine Service, - creating economic activity for private and public actors in the European Union.

The User Uptake Strategy has been set up to increase the loyalty and visibility of Copernicus Marine Service users, to attract new communities and to foster the service uptake. It includes the promotion of downstream services developed by private and public intermediate users, who can demonstrate how Copernicus Marine information is integrated into their own downstream services. The User Uptake (UU) programme shows the added value of the Copernicus Marine chain till the end-users.

The User Uptake programme addressed the widest possible panel of users and its objectives were to identity

Figure 1: The CMEMS User Uptake programme at a glance.

207

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

on demonstrations of operational and coastal downstream services including those linked to EU Marine Strategy Framework Directive (MSFD) and on an inventory of mobile applications that use Copernicus Marine products.

1. 2015-2021 PROGRAMME IMPLEMENTATION The User Uptake programme has taken the shape of several successive calls for tender to involve many private and public companies in the uptake of the marine products. Six calls for tender were issued and focused mainly on marine coastal or offshore services while one was dedicated to an inventory of existing mobile apps. A first batch of 18 contracts were selected (DEM1, DEM2, DEM3 and INV1) and started on March 2017; they focused

A second batch of 10 User Uptake contracts (DEM4) started on May 2018 and focused on new demonstrations of coastal and operational downstream services to complete geographical regions not covered by previous calls for tender. In 2019, a third batch of 12 contracts (DEM5) started on May 2019 and focused on new demonstrations of CMEMS downstream services to complete geographical regions not sufficiently covered by previous calls for tender, the Arctic Ocean and the Global Ocean in particular.

Figure 2: : Schedule of the 6 calls for tender launched in the framework of CMEMS User Uptake programme.

For each call for tender and after receipt of offers, Mercator Ocean analysed all proposals with respect to the criteria listed in the tender. Then a Tender Evaluation Board took place to evaluate collectively the received proposals. This Tender Evaluation Board was composed of external independent experts, European Commission members and the Mercator Ocean evaluation team. Over the period 2015-2021, Mercator Ocean has received and analysed 122 different proposals from 111 different organisations (64 private / 47 public - 21 EU countries / 3

non-EU). After selection phases, a total of 40 projects were selected from 46 different organisations (28 private / 18 public - 16 EU countries / 2 non-EU). The ratio of private to public companies remained almost the same after the selection (about 1/3 public and 2/3 private). Some entities applied as individuals and others as the leader or partner in a consortium. The number of consortia was increasing year after year with innovative mixing between public and private entities and different countries.

208

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

39 downstream services demonstrations (from DEM1, DEM2, DEM3, DEM4 or DEM5 calls for tenders) and an inventory of mobile applications using Copernicus Marine products (from INV1).

2. FORTY CONTRACTS SELECTED Table 1 below sums-up the contracts selected in the frame of the User Uptake programme. These contracts represent Organisation

Main market

Description

Domain

INV1

TEAMSURV (private, UK)

Inventory of mobile applications available on on-line application markets.

DEM1

SHOM (public, FR)

Extremes, Hazards & Safety

data.shom.fr - High-resolution forecasting system for marine safety activities in IBI coastal area.

IBI

DEM1

HIDROMOD (private, PT)

Extremes, Hazards & Safety

Marine Weather Information - High resolution forecasting system for port operations, oil spill, water quality - Lisbon

IBI

DEM1

NOLOGIN (private, SP)

Trade & Marine Navigation

DEM1

PLANETEK / EVENFLOW (private, IT/BE)

DEM1

NIMRD / ACTION MOD. (public/private, RO/PT)

DEM1

Environmental Control Panel - Coastal high resolution mobile application - 2 harbours (Barcelona, Algeciras).

MED

Ocean Health

RheticusTM Marine - Monitoring coastal water quality and eutrophication in Italian and Greek coastal areas.

MED

Ocean Health

iSWIM - Integrated Service for Water Quality Monitoring in Mamaia bay - Romanian coast.

