Management model for decision support when applying low quality waterin irrigation

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited.

In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy Agricultural Water Management 98 (2010) 472–481

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Management model for decision support when applying low quality water in irrigation M. Styczen a,∗ , R.N. Poulsen b , A.K. Falk c , G.H. Jørgensen c a

Soil and Environmental Chemistry, Department of Basic Sciences and Environment, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg C, Denmark Ecology and Environment Department, DHI, Agern allé 5 2970 Hørsholm, Denmark c Water Resources Department, DHI, Agern allé 5, 2970 Hørsholm, Denmark b

a r t i c l e

i n f o

Article history: Available online 19 November 2010 Keywords: Decision support system Crop modelling Low quality water Irrigation management Environmental and health risk

a b s t r a c t Use of low quality water for irrigation of food crops is an important option to secure crop productivity in dry regions, alleviate water scarcity and recycle nutrients, but it requires assessment of adverse effects on health and environment. In the EU-project “SAFIR1 ” a model system was developed that combines irrigation management with risk evaluation, building on research findings from the different research groups in the SAFIR project. The system applies to field scale irrigation management and aims at assisting users in identifying safe modes of irrigation when applying low quality water. The cornerstone in the model system is the deterministic “Plant–Soil–Atmosphere” model DAISY, which simulates crop growth, water and nitrogen dynamics and if required heavy metals and pathogen fate in the soil. The irrigation and fertigation module calculates irrigation and fertigation requirements based on DAISY’s water and nitrogen demands. A Water Source Administration module keeps track of water sources available and their water quality, as well as water treatments, storage, and criteria for selection between different sources. At harvest, the soil concentrations of heavy metals and pathogens are evaluated and the risk to consumers and farmers assessed. Crop profits are calculated, considering fixed and variable costs of input and output. The user can run multiple “what-if” scenarios that include access to different water sources (including wastewater), water treatments, irrigation methods and irrigation and fertilization strategies and evaluate model results in terms of crop yield, water use, fertilizer use, heavy metal accumulation, pathogen exposure and expected profit. The management model system can be used for analysis prior to investments or when preparing a strategy for the season. © 2010 Elsevier B.V. All rights reserved.

1. Introduction In many places around the world, fresh water is a scarce resource. Re-use of wastewater in irrigation is one option to alleviate the scarcity and improve nutrient recycling. However, use of low quality water for irrigation of food crops immediately raises concerns in consumers and authorities administering food quality and health. As a result, the water quality legislation controlling the use of reclaimed water for irrigation is quite strict in some countries. For example, the Italian law 152/06 admits a maximum of 0.1 cfu (colony forming unit) ml−1 Escherichia coli for direct wastewater use. This is considerably stricter than the WHO (2006)recommendations of 6–7 log unit reductions of the E. coli content in

∗ Corresponding author. Tel.: +45 45 41 45 91.

E-mail addresses: styczen@life.ku.dk, merete.styczen@post.cybercity.dk (M. Styczen).

1 Safe and high quality food production using low quality waters and improved irrigation systems and management. Contract-No. FOOD-CT-2005-023168.

0378-3774/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2010.10.017

raw wastewater (108 to 1010 cfu L−1 ), where part of the reduction may take place after application. To obtain such low E. coli-content, extensive and expensive pre-treatment of water is required. On the other hand, wastewater is used for irrigation in countries such as Pakistan (Ensink et al., 2004) and Mexico (Scott et al., 2000) and the resulting produce is important for local food security and as a source of income.

Several models exist that calculate irrigation water requirement, from a number of simple water balance models represented by e.g. CROPWAT (Smith, 1992), over single-field models such as DAISY (Abrahamsen and Hansen, 2000) or SALTMED (Ragab, 2002) to decision support systems, which again range from simple balance models such as Pl@nteInfo® (Jensen et al., 2000), SIMIS (Mateos et al., 2002) or IRRINET (CER, 2009, http://irrigation.altavia.eu/logincer.aspx) to systems that integrate remotely sensed data for precision farming (Pinter et al., 2003).

Fewer systems include nutrients and fertigation, e.g. Anastasiou et al. (2009) or Heiswolf et al. (2010). No models for calculation of crop water requirement were found that aim at incorporating hazards related to the use of wastewater, although SALTMED describes effects of salt on crop growth.

Author's personal copy M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

The hazards associated with wastewater mainly relate to pathogens, heavy metals and organic contaminants. These contaminants may be hazardous to the farmers working with the water and the consumers exposed to the produce. Particularly for heavy metals, accumulation may take place in the soil over time, thus increasing the risk of significant plant uptake (Chang et al., 1995).

WHO (2006) provides general guidelines of how to assess such risks. Furthermore, leaching of contaminants (McBride, 1997) as well as the nutrients added with wastewater is an environmental concern to consider.

In the EU-project “SAFIR” (www.safir4eu.org), summarized

by Plauborg et al. (2010a), it was attempted to combine research findings from the different research groups in an irrigation management model that could aid decision making regarding use of wastewater. The project concerned itself with irrigated tomatoes and potatoes, primary and secondary wastewater (and tap water), pathogens and heavy metals and four different water treatment methods at six field sites. Full irrigation, regulated deficit irrigation and the irrigation method “partial root drying” were investigated. The work with the management model for irrigation with wastewater aimed at covering the combinations investigated in the field. The model system integrates analysis of when to irrigate and fertigate, based on soil water content criteria and assessment of crop nitrogen requirements, and analyses of health and environmental aspects of the applied water. The effect of different treatments of low quality water is simulated.

Furthermore, profit calculations of the different tested scenarios are carried out. Thus, it is possible to evaluate combinations of water sources, water treatment methods, irrigation methods and strategies with respect to water use, hazards and costs. The model system may be used for pre-investment analysis or to evaluate a growing strategy for the next season. An earlier version of the model concept is described in Refsgaard and Styczen (2006).

This article describes the model system that was assembled.

Data obtained in the project has been used to derive the relationships and to calibrate sub-models; as yet the model system has not been validated on sites with independent data.

