TRUE Tube Map

Page 1

Euston Warren Street

Oxford Circus

Sound Pressure Level max exposure 98.47 [dB]

Knightsbridge

Earl‘s Court

Gloucester Road

Hyde Park Corner

South Kensington

West Bromton

Fulham Broadway

TRUE T Tube ube Map a proj pr oject ect by Lui Luisa sa Amann

1

Green Park


TABLE OF CONTENTS

UCL FACULTY OF THE BUILT ENVIRONMENT BARTLETT SCHOOL OF ARCHITECTURE

4

Introduction

4

The Tube Surfers | Background

MSc SPACE SYNTAX: ARCHITECTURE AND CITIES Coursework Submission Form

Word Count: 1,294

Student’s Name (IN CAPS): LUISA AMANN Coursework Title: TRUE Tube Map

Deadline for Final Submission: 25.4.2023 Actual Date of Submission (if different): (If different please state whether this has been agreed with the module leader and the appropriate documentation provided)

Student Statement • I confirm I understand the meaning of plagiarism and that I may be penalised for submitting work that has been plagiarised. • I confirm the work submitted is my own unaided work and that all sources used in its preparation and all quotations are clearly cited. • I confirm that all work will also be submitted electronically and that this can be checked using the JISC detection service, Turnitin. • I confirm that I understand that the work cannot be assessed unless I have handed in both hard copy and electronic versions.

Student Signature: ................................................................................. Your work will be returned unmarked if you fail to sign here

Spatial Computation | The Project

4

Project Development and Methodology

7

Step 2 | Cleaning The Data using Python

7

7

Step 1 | Base Set-Up in QGIS

Step 3 | Spatial Computation and Visualisation in Grasshopper

16

Conclusion

Closing Thoughts

16

Limitations

16

Additional Note

17

Other Uses of the Grasshopper Tool

16

Outlook

17

References

18

Figures and Tables

18

Datasources

18

Thank You

2 Document1

14

19

3


INTRODUCTION

In light of rapid urbanisation and growing population density, public transport plays an increasingly important role in overcoming problems such as traffic congestion, air pollution, social exclusion, and poor urban health (Rojas-Rueda 2012). To successfully promote a shift from private motorised to public transport, passenger comfort is a crucial yet long-neglected factor beyond travel times, cost, reliability, frequency, and accessibility (Ebrahimi and Bridgelall 2021, Imre and Celebi 2017). Public transport improvements often target the latter service quality attributes, even though experienced passenger comfort appears to be critical for raising passenger volume and well-being (Redman 2013, Liu 2021). The TubeSurfers | Background

In 2021, a group of UCL Engineering Ph.D. students (The TubeSurfers) won a ‘Curiosity Research Grant’ to collect and analyse environmental data (e.g. noise, temperature, humidity, air quality incl. PM2.5 and CO2) across the London Underground (LU). Their research objective was to develop an open-source Temperature and Resonance in an Underground Environment (TRUE) Tube Map to help Londoners take better decisions about their journeys and assist transport planners in effective mitigation planning. Spatial Computation | The Project

Earlier this year, a meeting was conducted in which they introduced me to their work and indicated open challenges concerning spatial data visualisation: Although the website draft displayed a total data overview (see Figure 1), it yet lacked • GOAL 1 - the option to analyse specific tube journeys in detail (“from Station A to B”) as well as • GOAL 2 - distribution curves illustrating data variation along segments.

In the course of further discussions and within the Spatial Computation and Dynamics module, headed by Dr. Tasos Veroudis at the Bartlett Faculty of the Built Environment, developing this add-on became subject of this project. Its detailed computational design is presented step-by-step in the following. 35

50 [dB]

115

Figure 1. Excerpt of Website Draft, Comfort Indicator Selected: Sound Pressure Level

4

5


PROJECT DEVELOPMENT AND METHODOLOGY STEP ° 1 | Base Set-Up in QGIS 1.1. Recreating the Tube Network

Transport of London (TfL) publicly shares its station locations as geo-coded KML feed. The point dataset was downloaded and imported to QGIS (geographic information system software). As comfort data was collected by the TubeSurfers exclusively in zone 1 to 2/3 for underground lines (covering 52% of the LU network), non-data nodes were deleted and segments drawn to replicate the network of interest (see Figure 2).

1.2. Adding X and Y Coordinates

To further process the result in Rhinocerus McNeel (3D modelling software), x and y coordinates were added to segments using the field calculator (expression: $y_at(0) and $x_at(0) for the line start point, $y_at(-1) and $x_at(-1) for the end point coordinate). The final attribute table was then exported as a .csv file.