NOVELTIS (private, FR)

Natural Resources & Energy

TidEA - Indicator and maps for implementation of coastal tidal technologies (Tidal Energy Assessment).

GLO

DEM1

HIDROMOD (private, PT)

Extremes, Hazards & Safety

Marine Weather Information - High resolution forecasting system for port operations, oil spill, water quality - Azores

GLO

DEM2

POLAR VIEW (public, UK)

Polar Environment Monitoring

Greenland Floe Edge Service - Greenland Community Ice Information to support traditional ways of life in the Arctic.

ARC

DEM2

FMI (public, FI)

Polar Environment Monitoring

BALFI - Baltic Sea landfast ice extend and thickness.

BAL

DEM2

DTU (public, DK)

Marine Food

OCEBIS - Ocean Biological Information Service for fishermen and researchers.

NWS

DEM2

TELESPAZIO (private, FR)

Ocean Health

Coastal Turbid Plume Survey - Predictive and integrated coastal water quality management tool for the Basque coast

DEM2

ISPRA / OGS (public, IT)

Ocean Health

CADEAU - Annual bulletin on Marine Environmental State in support of the application of the different EU directives.

MED

DEM2

HIDROMOD (private, PT)

Extremes, Hazards & Safety

Marine Weather Information - High resolution forecasting system for port operations, oil spill, water quality - Madeira

IBI

DEM3

AZTI (public, SP)

Policies, Ocean Governance & Mitigation

CHLO4MSFD - Satellite-derived Chl-a products for MSFD. Descriptors: D1 (Biological diversity) - D5 (Eutrophication)

IBI

DEM3

DELTARES (private, NL)

Policies, Ocean Governance & Mitigation

MSFD-EUTRO - Satellite-derived Chl-a products for MSFD. Descriptor: D5 (Eutrophication)

NWS

DEM3

CEFAS (public, UK)

Policies, Ocean Governance & Mitigation

DEM3

QUIET OCEANS (private, FR)

DEM4

CefMAT -Tool to assess the capability and limitation of CMEMS products to answer to MSFD descriptors.

IBI

NWS

Policies, Ocean Governance & Mitigation

Quonops - Operational ocean noise forecast. Descriptor: D11 (Energy including underwater noise)

GLO

NOVELTIS /QUIET OCEAN (private, FR)

Polar Environment Monitoring

ARCTIVITIES – Sea state and noise indicators in polar regions.

ARC

I-SEA / HYDROCOTE / N7MOBILE / PLANETEK (private, FR/PL/IT)

Ocean Health

HABRisk - Surveillance and drift forecast for algal bloom. Demonstration for beach closure of the Polish coast.

BAL

209

#57

Organisation

Main market

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Description

Domain

DEM4

I-SEA / HYDROCOTE / N7MOBILE / PLANETEK (private, FR/PL/IT)

DEM4

UNIVERSITY OF LATVIA (public, LV)

DEM4

RBINS / MET NO / METEO-FR (public, BE/NO/FR)

DEM4

ARGANS (private, UK)

DEM4

SOCIB / AZTI / RPS (public/private, SP/ USA)

DEM4

COLOMBOSKY (private, IT)

Marine Food

JellyX - Monitoring tools for the identification of jellyfish swarms.

DEM4

TERRASIGNA / NIMRD (private/public, RO)

Marine Food

SkyFISH - Habitat sustainable indices for fishing activities and aquaculture.

DEM4

UFPR / ENVEX / IST (public/private, BR/PT)

Natural Resources & Energy

Sea Observatory - Met-ocean forecasts from regional and local coastal operational modelling systems

GLO

DEM4

IHCANTABRIA (public, SP)

Natural Resources & Energy

RENAQUA DSS - Location opportunities for aquaculture farming and wind or wave renewable energies.