473

2. Theory and model description The model system, which is described in detail in Styczen et al.

(2009) consists of (Fig. 1): (1) The Plant–Soil–Atmosphere model (PSA-model), DAISY (Abrahamsen and Hansen, 2000; Hansen et al., 1990). This model was further developed under the SAFIR-project, as described in Plauborg et al. (2010b), (2) an Irrigation and Fertigation Strategy module (IF-module), (3) a Water Source Administration module (WSA-module), (4) a Risk Assessment module (RA-module) and (5) an Economy module.

The PSA-model is linked to the IF-module and to the WSAmodule via an OpenMI-interface (Gijsbers, 2004), which is a standardized framework for linking environment-related models.

OpenMI allows the user to let the prediction of one model depend on the state predicted by another model. An overview of the model system is given in Fig. 1.

The cornerstone in the model system is the “Plant–Soil–Atmosphere” (PSA)-model. It simulates water, carbon and nitrogen dynamics (including crop growth) and if required, heavy metals and pathogen fate in the soil. The IF-module repeatedly questions the PSA model for its water and nitrogen demands.

In turn the IF-module requests water from the WSA-module, which keeps track of available water sources and their water quality, water treatments, storage, and criteria for selection between different sources. After receiving water of a certain quality from the WSA-module, the IF-module supplies water and nitrogen back to the PSA, according to a defined irrigation strategy. At harvest, the soil concentrations of heavy metals and pathogens are calculated and the risk to consumers and farmers assessed. Crop profits are assessed, considering fixed and variable costs of input and output.

Using a Microsoft Excel result presentation, users get access to key performance figures from the model simulation, including crop

Fig. 1. Overview of the management model system made within the SAFIR project. The Plant–Soil–Atmosphere model used in the project is DAISY (Abrahamsen and Hansen, 2000).

Author's personal copy 474

M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

yields, water use, risk assessments and economy, but users may also retrieve detailed model outputs.

2.1. The plant soil atmosphere model The PSA-model code (DAISY) is calibrated for the crop and site to be modelled. DAISY calculates water, nitrogen and carbon dynamics in the plant and soil. It allows several different process descriptions for water flow, evapo-transpiration, crop growth and solute transport. In total it is able to simulate about 100 different processes with about 200 process models, implying that different process models are available for some of the processes. Under the SAFIR project, DAISY was extended with the possibility of two dimensional (rather than one-dimensional) simulations, and a framework for modelling soil vegetation atmosphere transfer (SVAT) was added (Plauborg et al., 2010b; Ragab, 2009-Annex 3.1, 3.3 and 3.6). This means that the exchanges of sensible heat, water vapour, and CO2 between the canopy, soil, and atmosphere are simulated. Stomata opening is a function of various parameters, including abscisic acid in xylem sap, which is generated in the root system as a function of the water uptake and pressure potential in the soil. This process is particularly important when describing “partial root drying”. The SVAT model requires hourly weather data (precipitation, global radiation, diffuse radiation (if available), wind speed, vapour pressure, and air temperature). While most sprinkler and drip-irrigation applications can be simulated using the one-dimensional model, “partial root drying” with irrigation on alternate sides of the plant is possible only with a two-dimensional model.

DAISY solves Richards’s equation, using data of the hydraulic properties of the soil. Macropore flow may be specified. Furthermore, data on texture, organic matter, and N-content are required.

DAISY is equipped with a number of parameterised crop models, but the SAFIR project only dealt with potatoes and tomatoes. The potato crop model used was based on Heidmann et al. (2008) and updated to the new SVAT-model, while the tomato models (fresh and processed tomatoes) were developed during the SAFIR project (not published).

DAISY allows specification of whether the irrigation water is applied by sprinkler or by drip irrigation to the surface or at a given depth below the surface. Furthermore, tillage and fertilizer or manure-additions are specified. It calculates organic matter turnover, typically including six organic pools, and the mineral nitrogen balance, including deposition, fertilization, ammonification, immobilization, nitrification, denitrification, plant uptake, and leaching. Through the OpenMI-interface, DAISY links to the IFmodule.

The PSA-model supplies: -

Water content in the root zone (mm).

Water content at field capacity (mm).

Water content at wilting point (mm).

Critical N-content in the crop (kg ha−1 ).

Potential N-content in the crop (kg ha−1 ).

Actual N-content in the crop (kg ha−1 ).

DAISY’s crop development stage.

The PSA-model receives:

- Irrigation water to apply in the next time-step (hourly) (mm).

- Nitrogen content of irrigation water, distributed on ammonia-N and nitrate-N (mg L−1 ).

- Heavy metal content (one or more constituent) of irrigation water (mg L−1 ).

- Pathogen content of the irrigation water (mg L−1 ). The E. coli concentration is used as pathogen indicator. The normal unit for E. coli would be cfu per l (or in guidelines or legislation per ml or 100 ml),

which DAISY does not handle. To move from cfu L−1 to mg L−1 a conversion factor of 10−9 mg cfu−1 is used.

DAISY models were calibrated for the Danish (Plauborg et al., 2010b), Italian (A. Battilani, pers. com, 2009), Serbian, and Cretan (F. Plauborg pers. com, 2009) field trial sites in the SAFIR project and were therefore available for the management model work.

2.2. The irrigation and fertigation strategy module The IF-module receives information from the PSA-model about crop development stage, soil water content and nitrogen content.

Based on this and the user defined irrigation and fertigation strategy, the IF-module calculates how much water should be supplied to the PSA-model and if fertigation should be added. The IF-module passes on the water request to the WSA module, which abstracts water from its defined sources, if available.

Irrigation can be based on calculations of the relative water content (RWC) in the root zone or as a depth of irrigation water. The relative water content is defined as ( act − wp )/( fc − wp ), where act is the actual water content, wp is the water content at the wilting point and fc is the water content at field capacity. In the prescribed irrigation scheme option, the irrigation depth (in mm) is specified as a time series. If the demand cannot be fulfilled in one time step (due to limiting factors such as irrigation system capacity or water availability) the remaining demand is requested during the next time step. If irrigation is based on relative root zone water content, irrigation is triggered when the relative water content in the root zone, simulated by the model reaches a user defined lower limit (threshold). This lower limit, specifying the allowable depletion, depends on the crop and its development stage and is tabulated. A similar table relates upper limit values and crop development stages and is used to define the threshold for stopping irrigation. A value of one indicates that irrigation continues until field capacity is reached. The depth of the root zone is defined by the simulated root development at any given time. However, just after planting, irrigation requirement is calculated for 20 cm depth, even if roots are shallower.