1.3. Setting up the Surrounding

To embed the tube network in its larger environment, building data for the Greater London Area was downloaded from Ordnance Survey (.shp, tile TQ), converted in QGIS (.dxf) and imported in Rhino (.3dm).

STEP ° 2 | Cleaning The Data using Python

The raw data collected by the TubeSurfers is compiled in two .csv files – one indicating timestamps, the other recording comfort indices (see Table 1 and 2, next page). With a total of >540,000 datapoints for >87,000 timestamps it was imperative to clean and simplify the data for further processing. Accordingly, Anaconda and JupiterLab were utilised to devise a Python script.

2.1. Cleaning the Timestamps.csv

When collecting the data, some indicators were measured in different intervals while traveling traveling back and forth through the LU network (for details on equipment used and measurement intervals see Table 3).

Station

Segment

6

Figure 2. Georeferenced Stations, Tube Network Segments and Surrounding (in QGIS)

7


D

Indicator

DataType

Sound Pressure Level

Equipment

Zoom H4n Pro Handy Recorder with Ryeote Wind Jammer

Parameter

Measurement Interval

SPL (dB)

0.5 s

Temperature Air Quality

PCE-WB 20SD Dry Bulb Thermometer PCE-AQD 20 Air Quality Logger

TA (°C), RH (%) PM2.5 (uq/m3), CO2 (ppm)

1s 2s

Table 3. Range of Indicators and Equipment used to measure passenger comfort on the LU

To generate data distribution curves along each segment (GOAL 2), datapoints where the train stood still at the station are deleted. Additionally, duplicate datapoints (for when the researcher travelled through the same tube line opposite way) are removed. P ython Script import pandas as pd Timestamps = pd.read_csv(‘Timestamps.csv’, sep=’,’) df = pd.DataFrame(Timestamps) df2 = df.dropna() #delete datapoints for which the trains stood still at the station df2 = df2[df[‘FromStation’] != df2[‘ToStation’]] #remove datapoints for which the researcher travelled through the same tube line opposite way filtered = df2[[‘FromStation‘, ‘ToStation‘]].drop_duplicates() filtered[‘tmp‘] = filtered.apply(lambda row: ‘‘.join(sorted([row[‘FromStation‘], row[‘ToStation‘]])), axis=1) filtered = filtered.drop_duplicates(‚tmp‘) df2[‘tmp‘] = df2.apply(lambda row: ‚‘.join([row[‘FromStation‘], row[‘ToStation‘]]), axis=1) df3 = df2[df2[‘tmp‘].isin(filtered[‚tmp‘])] df3 = df3.drop(‘tmp‘, axis=1) #print results to csv df3.to_csv(‘Timestamps_clean.csv‘)

2.2. Linking Comfort Indices and Timestamps to Create one Master File Table 1. Excerpt of original datafile indicating comfort indices ( > 540,000 datapoints)

Table 2. Excerpt of original datafile indicating timestamps ( > 87,000 datapoints)

8

In an interim step, comfort indices are appended to respective timestamps in QGIS using the ‘Join’ Option. The resulting merged .csv file is ready for further processing in Python.

9


2.3. Dropping Empty Fields

In an effort to further simplify the dataset, only timestamps for which all indices are sampled (i.e. every 2 seconds) are retained. Other rows with partially empty fields are dropped. P ython Script Timestamps_with_values = pd.read_csv(‘Timestamps_with_values.csv’, sep=’,’) df4 = pd.DataFrame(Timestamps_with_values) #clean rows where column CO2, PM25 or HealthIndex are empty (no data) df4 = df4.dropna(subset = [‘CO2’, ‘PM25’, ‘HealthIndex’], how=’all’)

2.4. Preparing the Master File for Grasshopper

For further processing in Grasshopper, comfort indices are grouped in a way that each row contains all relevant information for one segment. Moreover, the number of datapoints per segment is calculated. The clean, final ‘Datapoints’.csv (see Table 4) has 155 rows. P ython Script #count datapoints per segment df5 = df4.groubby([“FromStation”, “ToStation”]).size().reset_index(name=”Datapoints”) df5.to_csv(‘Datapoints_per_Segment’) #group comfort indices per segment in one row CO2 = df4.groupby([“FromStation”, “ToStation”]) [‘CO2’].apply(list) Humidity = df4.groupby([“FromStation”, “ToStation”]) [‘Humidity’].apply(list) PM25 = df4.groupby([“FromStation”, “ToStation”]) [‘PM25’].apply(list) Sound = df4.groupby([“FromStation”, “ToStation”]) [‘SPL Sound’].apply(list) Temperature = df4.groupby([“FromStation”, “ToStation”]) [‘Temperature’].apply(list) HealthIndex = df4.groupby([“FromStation”, “ToStation”]) [‘HealthIndex’].apply(list) #print results to csv df5 = (CO2), df6 = (Humidity), df7 = (PM25), df8 = (Noise), df9 = (Temperature), df10 = (HealthIndex) df5.to_csv(‘CO2.csv’) df6.to_csv(‘Humidity.csv’) df7.to_csv(‘PM25.csv’) df8.to_csv(‘Noise.csv’) df9.to_csv(‘Temperature.csv’) df10.to_csv(‘HealthIndex.csv’)