GLO

DEM5

SUEZ / AZTI / TELESPAZIO (private, FR/SP)

Ocean Health

FML-TRACK - Monitoring of Floating Marine Litter to guide collect operations.

IBI

DEM5

NOC (public, UK)

Marine Food

SSW-RS - 25-year physical reanalysis on the Scottish shelf waters.

NWS

DEM5

FO.R.T.H / MEEO / ORION (public/private, GR/ IT/CY)

Trade & Marine Navigation

COASTAL CRETE - High-resolution forecasting system for marine safety activities in the area of Crete Island.

MED

DEM5

ARTELIA (ex-OO) (private, FR)

WaX-Coast - Coastal extreme waves and statistics information for insurance and reinsurance companies.

MED

DEM5

PAIC (private, LV)

Trade & Marine Navigation

HYWAS-PORT - Operational service for safe navigation in the approaches to ports and within ports.

BAL

DEM5

DRIFT+NOISE / MET NO (private/public, GE/NO)

Polar Environment Monitoring

IcySea - Sea-ice information in line with the Polar Code in the Svalbard region.

ARC

DEM5

POLAR VIEW (private, DK)

Polar Environment Monitoring

Polar Code Service - Aggregation of sea-ice information considering the Polar Code recommendations.

ARC

DEM5

DMI (public, DK)

Polar Environment Monitoring

SHIPcAPP - High quality sea-ice information for shipping around the Cape Farewell.

GLO

DEM5

DELTARES (private, NL)

Ocean Health

COASTSERV - Interface to identify relevant datasets for a specific South African coastal location.

GLO

DEM5

CITICAN / IHCANTABRIA (private, SP)

SOSeas - Assessment tool for predicting the dynamic risk of drowning on Brazilian beaches.

GLO

DEM5

D-ICE ENGINEERING (private, FR)

SATORI - Tool to increase ship-routing performances, in hindcast mode, for ships with hybrid propulsion.

GLO

DEM5

ARTELIA (private, FR)

StArt - Statistical met-ocean information to qualify variability and extreme values of site conditions.

GLO

HABRisk - Surveillance and drift forecast for algal bloom. Demonstration for beach closure of the Polish coast

BAL

Extremes, Hazards & Safety

Present to Present - Drift simulation for prevention of pollution in marine environment.

BAL

Extremes, Hazards & Safety

NOOS-Drift - On-demand drift forecast service with improved estimation of uncertainty.

NWS

LITTER-TEP - Source and landing of macro litter thanks to drift forecast.

NWS

Ocean Health

Ocean Health Extremes, Hazards & Safety

Extremes, Hazards & Safety

Extremes, Hazards & Safety Trade & Marine Navigation Natural Resources & Energy

IBISAR - Most accurate current prediction for search & rescue and emergency operations.

Table 1: : List of the 40 contracts selected in INV1, DEM1, DEM2, DEM3, DEM4 or DEM5.

210

IBI MED

#57

The 39 User Uptake service demonstrations are representative of all the communities targeted by Copernicus Marine marketing activities (Abadie et al., this

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

issue) and serve as exemplary and detailed use cases. All Copernicus marine geographical regions are covered and 9 out of 12 blue markets are represented (see Figure 3).

Figure 3: Geographical and blue markets distribution of the User Uptake services demonstrations.

211

#57

3. PRODUCTS USED IN USER UPTAKE SERVICES Figure 4 shows, for each entity involved in the programme (33 leaders’ entities out of 39 demonstrations – 5 entities

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

were awarded with 2 or more contracts), the number and the type of products downloaded and used by User Uptake contractors. Each colour represents a type of products: in deep blue analysis and forecasts model products; in deep orange reanalysis model products; in light blue near real time observation products; and in light orange reprocessing observation products.

Figure 4: Product’s breakdown per type and per service.