The IF-module operates with four crop development stages as defined by Doorenbos and Pruitt (1975), and consequently also four irrigation stages, defined for each crop, i.e. potatoes, processing tomatoes and fresh tomatoes. The irrigation stages are linked to DAISY growth stages, which in turn are linked to other types of development as for instance the root depth, and flower and fruit formation. The irrigation start and stop trigger values used can be defined separately for the four stages, depending on the irrigation strategy (full irrigation, deficit irrigation, partial root drying), the irrigation method (sprinkler, drip) and the crop (potatoes, fresh tomatoes, processing tomatoes). Fig. 2 illustrates this principle for two irrigation cases.

The threshold values for initiating and ending irrigation as a function of crop, irrigation method, and irrigation strategy, were determined partly through discussions with SAFIR participants and partly from a study of the measurements of soil water in the experimental plots (field and laboratory) in combination with the soil retention properties. In addition to an upper and lower threshold, “partial root drying” requires a threshold for when to change irrigation from one side of the plant to the other. The findings from the SAFIR field experiments are summarized in Jensen et al. (2009) and tables of threshold values for soil water suction are available in Styczen et al. (2009).

Fertigation is application of nutrients dissolved in water through an irrigation system. This method may be combined with e.g. an initial application of fertilizer at planting. In the model system, fertigation is based on the N-status in the crop. Phosphorus

Author's personal copy M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

475

Fig. 2. Schematic illustration of upper and lower thresholds that change over the season. For sprinkler irrigation, the difference between the upper and lower threshold is large while for drip irrigation, the difference is smaller. In the example “RWCUpper” is the threshold at which irrigation stops, while RWCLower-1 and RWCLower-2 are the thresholds at which irrigation is initiated for sprinkler and drip-irrigation, respectively. RWCAct-1 and RWCAct-2 shows development in actual relative water content in the root zone as a consequence of irrigation simulated for the two cases.

and other nutrients are not considered because their dynamics are not described in DAISY. However, the application of phosphorus or other selected nutrients dissolved in irrigation water can be accumulated and logged by the IF-module and the total application at a given time can thus be compared to crop requirements.

DAISY calculates N-uptake and distribution in leaves, stem, roots and storage organs (Hansen and Abrahamsen, 2009). A certain concentration of N is required in each of these organs to allow optimal growth. At any time, these critical concentrations can be multiplied with the amount of dry matter in each organ and summed up to produce the critical nitrogen content of the plant (NCropCr ).

Similarly, a maximum concentration is defined for each organ, from which the potential nitrogen content in the plant (NCropPt ) can be calculated. The difference between the two may be interpreted as a luxury uptake or a store of N that the plant can draw on later in the growth period. In addition, the simulated “actual” N-content in the plant (NCropact ) can be calculated and compared to the critical and potential content. If NCropact lies above NCropCr the crop is growing optimally, and so the strategy is simply to try to fulfil this requirement. Below NCropCr , the plant suffers from N-stress and the growth is retarded.

The analysis carried out by the IF-module to assess whether the plants will require addition of N through fertigation to ensure the nutritional status till next irrigation is the following: If NCropAct > NCropCr + NCropCr ∗ Ndays ⇒ no action is required, If NCropAct ≤ NCropCr + NCropCr ∗ Ndays ⇒ initiate fertigation.

NCropAct and NCropCr are calculated as kg N ha−1 . CropNCr is the daily nitrogen-requirement that will keep the crop at the critical nitrogen level. To obtain this value, the model evaluates the daily changes in NCropCr and uses them for the extrapolation. Ndays is the typical time space between two irrigations. For many drip irrigation installations this will be 1 or 2 days. The selected strategy is most appropriate for relatively frequent irrigations; if Ndays become large, the ability to predict the crop requirement becomes poorer.

The amount of nitrogen to be added is calculated as: Addition = NCropCr − NCropAct + NCropCr ∗ Ndays + “Security factor” The security factor is defined as a user specified amount of nitrogen (kg ha−1 ). It can decrease during the growing season, e.g. starting with 5 and ending with 0 kg ha−1 . Fig. 4D shows an exam-

ple of the development in crop nitrogen status during the growing season, using these principles.

If rainfall occurs there may be instances where fertigation is required although the relative soil water content is adequate. This can potentially lead to over-irrigation. In the model, the demand for nitrogen can trigger irrigation with the minimum quantity of water required to dissolve the fertilizer and avoid damage to roots, or for sprinkler irrigation, to avoid damage to the crop foliage. If the irrigation water already contains nitrogen, this initial content of N in the water is taken into account before dosing the fertigation solution. Still, a specified maximum content of N cannot be exceeded.

2.3. The Water Source Administration module The WSA-module defines the water sources available for irrigation (typically groundwater, river water or different waste water) as well as certain water treatment methods. Reservoirs for storage of water may be added in the model between the source and the field to be irrigated. A flow source is characterised by its flow rate (time series) and concentrations of chemicals that the model system is desired to handle. A water treatment partly or fully removes selected chemical species or pathogens. It requires a flow capacity, and a factor specifying a “pass through rate” for the chemical.

Simple chemical reactions can be defined as well. Water quality for primary and secondary wastewater used in test simulations was based on measurements at the different SAFIR experimental sites, particularly Italy and Crete. The treatments that were parameterised based on results generated in the SAFIR project (Battilani et al., 2010) were a gravel filter, a UV-lamp, and a heavy metal removal device (a high porosity adsorber matrix based on ferric hydroxides) applied to secondary wastewater and a small-scale compact pressurized Membrane Bioreactor, applied to primary wastewater.

A reservoir requests water from an upstream source until it is full (or reaches a defined level), regardless of the current irrigation demand. It is characterised by volume, outflow capacity, evaporation of water and decay of pathogens. While chemical reactions may take place in reservoirs, particularly if contaminated water is stored, this has not been included in the management model. Furthermore, data from the project did not allow a parameterisation of such reactions.