10

Table 4. Excerpt of final (clean) Datapoints.csv (155 datapoints)

11


IMPORT CSV WITH DATAPOINTS script below then recreates tube network in rhino

CHOOSE COMFORT INDEX OF INTEREST

RESULT IN NUMBERS

Joins nurbs curves (tube line and comfort rating distribution) to create a ruled surface

ValueList contains station names with respective x and y coordinate (journey selection)

script divides segments into their respective # of datapoints, assigns comfort rating as z-value to points and reconnects them to create a distribution curve

GOAL 1 - FROM/TO FILTER

CHOOSE JOUNREY START AND END STATION

12

script calculates avg/max/min exposure on route for given comfort index

GOAL 2 - ILLUSTRATE DATA VARIATION ALONG SEGMENT

colorcodes distribution surface to visualise data in detail/3D

ShortestWalk Component calculates shortest route from start to end point in a network

13


STEP ° 3 | Spatial Computation and Visualisation in Grasshopper

CO2 max exposure 21 ppm

3.1. Importing of Segments for all Tube Lines

Buildings.3dm and Grasshopper are opened in Rhino. With the ‘Datapoints.csv’ finally ready and clean, it is selected as input in Grasshopper to recreate the LU network using the prepared x/y coordinates (from 1.2.).

King‘s Cross St. Pancras Regent‘s Park

3.2. GOAL 1: Filter Option From-To

Tottenham Court Road Russel Square

Barbican

Moorgate

Farringdon Holborn

Baker Street Oxford Circus

Using the ‘value list’ component, ‘From’ and ‘To’ options are created to allow individualized journey analysis (GOAL 1). Each value list comprises station names with x/y coordinates. The open-source ‘Shortest Walk’ add-on is downloaded (food4Rhino) and applied. It calculates the shortest route from line start (From) to line end point (To) in a network (the London Underground, see 3.1).

Edware Road

Wetbourne Park

3.3. GOAL 2: Visualisation of Detailed Distribution Curve

Segments which make up the specific, chosen journey are divided by the number of datapoints they have. Comfort indices are imported to Grasshopper and dispatched in such a way that users can pick their comfort type of interest. The relevant values are then assigned as hight (z) to each associated, deconstructed point. Merging the refined nodes into a nurbs curve creates the desired detailed distribution line (GOAL 2). To illustrate the outcome, a gradient coloured mesh surface is constructed that can be baked into Rhino (for exemplary visulisation result see Figure 3). Average, maximum and minimum exposure during the chosen journey are calculated and can be found in the ‘result in numbers’ section.

Royal Oak

Paddington

Ladbroke Grove Latimer Road Goldhawk Road Wood Lane

Shepherd‘s Bush Market

CO2 min exposure 16 ppm

Hammersmith

0

[ppm]

21

Figure 3. Exemplary Distribution Curve Visualised, Comfort Indicator Selected: CO2

14

Bank


CONCLUSION

Closing Thoughts | Discussion

Outlook | Further Development

The newly-developed interactive add-on offers the possibility to explore environmental stress exposure along self-selected tube journeys – both quantitatively and visually. This allows passengers to manage their expectations before boarding the train or to modify route planning in advance according to their comfort thresholds. Hopefully, raising public awareness via the soon-to-be-published website will also increase the efforts of transport planners to curb current problems. Generally, the add-on aims to help draw attention to and communicate the impacts of commuting (with the LU) on people’s health and well-being, which has become an increasingly pressing issue in recent years (Wen et. al. 2020).

The current add-on allows for selecting one indicator at a time. Future versions may show multiple indicators simultaneously through vertically arranged layers or through developing a composite comfort index that allows exploring the interrelations between comfort indices. In addition, it could be interesting to develop an option to visualise only data points that exceed thresholds or a journey optimisation tool to computationally review alternative routes prompting the least congested. Moreover, two-dimensional line graphs could be added. Practically, the data file created as part of this project (‘Datapoints.csv‘) provides the perfect basis for such data visualisation, e.g. directly in Excel.