Most of the products downloaded are analysis and forecasting model products (Model ANA&FCST in deep blue) and near real time observations (OBS NRT in light blue) as the calls for tenders required operational services. Most of the contractors use at least five products. One contractor uses almost twenty different products as its application can be used all over Europe and the world. To go into more detail, Figure 5 (left) shows that the projects use 84 different products from the catalogue (out of 180); and, except for the MultiObs product, all product families are represented. The User Uptake contractors clearly explain in their demonstration, the usefulness of the marine products and how they are integrated in their

downstream service. It is also possible to collect the type and the number of variables used (see Figure 5 right, with biogeochemical variables in green, physical variables in blue and sea-ice variables in grey). The marine products are integrated in the User Uptake services in various different ways: downscaling with high resolution coastal model or multi model approach, with satellite derived products, with drift models, doing statistics with extreme events, calculating indices or risk indicators, with a combination of products (models, insitu, satellites) and/or artificial intelligence with machine learning or neural networks.

212

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

Figure 5: Number and type of marine products (left) and of variables (right) used in User Uptake Demonstrations.

4. PROMOTION AND DEMONSTRATIONS

and other major players in ocean knowledge through conferences, workshops, articles, European projects, posters, social media and/or training courses.

When the services are developed and operational, Copernicus Marine makes them visible on its portal in the dedicated «Use Case demos» section, thus highlighting the usefulness of Copernicus Marine information (Abadie et al., this issue).

In 5 years, about 30 events organised by Copernicus Marine have highlighted the User Uptake programme and presented all developed services (or a particular service when linked to a particular event by targeted activities or a geographic region). Promotional activities were also carried out by contractors in more than 100 national or international conferences or workshops.

The User Uptake concrete examples have been promoted and disseminated by the contractors, by Copernicus Marine

Figure 6: ”Use Case Demos” section dedicated to the User Uptake programme on Copernicus Marine website.

213

#57

5. CONCLUSION AND PERSPECTIVES First calls for tender were launched in 2016 and immediately met success. Five years later, the User Uptake programme is a key element to support the development and the promotion of new applications. It is a booster for the selected companies. The User Uptake programme was built on a mutually profitable approach. On the one hand, the Copernicus Marine Service supports the development of downstream services so that they can best benefit from Copernicus Marine Service products and services. In return, contractors promote the Copernicus Marine Service products all over Europe throughout their services. During workshops, conferences, or all other kinds of events, the contractors remind users that their services exist thanks to the open and free Copernicus Marine Service information. Moreover, Mercator Ocean has collected detailed and concrete end-of-contract feedback to feed the User Requirement Document (Delamarche et al., this issue). Mercator Ocean has also collected feedback on the socioeconomic impact of the Copernicus Marine Service. The sustainability of the Copernicus Marine Service is essential

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

for users developing operational services. At the end of 2020, it is estimated (from statistics reported by contractors) that 24 User Uptake services serve more than 50,000 users. This estimation only considers DEM1, DEM2, DEM3 and DEM4 services one year after their launch. Twothirds of the services serve more than 100 users while three services (from SHOM, NIMRD and FMI with services for the public where registration is not required) have a total of almost 40,000 end-users. The User Uptake programme has grown in maturity, and beneficiaries have evolved showcasing more diversity, more private companies, and spread to more countries. It has offered many opportunities to create innovation. For the future, and to address marine environmental concerns, there is still a need to highlight the way highquality marine products and information are processed and used to support policymakers, businesses, and citizens. During the 2021-2027 period, the EU Space Programme Agency (EUSPA) will foster synergies amongst the EU space programmes (Galileo, EGNOS, Copernicus) promoting and supporting the downstream/applications market development and user uptake sectors. Mercator Ocean will closely interact with EUSPA on activities related to Copernicus Marine Service user uptake and user engagement with respect to the downstream private sector.