From each of these elements or combined elements, time series of the abstracted flow as well as concentrations of nutrients and, if required, heavy metals and E. coli are generated and logged. A request for water from the PSA via the IF-module can only be fulfilled if the flow of water though the system is sufficient.

Author's personal copy 476

M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

Table 1 Die-off rates for E. coli on tomatoes as a function of temperature. The reduction factor scale assumes that the rate is given at 20 ◦ C. T (◦ C) Reduction factor (T0 = 20)

5 0.25

Days−1 Days

k(mod) T90

0.1439 16

Reduction per day

%

13.4

10 0.5 0.2878 8 25.0

2.4. Risk assessment related to pathogens

If t < 0: ft = 0.

If 0 ≤ t ≤ 20: ft = 0.05 × t If t > 20: ft = 0.1732 × e(0.07871×t) + 0.1624 The formula used for calculation of the number of E. coli present on the fruit surface at a given time if 1 mm of water stays on the fruit after each irrigation is Not =

(Ct /(1 × 10−9 )) 1000

20 1

0.4317 5.3 35.1

0.5756 4 43.8

25 1.4

30 2

0.8059 2.9 55.3

1.1513 2 68.4

mm expected to stay on the crop, the contamination at harvest is calculated and incorporated into the risk assessment.

For potatoes or tomatoes lying on or in the soil, contamination is related to the concentration of E. coli in the soil. In order to estimate the concentration in the soil, E. coli is added with irrigation water to the DAISY-model and is subject to die-off and filtration.

Die-off in soil was parameterised based on the review by Ensink and Fletcher (2009) as a function of temperature and water content.

The water content dependency is described as a stepwise linear function (Table 2). The calculated rates are used in a first order decay-function.

DAISY describes filtration of colloids in soil as a first order reaction, where the coefficient depends on the geometry of the matrix, the particle size, the flow velocity, the electrolytic composition of the water and the surface potential on particles and pore surfaces. Filtration in the soil matrix (micropores) is described as a 1.order reaction, where the filtration coefficient can be expressed as the colloid deposition rate-coefficient (s−1 ) divided by the velocity of colloid particles in the porous medium (m s−1 ), similar to the MACRO-model (Jarvis, 1994).

The treatment of pathogens follows to a large extent the approaches used by WHO (2006). E. coli is used as an indicator organism and other pathogens are expected to follow the presence of E. coli in a certain ratio. Until application on the plants or soil, E. coli numbers are calculated in the WSA module as described previously. In case of sprinkler irrigation of supported tomatoes, the risk calculation is carried out as post processing based on the concentration of E. coli in applied irrigation water and the air temperature. Based on Ensink and Fletcher (2009), Feachem et al.

(1983) and Bell and Bole (1978), the relationship between temperature and E. coli die-off on plants was estimated as shown in Table 1.

The temperature factor (ft ) is parameterised as:

15 0.75

F = fc × v × c × ,

where F is the filtration (g m−3 h−1 ), fc is a reference filtration coefficient (m−1 ), c is the concentration of the colloid in question (g m−3 ), is the water content (m3 m−3 ) and v is the pore water velocity (m h−1 ). DAISY includes two micropore domains allowing two different fc -values and no filtration in the macropores (Abrahamsen, 2010).

Using parameter values found by other authors (Baun et al., 2007; Jarvis et al., 1999; McGechan et al., 2002; Villholth et al., 2000), the filter coefficient in micropores was estimated to be in the order of 40–100 m−1 for colloid size-particles. This equals logreduction values of 1.6–2 per m. The particles considered as colloids were 0.2 ␮m in the study by Jarvis et al. (1999) and 0.02 ␮m for Baun et al. (2007), indicating that higher values may be appropriate for larger colloids such as bacteria.

+ Not−1 × ekft t

where Not is the number of E. coli in on the fruit surface (cfu), at the time t under the assumptions given above, Ct is the concentration of E. coli in irrigation water in (mg L−1 ) added at the time t (this figure is divided by the weight of a cell (cfu mg−1 ) and by 1000 to obtain cfu ml−1 ), k is the rate of decay in h−1 , ft is the temperature modification factor and t is the time step of 1 h used in the DAISY output files. The calculation sums the E. coli added at each irrigation event taking into account die-off, and the calculation is carried out from the first irrigation, although no fruit may be present. The high die-off rates ensure that the most recent irrigations dominate the calculations. Depending on the number of

Table 2 Die-off rates for E. coli in soil as a function of temperature and suction, expressed as T90-values and die-off rate coefficients.

Soil (pF)

Factorm

0 1 2 3 4 5 6 7

0.6 0.8 1 3 5 7 9 11

0 1 2 3 4 5 6 7

0.6 0.8 1 3 5 7 9 11

Temp. (◦ C) ft

0 0

5 0.25

10 0.5

15 0.75

20 1

25 1.4

T90 166.7 125.0 100.0 33.3 20.0 14.3 11.1

83.3 62.5 50.0 16.7 10.0 7.1 5.6

55.6 41.7 33.3 11.1 6.7 4.8 3.7

41.7 31.3 25.0 8.3 5.0 3.6 2.8

29.8 22.3 17.9 6.0 3.6 2.6 2.0

0.0276 0.0368 0.0461 0.1382 0.2303 0.3224 0.4145 0.5066

0.0414 0.0553 0.0691 0.2072 0.3454 0.4835 0.6217 0.7598

0.0553 0.0737 0.0921 0.2763 0.4605 0.6447 0.8289 1.0131

0.0774 0.1032 0.1289 0.3868 0.6447 0.9026 1.1605 1.4183

Die-off rates (day s−1 ) 0.0000 0.0138 0.0000 0.0184 0.0000 0.0230 0.0000 0.0691 0.0000 0.1151 0.0000 0.1612 0.0000 0.2072 0.0000 0.2533

Author's personal copy M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

The topsoil concentration of E. coli at the time of harvest is used for assessing the contamination of tomatoes lying on the ground.