Other Uses of the Grasshopper Tool | A Contribution to Analytical Design

Additional Note

It is noteworthy that this add-on offers many possible applications beyond this project. Developed based on a flexible input file, the Grasshopper tool can be applied to the processing and visualisation of truly any point data distributed along segments (traffic volumes, emissions, air quality, street tree density, etc.). Transport and urban planners, for example, can benefit from this in the context of evidence-based design efforts. Although this project regards a comparatively simple network, it thus serves as a prototype for modeling point network data in general.

A prerequisite of the course, in which this project was developed was for the result to be a grasshopper tool. In the next step, the code will be embedded in the current website draft. This task is outside the scope of this report but will be addressed by other members of the TubeSurfer Team who specialise in computer sciences. This work was supported by a fellowship of the German Academic Exchange Service (DAAD).

Limitations | Data Collection

Data collection took place on seven working days between mid-January and mid-February 2022 during off-peak times, i.e. between 10am and 4pm in a mixed-methods approach. Each tube line was traveled once or twice. Yet, PM 2.5, to name but one example, is influenced by the time of year and day, train frequency, and passenger density (Cao et. al. 2017, Hickman et. al. 2018). To refine the data quality, peak times and other seasonal measurements should be taken. Further, the user interface could be extended through additional comfort indicators and data exploration features. This may include adding more variable layers to the map to visualise network-related infrastructural data (e.g. tunnel depth, speed, level of service). To provide a holistic overview of London‘s public transport, comfort indices in other tube zones (outside 2/3), overground and bus lines should be measured and included in the analysis.

16

17


REFERENCES Cao, S. et. al. 2017. An investigation of the PM2.5 and NO2 concentrations and their human health impacts in the metro subway system of Suzhou, China. Environmental Science: Processes & Impacts. Vol 19. pp. 666–675.

UCL Department of Civil, Environmental and Geomatic Engineering. 2021. Introducing the ‘Tube Surfers’ team. https://www.ucl.ac.uk/civil-environmental-geomatic-engineering/news/2021/jul/introducing-tube-surfers-team. Accessed: 23.4.2023.

Ebrahimi, S. and Bridgelall, R. 2021. A fuzzy Delphi analytic hierarchy model to rank factors influencing public transit mode choice: A case study. In: Research in Transportation Business and Management. Vol. 39. Elsevier Ltd.

Wen, Y. et. al. 2020. Environmental and Health Effects of Ventilation in Subway, 44 Stations: A Literature Review. In: International Journal of Environmental Research and Public Health. Vol. 17. No. 1084. doi:10.3390/ijerph17031084.

Hickman, A. et. al. 2018. Evaluation of air quality at the Birmingham New Street Railway Station. In: Proceedings of the Institution of Mechanical Engineers. Journal of Rail and Rapid Transit (Part F). Vol. 232. pp. 1864–1878. https://doi.org/10.1177/0954409717752180.

Figures and Tables

All figures and Tables are created by the author.

Imre, S. and Celebi, D. 2017. Measuring Comfort in Public Transport: A case study for Istanbul. In: Transportation Research Procedia. pp. 2441–2449. doi:10.1016/j.trpro.2017.05.261.

Datasources

Comfort Indices measure along the London Underground: ©TubeSurfers Geocoded Tube Station Locations: ©Transport for London Building Data Greater London Area: ©Ordnance Survey ShortestWalk Plugin: ©food4Rhino

Liu, Q. et. al. 2021. Exploring the relationship between the commuting experience and hedonic and eudaimonic well-being. In: Transportation Research. Transport and Environment. Vol. 99. No. 103026. doi:10.1016/j.trd.2021.103026. Redman, L. et. al. 2013. Quality attributes of public transport that attract car users: A research review. In: Transport Policy. Vol. 25. pp. 119–127. doi:10.1016/j.tranpol.2012.11.005.

Thank You

To Nick Tyler Tyler,, Dir Director ector of the UCL Centre Centre for Transport Transport Studies, for intr introducing oducing me to, and all T TubeSurfers ubeSurfers (Jordy (Jordy Adjagboni, Alexander Blakesley, Blakesley, Julia Foellmer Foellmer,, Jan Luca Kastle and Octavio Zamudio Lopez) - for joyfully welcoming me in your team, your inspiration, enthu enthu-siasm and, of course, collecting the data. To Shriyansh Jain, Mathias Jackermeier Jackermeier,, Jakob Smeretschnik Smer tschnig - -for your time and comments - and Dr Dr.. Tasos V Var aroudis oudis - for sharing your expertise in Grasshopper, Grasshopper, constructive feedback and support.

Rojas-Rueda, D. et. al. 2012. Replacing car trips by increasing bike and public transport in the greater Barcelona metropolitan area: A health impact assessment study. In: Environment International. Vol. 49. pp. 100–109. doi:10.1016/j.envint.2012.08.009.

18

19


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.