214

DIGITAL COMMUNICATIONS AND SOCIETAL AWARENESS ACTIVITIES QUADE, G. , ABADIE, V., BASTIDE, L., THOMAS-COURCOUX, C., CROSNIER, L., LEGROS, V., QUÉAU, A., TRONCI, M. Mercator Ocean international (MOi), France

OVERVIEW Communication assets (both on and offline), ocean literacy tools, and societal awareness can be divided into three major categories (although these all overlap substantially): 1) Digital (Website, Tools & Social Media), 2) Editorial & Press Relations, 3) Ocean Literacy & Outreach. These are designed to deliver Copernicus Marine Service expertise to a wider audience through the translation from scientific language and findings for our different target audiences, then to distribute and transfer such tools in an effort to drive uptake. During the first [2015-2021] Copernicus phase, we have produced over 3000 videos, 360 articles & publications, and launched the redesigned website with a suite of new ocean literacy tools. This work is in conjunction with our events, training (Giordan et al., this issue), marketing (Abadie et al., this issue), and user uptake program (Durand et al., this issue) actions. Moreover, our social media platforms have helped us to cultivate a growing community of scientists, policy-makers, and ocean enthusiasts.

1. MAIN ACHIEVEMENTS 1.1 The main achivements are on the field of digital activities: Website, Tools and Social Media The Copernicus Marine Service website has evolved substantially over the course of the 2015-2021 Copernicus time period in terms of the static and dynamic content. In

January 2021, a new version of the website was launched that was made in line with guidelines given to the different Copernicus services to homogenise their web portals. This was done for both aesthetic and UX (user experience/ usability) purposes. Currently the product catalogue is decorrelated and it will also soon undergo a revamp in 2021. The traffic to the site has increased steadily since 2016, with unique visits more than doubling. The editorial site (not including the ocean data products) has revealed some surprising trends. For example, the news and events and educational sections were among the most popular pages of the site. For this reason, the website revamp includes a new suite of sections (described below) dedicated to ocean literacy as well as a more robust editorial publishing calendar. Our main social media accounts are through Twitter, Linkedin, and Youtube. Twitter represents a channel where we connect well with our technical user-base as well as with policy-makers, organisations, and ocean science / conservation enthusiasts. The base of followers has steadily increased from 2017 and we maintain an active and engaged community. Our Youtube channel has over 3,200 videos, which include, but are not limited to: scientific content, animations, ocean literacy outreach tools, ocean data animations, conference presentations, and training and workshops. Another major achievement includes the multiple viewing tools that are now available for users to visually explore our different datasets. The MyOcean Viewer (pro version) was launched in 2020, which includes the ability to create cross sections, select regions, and generate graphs with selected variables. Layering, superimposing layers with different opacities, is also possible allowing users to compare multiple datasets. In addition, the selected maps and time frames can be exported as videos, images or embedded elsewhere. The original viewer as well as the Pretty View viewer are also available via the product portal.

215

#57

Figure 1: Growth of unique visits on Copernicus Marine Service webportal.

Figure 2 & 3: Community growth on social media.

Figure 4: MyOcean Viewer data visualisation tool.

216

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2. EDITORIAL, NEWSLETTER, AND PRESS RELATIONS

advances in the field of operational oceanography, ocean policy and conservation, and the work of our partners. In 2019 we launched a quarterly newsletter. We also regularly participate in publications in the Copernicus Observer.

News articles posted on our site and on the Copernicus Observer have been identified as a powerful way to regularly deliver scientific, technical and non-technical educational content to our wide range of users. We increased the number of articles published in 2020 by 50% as compared to 2016. Our website has some 240 articles and 120 event articles. In our editorial content we cover many topics including but not limited to: information on the state of the ocean, advances in our services and products,

Having identified the enormous impact of press and mainstream news in the widespread distribution of information, in 2021 we significantly changed our strategy and internal workflow. We have gone from relying heavily on in-house editorial production and distribution, to working with several top science reporting and press relation agencies in Europe. The fruits of this work will be felt in the next [20212027] Copernicus phase, where we will continue to rely on expert European partners for press relations. This includes the translation and distribution of press releases and editorial content into major European languages for maximum uptake.