The basis for calculation of the contamination of potatoes is the content of E. coli at the depth of the potatoes at the time of harvest.

The actual risk assessment for consumers follows the WHO (2006) guidelines for Quantitative Microbial Risk Analysis (QMRA), using dose response curves to estimate the disease risk when the exposure is known. The concentrations and exposure figures are transferred to Excel spreadsheets, pre-programmed with the QMRA analysis and parameterised with default figures related to amounts of water or soil sticking to the produce (default set to 1–1.5 ml of water per irrigation on 100 g of tomato, 5–10 mg soil/100 g tomatoes on the ground and 10–50 mg soil/100 g potatoes) as well as consumption of the crop in a particular country and disease related parameters for rotavirus, campylobacter, cryptosporidium and giardia. All results of the risk analysis should be less than 10−3 to represent an acceptable risk.

The estimation of risk to farmers when using contaminated water follows the same principles, except the value used for risk assessment is the range between median and maximum E. coli soil concentration (g ha−1 ) in the topsoil extracted for the growing season, i.e. from sowing to harvest. These values are transferred and converted into cfu before they are included in the QMRA. 2.5. Risk assessment related to heavy metals Like pathogens, heavy metals follow the water flow from the water source to the field through water treatments or other defined elements in the WSA-module, as described earlier. When irrigation water contaminated with heavy metal ions is supplied to the soil, the subsequent fate is calculated by DAISY using a Freundlich sorption isotherm. The approach was adopted after geochemical modelling of Pb at the SAFIR experimental site in Crete (Pettenati et al., 2009) showed that this provided a good approximation of the data. The geochemical modelling also showed that parameters for the lead isotherm depended on soil mineralogy, pH and constituents of the irrigation water. The use of sorption isotherms for assessment of accumulation of heavy metals is not new, although linear partition coefficients (Kd = mass of adsorbate sorbed/mass of adsorbate in solution) are more commonly used. Allison and Allison (2005) have provided a comprehensive review of linear partition coefficients for metals in surface water, soil and waste.

The model system considers five values for risk assessment of indicated heavy metals: 1.

2.

3.

4.

5.

The concentration in the irrigation water.

The initial concentration in the soil.

The final concentration in the soil.

The increase in concentration due to irrigation with waste water.

Concentration in leaching water, if relevant.

The concentration in irrigation water is evaluated in relation to the environmental thresholds given in Table 3. If the concentration of a heavy metal in irrigation water is below the limit for prolonged use, it is considered safe, if the concentration is between the limit for prolonged use and acute toxicity, caution is required.

Concentrations above the limit for acute toxicity are considered dangerous.

At the end of the simulation, the content in the root zone is evaluated. Somewhat arbitrarily, the use is considered safe if the soil concentration is less than 70% of the maximum tolerable soil concentration. Caution is required if the concentration is rising to between 70 and 100% of the maximum tolerable soil concentration, and particularly in this range, it is important to look at the increase in concentration over a growth season. The use is unsafe if the concentration exceeds the maximum tolerable soil concentration.

477

Table 3 Limiting contents of a number of chemical elements in irrigation water and soil based on a literature review carried out by A. Battilani (pers. com, 2009). The figures are a combination of WHO (2006) guidelines and other guidelines, and the resulting figures are either equal to or stricter that the WHO (2006)-recommendations.

Element

Irrigation water

Soil

Prolonged use (mg l−1 )

Acute phytotoxicity (mg l−1 )

Concentration in soil (mg kg−1 )

Arsenic B

0.1 0.5

8.0 nd

Cd Cu Cr(VI) Fe Mn Hg Mo Ni Pb Zn

0.01 0.2 0.1 0.2 0.2 0.002 0.01 0.2 2.0 2.0

2.0 Crop dependent, 0.5->6 0.05 5.0 1.0 20.0 10.0 0.002 0.05 2.0 5.0 5.0

1.0 150.0 nd nd nd 1.0 nd 50.0 84.0 150.0

In some cases, the concentration of heavy metal in infiltrating water or drain water may pose a problem. The concentration in leaching water is therefore presented graphically. However, if the simulation only covers the growth season, the amount of water leaching may be insignificant and not really represent average conditions for a year.

2.6. Profit calculation The model system provides profit calculations as postprocessing. The calculation, which is carried out in Microsoft Excel is simple; in short, the value of the crop yield is calculated as a product of the quantity produced and a (time-varying) price, and compared to the fixed and variable costs involved in irrigation and fertigation.

The inputs required are • The area irrigated.

• Fixed costs and costs per m3 related to each water source.

• Fixed costs and costs per kg N related to fertilizer and fertigation solution, including cost of application and depreciation of relevant equipment.

• Price of the harvested crop, which typically varies over time.

With respect to water sources, the fixed costs may be depreciation of equipment of different types; while the cost per m3 could be a cost paid to the water authorities or be related to water treatment, energy or labour cost. The user can specify prices and costs; however, figures implemented in the supporting tables stem from the SAFIR field sites and are reported in Pedersen (2009). The amount of water used per source is calculated in the WSA-module, while the amount of N used is available from the IF-module.

Typically, the price of a crop depends on the quality of the crop.

A fraction of the crop has to be allocated to each quality class as the model does not estimate quality as a function of irrigation and fertigation practices.

Particularly the price of fresh tomatoes varies with availability.

For processing tomatoes the farmer usually has a contract with a fixed price. For potatoes, the variation depends on the market.

2.7. Result presentation An Excel result presentation was designed as the immediate user interface for extracting and presenting model results. The result presentation consists of a number of sheets, attributed to data

Author's personal copy 478

M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

Fig. 3. Screen dump of the presentation of main results in the DSS. Values related to risk assessment are shown in green, yellow or red in order to ease the interpretation.

extraction, summarizing main results, predefined time series and accumulated plots, intermediate calculations and access to “raw” model results.

The result presentation requires the user to execute a single data processing routine, which runs a number of macro’s that extracts results from the model system and runs the risk assessment calculations whereby the result presentation is populated automatically.

Subsequently, the user can view a tabulated summary of the main results as shown in Fig. 3 as well as a range of different predefined plots. Using Excel, the experienced model user have flexibility to customize and add own presentations as extensive model output logging is available in DAISY. Fig. 4 shows a selection of the predefined plots in the result presentation. The figure shows a simplified hypothetical case of regulated deficit irrigation in potato with subsurface drips, with access to clean water and secondary waste water (individual water sources not shown), but without rainfall, and a fixed groundwater table 150 cm below the surface on a clay soil. Fertilizer was not applied at sowing. The minimum irrigation frequency and the fertilizer demand forecast are set to 5 days. Plot A shows the rainfall and applied irrigation water during the growth season. Plot B shows the development in the relative soil water content, also indicating the changing start and stop thresholds throughout the growing season of the crop. Plot C illustrates the applied amount of nitrogen fertilizer and the actual content of nitrate-N and ammonium-N in the soil, but also the quick uptake of nitrate-N and the slow transformation of ammonium, while plot D shows the resulting nitrogen status in the crop.

3. Discussion Considering the problems of scarce water resources in many parts of the world and the benefits of recycling of nutrients it is worth evaluating how low quality water can be used safely in food production. It is thought that a model system that can analyse the consequences of application of low quality water for crop production as well as with respect to health and environment, taking into account the local climate, crops, soils, irrigation techniques, agricultural practices, and water quality, can provide local decision makers, farmers, and consumers with a better basis for judging where, when, and how to utilize this resource. With the possibility of evaluating effects of restrictions such as application of water treatments, restrictions of application time, or restrictions of waste water amounts, safe scenarios can be identified and used as basis for local implementation.

As with all model systems, the quality of the predictions depends on proper calibration and validation of the implemented model. It means that the PSA-model has to be calibrated to local crop and soil conditions in order to provide good estimates of water and nitrogen requirement during the growing season and the resulting yields. Similarly, local data for waste water quality, sorption of heavy metals, prices, etc. are required for the predictions to be credible. Use of the model system thus requires an initial investment in local adaptation of the system by an expert user. However, as water quality and water treatments are user defined, a library of default values can be included in the database of the WSA-module over time.

Author's personal copy M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

479

Fig. 4. Selection of the predefined plots in the result presentation based on a simplified hypothetical case of regulated deficit drip-irrigation in potato. For details, refer to the text. Plot A shows the rainfall and applied irrigation water during the growth season. Plot B shows the development of the relative soil moisture content, also indicating the changing start and stop trigger values throughout the growing season of the crop. Plot C shows the applied amount of nitrogen fertilizer-N and the actual content of nitrateand ammonium-N in the root zone, but also the quick uptake of nitrate-N and the slower transformation of ammonium. Plot D shows potential (NCropPt), actual (NCropAct) and critical nitrogen (NCropCr) content in the plants during simulation. The actual N-content is kept above, but close to the critical nitrogen content during the simulation.

DAISY, the PSA-model code applied here, is an advanced model system in itself and the process models used in the project requires considerable skill and good data, which in many cases will be seen as a drawback to general use. DAISY can be applied in a onedimensional version, with a transpiration calculation that depends on reference evaporation, root distribution and soil water content (not including abscisic acid), for ordinary sprinkler or drip irrigation, but not for “partial root drying”, which requires two dimensions and simulation of abscisic acid production in roots.

The simpler transpiration calculation requires daily weather data and has been parameterised for several crops. The models for Ndynamics and contaminant transport are independent of the choice of transpiration model.

When the complete model system has been parameterised by expert users for soils, crops and climate in a given area, scenarios can be run by computer literate agricultural engineers or similar.

The end users of simulation results would be agricultural advisory staff and farmers, but also administrators’ deciding on acceptable practices for the use of wastewater.

The joint management of water and N, focussing on both crop growth and environmental effects is important. Managing the NCropAct – value to closely follow the course of NCropCr , particularly towards the end of the growing seasons should optimize fertilization and minimize nitrate in the soil at final crop harvest, which should reduce the risk of leaching. However, if wastewater is applied all through the growing season excess nitrate- and ammonia-N may build up, increasing the risk of leaching. From

the result presentation, the user can evaluate when a shift from wastewater to a water source with less nitrogen would be appropriate.

The combination of irrigation modelling and detailed modelling of pathogens in soil is, to our knowledge, new. The processes selected for E. coli (die-off and filtration) mirror to a large extent the findings from field studies as discussed in Forslund et al. (2010).

In the model, E. coli can move downwards in large pores with saturated flow, while its movement is severely restricted in smaller pores. A consequence of this is that E. coli cannot move upwards with evaporation from a buried drip-point to the surface. Subsurface drip thus reduces the contamination of above-ground plant material substantially. On the other hand, die-off rates may be slower below the surface due to higher soil water content and lower temperatures than at the surface.

At present, the risk assessment scheme assumes that other pathogens follow E. coli in a fixed ratio. If parameters for filtration, sorption and die-off are known for other pathogens, they may be simulated separately in DAISY. The evaluation of pathogen risk in the management model system is suitable for fresh tomatoes, but too conservative for processing tomatoes and potatoes. Processing tomatoes are washed and heated, which reduces the contamination that may reach the consumer. Peeling of potatoes decreases the contamination by two E. coli log-units and boiling by six to seven E. coli log-units, so the actual risk when eating the potato is extremely small and mainly related to cross contamination during handling of the potatoes.

Author's personal copy 480

M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481

The developed system is quite flexible in the sense that it can be adapted to local requirements. It can be used to estimate irrigation/fertigation alone or expanded to include heavy metals and pathogens as long as these can be parameterised. The profit calculation can be adjusted by the user if required. For example, the cost of washing produce could be added if the produce is pathogen contaminated, or the saved cost of P-fertilizer due to the use of wastewater could be included.

There are several options for expanding the capabilities of the model system. An obvious choice would be to generate parameters for the SVAT-growth model and the irrigation–fertigation module for more crops. In particular vegetable crops would be of interest because they are often irrigated due to quite high crop water requirements and also often eaten raw or without much preparation by consumers.