Figure 5: Revamped website announcement.

217

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

2.1 Ocean State Report Summaries, videos and content

2.2 The Blue Book: Copernicus for a Sustainable Ocean

The first Ocean State Report was launched at the UN Ocean Conference in June 2017. During 2015-2021, there were four Ocean State Reports (OSR) published and 4 associated summaries. Starting with the OSR2, there were increased resources dedicated to creating these tools that synthesize and simplify the reports’ main findings for the target audiences (i.e., the general public, policymakers, etc.). After the great success of the 2nd summary, we have each year further developed the activity, with the addition of educational videos, social media campaigns, and press relation initiatives.

The Blue Book “Copernicus for a Sustainable Ocean”, published in November 2019 promotes and tells the story of the Copernicus Marine Service, which has been implemented and operated by Mercator Ocean International since 2015. It is designed to inform all European citizens including policymakers, students and youth on how the Copernicus Marine Service benefits society as a whole. Among its contributors are decision-makers, entrepreneurs, experts, citizens, and scientists from all over the world that are directly involved in ocean-related issues. It demonstrates the impact and benefits of the Copernicus Marine Service according to such ocean-related issues.

OCEAN STATE REPORT SUMMARY

Implemented by

Figure 6 & 7: The annual OSR Summary and the Blue Book, two cross-cutting publications released by the Copernicus Marine Service.

218

#57

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

3.3 Policy Tools

3. OCEAN LITERACY AND OUTREACH Ocean literacy, education and outreach are important parts of the Copernicus Marine Service’s strategy for reaching a wider audience and providing information about the ocean’s physical state and the challenges it faces. 3.1 Societal Awareness Through outreach events, partner initiatives, and museum exhibitions, we estimate that our actions have reached out to about 2.2 million citizens. The Copernicus Marine Service provides access to marine data and ocean literacy information, as well as technical support to build educational tools. During awareness events, scientists from the Copernicus Marine Service are also able to engage in sharing ocean literacy tools and materials. A key part of this is developing partnerships with nonprofit organisations, as well as leading and participating in awareness events. Through ocean literacy networks, nonprofit partners (such as Children for the Oceans, Coral Guardian, Emily Penn’s eXXpedition) and events, the Copernicus Marine Service and Mercator Ocean share scientific knowledge with civil society and promote sustainable ocean initiatives. 3.2 SDG 14 Sulitest ocean literacy module on behalf of the EU The European Union has a demonstrated commitment to the 2030 Agenda on Sustainable Development that specifically highlights actions on SDG 14. As a part of addressing United Nations Sustainable Goals (SDGs), Mercator Ocean, in the frame of the Copernicus Marine Service has joined Sulitest to create an educational module on SDG 14 on behalf of the EU and it will be translated into 10 languages. It covers key mechanisms of and challenges facing the ocean. For example, it will highlight ocean acidification, for which the Copernicus Marine Service has developed a suite of datasets and indicators (including contributions to EUROSTAT/EEA in support of SDG 14).