In addition, the model system could be developed into an online forecast system, where the model is updated daily with actual rainfall and irrigation/fertigation actions and e.g. a 5-day weather forecast. As the DAISY model is able to hot-start from a saved resultfile, there is no serious technical problem in doing so, except for an extra layer of complexity for the user. However, development of a shell that keeps track of model runs, result files and weather forecasts is rather expensive and requires local interest and agreement with a relevant supplier of forecast data.

The target group for an on-line system, where soil, crop and irrigation system information is implemented at the start of the season would ideally be farmers, but it requires computer-literacy and time which farmers cannot be expected to have. Therefore agricultural consultants are a more realistic target group. On-line modelling of field trials, which in many places are the foundation of the agricultural consultants’ advice, could, for example, be an interesting option. Near real time model outputs could provide supplementary information on which agricultural consultants could refine their advice. Design of such an online system requires a close dialogue and collaboration with the future users in order to ensure the model system provides relevant and user-friendly information as also highlighted by Jørgensen et al. (2007).

4. Conclusion The management model developed is able to schedule irrigation and fertigation according to modelled soil water conditions and plant N-status. In addition it includes risk assessments for pathogen exposure and heavy metal contamination when low quality water is used for irrigation, assisting the user in assembling safe schemes of wastewater uses.

The model system is so far an expert system with a high degree of complexity, but also flexibility to adapt the model system to specific cases. However, a customized system can be applied by trained agricultural consultants.

Acknowledgements The management model for decision support is developed within the SAFIR project funded by EU (Contract-No. FOODCT-2005-023168) as part of the sixth Framework Programme (2002–2006). The DSS design and development builds on discussions with and information from all partners in the project. In particular thanks to project co-ordinator Finn Plauborg, Faculty of Agricultural Sciences at Århus University, Denmark, Adriano Battilani, Consorzio di Bonifica di secondo grado per il Canale Emilliano Romagnolo, Italy, Søren Hansen and Per Abrahamsen, Faculty of Life Sciences, University of Copenhagen, Denmark, Wolfgang Kloppmann and co-workers at Bureau de Researches Géologique et Minières, France, Jeroen Ensink and Tony Fletcher at London School

of Hygiene and Tropical Medicine, UK and Søren Petersen and Jens Erik Ørum at Food and Resource Economics Institute, University of Copenhagen, Denmark who all have contributed significantly to the concept and data included in model system.

References Abrahamsen, P., 2010. Daisy Program Reference Manual, Version 4.89. URL: http://www.Daisy-model.org/ (accessed on 7/9 2010).

Abrahamsen, P., Hansen, S., 2000. Daisy: an open soil–crop–atmosphere system model. Environ. Model. Softw. 15, 313–330.

Allison, J.D., Allison, T.L., 2005. Partition Coefficients for Metals in Surface Water, Soil and Waste. EPA/600/R-05/074, U.S. Environmental Protection Agency, Office of Research and Development, Washington, DC 20460. 93 p.

Anastasiou, A., Savvas, D., Pasgianos, G., Sigrimis, N., Stangellini, C., Kempkes, F.L.K., 2009. Decision support for optimised irrigation scheduling. In: Tüzel, Y., Öztekin, G.B., Meric, M.K. (Eds.), ISHS Acta Horticulturae 807: International Symposium on Strategies Towards Sustainability of Protected Cultivation in Mild Winter Climate, pp. 253–258.

Battilani, A., Steiner, M., Andersen, M., Back, S.N., Lorenzen, J., Schweitzer, A., Dalsgaard, a., Forslund, A., Gola, S., Klopmann, W., Plauborg, F., Andersen, M.N., 2010.

Decentralised water and wastewater treatment technologies to produce functional water for irrigation. Agric. Water Manage. 98, 385–402.

Baun, D.L., Styczen, M., Lønborg, M.J., Holm, J., Clausen, T., Grøn, C., Koch, C.B., Gjettermann, B., Petersen, C.T., Spliid, N.H., 2007. Kolloid-faciliteret transport af glyphosat og pendimethalin. Kvantificering og modellering. Bekæmpelsesmiddelforskning fra Miljøstyrelsen, nr. 107. Danish Environmental Protection Agency, Copenhagen, (in Danish) 268 p.

Bell, R.G., Bole, J.B., 1978. Elimination of faecal coliform bacteria from soil irrigated with municipal sewage lagoon effluent. J. Environ. Qual. 7, 193–196.

Chang, A.C., Page, A.L., Asano, T., 1995. Developing Human Health-related Chemical Guidelines for Recycled Wastewater and Sewage Sludge Applications in Agriculture. World Health Organization, Geneva, 114 p.

Consorzio 2de Bonifica di second grado per il Canale Emilliano Romagnolo (CER). IRRINET: IT Irrigation Advisory Services for Farm Water Management. Description available on http://www2.altavianet.it/irriweb2/Docs/CER/ rrinet testo documentazione english version.pdf, access to the service at http://www.consorziocer.it/ (accessed on 11.10 2010).

Doorenbos, J., Pruitt, W.O., 1975. Guidelines for Predicting Crop Water Requirements, FAO Irrigation and Drainage Paper No. 24, FAO, Rome (Italy). Land and Water Development Div., 193 p.

Ensink, J.H.J., Mahmood, T., van der Hoek, W., Raschid-Sally, L., Amerasinghe, F.P., 2004. A nationwide assessment of wastewater use in Pakistan: an obscure activity or a vitally important one? Water Policy 6, 197–206.

Ensink, J., Fletcher, T., 2009. Survival and Transport of Helminth Eggs and Faecial Coliforms in Soil and Agricultural Produce. Deliverable 5.4 of the SAFIR Project.

Available at: www.safir4eu.org (accessed on 11.10 2010). London School of Hygiene and Tropical Medicine, UK, 9 p.

Feachem, R., Bradley, D., Garelick, H., Mara, D.D., 1983. Sanitation and Disease: Health Aspects of Excreta and Wastewater Management. John Wiley & Sons, Chichester, UK/New York, NY, USA, 501 p.

Forslund, A., Ensink, J.H.J., Battilani, A., Kljujev, I., Gola, S., Raicevic, V., Jovanovic, Z., Stikic, R., Sandei, L., Fletcher, T., Dalsgaard, A., 2010. Faecal contamination and hygiene aspect associated with the use of treated wastewater and canal water for irrigation of potatoes. Agric. Water Manage. 98, 440–450.