A new section of our site is dedicated to understanding marine policies on a European and international level. This can help users to see the larger picture and how our ocean data is key for the deployment and implementation of a wide range of policies in areas such as the Arctic region, environmental conservation, climate change, biodiversity protection, water quality, coastal regions, plastic pollution and international cooperation. Since 2020, the Copernicus Marine Service has been working to support the European Green Deal, a roadmap for making the EU’s economy sustainable, and which provides a set of policy initiatives and climate action plans. It is a new growth strategy that aims to transform the EU’s economy to ensure a sustainable future. For example, through the EU4OceanObs project, which serves to strengthen international ocean governance and commitment for a comprehensive global ocean observing system, the Copernicus Marine Service is recognised for providing data management infrastructure and access portals, both key parts of the value chain. 3.4 Plastic Pollution Marine plastic litter is damaging to marine ecosystems and wildlife and it is an issue that countries around the world are increasingly interested in addressing. We released a new section on our website on this topic that presents themes to understand marine plastic pollution, monitoring methods, sources and impacts, exploring mitigation and reduction policies and initiatives, and how the Copernicus Marine Service can support this cause. 3.5 Monitoring the Ocean The Copernicus Marine service data comes from global and regional numerical models, in situ observations, and satellite observations. The ocean variables provided by the Copernicus Marine Service can be divided into three main categories, referred to by their associated colour: the Blue (physical), the Green (biogeochemical) and the White (sea ice) Ocean. The Copernicus Marine Service has developed specific content to explain these marine variables and make them more easily available to users. Over 20 educational videos are a part of this series designed to demystify different ocean parameters and their applications in the real world.

219

#57

4. STATUS OF THE END OF COPERNICUS 1

MERCATOR OCEAN JOURNAL SEPTEMBER 2021

4.3 Translation of the Copernicus Marine website static content into the EU languages - under the first [2015-2021] Copernicus phase, translation has begun for five major EU languages: French, Italian, German, Spanish and Portuguese, - under the next [2021-2027] Copernicus phase, the site’s static content will be translated into the remaining official EU languages.

In 2020 as well as Q1 of 2021, we have launched various projects and initiatives to continue implementing our communication strategy. These projects are at varying stages and will all be deployed in 2021. Among these include: 4.1 Educational Explainers We are working on a section of our website that allows us to explain various oceanic phenomena, and concepts in the world of operational oceanography to the general public and policy-makers. This template within our website will be used to continue to grow our repertoire of explainers. The initial topics selected include: ocean warming, extreme events, operational oceanography, in situ, numerical models, and satellites. 4.2 Ocean Climate Bulletin Our Ocean Monitoring Indicators (OMIs) catalogue is widely appreciated not only in the scientific community by policymakers and the general public as they provide trends on the health of the ocean. In response to user demand, we are working on better highlighting about 10 key OMIs that include data that is more recent than in the existing catalogue. Moreover, they will be presented with pedagogical materials, videos and interactive graphs.

4.4 Communications and press There will be continued improvement of our editorial content developed for our website, the Copernicus Observer, and to be published and covered by specialist and mainstream news sources. For example, we will have a campaign on the Copernicus Marine Service achievements in the first [2015-2021] Copernicus phase. We will also publish several spots in magazines to show the Copernicus Marine contributions to monitoring the ocean state, pollution, and coastal regions, for example.

5. FUTURE PROSPECTS ON COPERNICUS 2

In the last years, we have prepared ourselves well to ramp up to achieve our goals for the next [2021-2027] Copernicus phase. Having nurtured our internal and external resources, we are better positioned to scale up our activities. Our strategy includes a focus on the three main facets of communication: Digital (website, tools and social media); editorial & press relations; and ocean literacy & outreach. The most growth-focus will be on the editorial and press relations, as the latter is a relatively nascent activity for our team and the potential for impact is the highest. There will also be increased focus on developing dedicated digital tools for different audiences to explore their ocean interests. Communications and tool partnerships have also been identified as important elements to cultivate.

Figure 8: Important facets of communication used by the Copernicus Marine Service.

220

JOIN THE COPERNICUS MARINE SERVICE COMMUNITY Webportal of Mercator Ocean International mercator-ocean.fr/en

Mercator Ocean @MercatorOcean

Webportal of Copernicus Marine Service marine.copernicus.eu

MercatorOcean

Service desk email sericedesk.cmems@mercator-ocean.eu Collaborative forum forum.marine.copernicus.eu

Mercator Ocean mercator-ocean Copernicus Marine Service Copernicus Marine Service

221

Copernicus Marine Service @CMEMS_EU

Copernicus Marine Service Copernicus Marine Service