Gijsbers, P.J.A., 2004. In: Liong, Phoon, Babovic (Eds.), The OpenMI Architecture – Details, 6th International Conference on Hydroinformatics. World Scientific Publishing Co. Pte. Ltd., Singapore, pp. 1819–1826.

Hansen, S., Jensen, H.E., Nielsen, N.E., Svendsen, H., 1990. Daisy: Soil Plant Atmosphere System Model. NPO Report No. A 10. Danish Environmental Protection Agency, Copenhagen, 272 p.

Hansen, S., Abrahamsen, P., 2009. Modeling water and nitrogen uptake using a single-root concept. Exemplified by the use in the Daisy model. In: Ma, L., Ahuja, L.R., Bruulsema, T. (Eds.), Quantifying and Understanding Plant Nitrogen Uptake for Systems Modelling. CRC Press, Florida, USA, pp. 169–195.

Heidmann, T., Tofteng, C., Abrahamsen, P., Plauborg, F., Hansen, S., Battilani, A., Coutinho, J., Doleˇzal, F., Mazurczyk, W., Ruiz, J.D.R., Takáˇc, J., Vacek, J., 2008.

Calibration procedure for a potato crop growth model using information from across Europe. Ecol. Model. 211, 209–223.

Heiswolf, S., Wright, R., Moody, P., Pattison, T., 2010. Vegetable production in the dry tropics – nutrient and soil management strategies from Queensland, Australia.

In: Larsen, R.U. (Ed.), ISHS Acta Horticulturae 852: IV International Symposium on Ecologically Sound Fertilization Strategies for Field Vegetable Production, pp.

97–106.

Jarvis, N., 1994. The MACRO Model – Technical Description and Sample Simulations. Reports and Dissertations, vol. 19. Department of Soil Sciences, Swedish University of Agricultural Sciences, Uppsala, 51 p.

Jarvis, N.J., Villholth, K.G., Ulen, B., 1999. Modelling particle mobilization and leaching in macroporous soil. Eur. J. Soil Sci. 50, 621–632.

Jensen, A.L., Boll, P.S., Thysen, I., Pathak, B.K., 2000. Pl@nteInfo® —a web-based system for personalised decision support in crop management. Comput. Electron. Agric.

25, 271–293.

Jensen, C.R., Battilani, A., Plauborg, F., Psarras, G., Janowiak, F., Stikic, R., Jovanovic, Z., Li, G., Qi, X., Liu, F., Andersen, M.N., 2009. Statistical analysis of the field

Author's personal copy M. Styczen et al. / Agricultural Water Management 98 (2010) 472–481 experiments and conclusion on the new irrigation/fertigation techniques and methods of growing potatoes and tomatoes. University of Copenhagen, Faculty of Life Sciences, Copenhagen, Denmark. p. 20. Deliverable 2.5 of teh SAFIR project. Available at: www.safir4eu.org (accessed on 6.11 2010).

Jørgensen, L.N., Noe, E., Langvad, A.M., Jensen, J.E., Ørum, J.E., Rydahl, P., 2007.

Decision Support Systems: Barriers and Farmers’ Need for Support. EPPO/OEPP Bulletin, 37(2), August 2007. Blackwell Publishing, pp. 374–377.

Mateos, L., López-Cortijob, I., Sagardoyb, J.A., 2002. SIMIS: the FAO decision support system for irrigation scheme management. Agric. Water Manage. 56, 193–206.

McBride, M.B., 1997. Mobility and solubility of toxic metals and nutrients in soil fifteen years after sludge application. Soil Sci. 162 (7), 487–500.

McGechan, M.B., Jarvis, N.J., Hooda, P.S., Vinten, A.J.A., 2002. Parameterization of the MACRO model to represent leaching of colloidally attached inorganic phosphorus following slurry spreading. Soil Use Manage. 18 (1), 61–67.

Pedersen, S.M., 2009. Final Report Prepared on the Potential of Implementation of Improved Irrigation Systems for Use of Low Quality Water Resources. Deliverable 6.5 of the SAFIR Project. Available on http://www.safir4eu.org/safir.asp (accessed on 11.10 2010). Institute of Food and Resource Economics, University of Copenhagen, 18 p.

Pettenati, M., Surdyk, S., Cary, L., Kloppmann, W., 2009. The Sorption Isotherm Concept Revisited Through Stringent Geochemical Modeling and Application to Agronomical Models: The Case of Lead, pp. 88–96 in “Safir Deliverable D4 3: Mobility and Accumulation of Selected Toxic Elements in the System Irrigation Water–Soil–Subsoil–Crop–Food Under Irrigation with Low Quality Water (report), BRGM, France, p. 109. Deliverable 4.3 of the Safir project. Available at www.Safir4eu.org (accessed on 11.10.2010).

Pinter, P.J., Hatfield, J.L., Schepers, J.S., Barnes, E.M., Moran, M.S., Daughtry, C.S.T., Upchurch, D.R., 2003. Remote sensing for crop management. Photogramm. Eng. Rem. Sens. 69, 647–664.

Plauborg, F., Andersen, M.N., Liu, F., Ensink, J., Ragab, R., 2010a. Safe and high quality food production using low quality waters and improved irrigation systems and management: SAFIR. Agric. Water Manage. 98, 377–384.

481

Plauborg, F., Abrahamsen, P., Gjettermann, B., Mollerup, M., Iversen, B.V., Liu, F., Andersen, M.N., Hansen, S., 2010b. Modelling of root ABA synthesis, stomatal conductance, transpiration and potato production under water saving irrigation regimes. Agric. Water Manage. 98, 425–439.

Ragab, R., 2002. An integrated modeling approach for irrigation water management using saline and non-saline water: the SALTMED model. In ISHS Acta Horticulturae 573: Aksoy, U., Arnac¸, D., Anac¸, S., Beltrão, J., Ben-Asher, J., Cuartero, J., Flowers, T.J., Hepaksoy, S. (Eds.), International Symposium on Techniques to Control Salination for Horticultural Productivity, pp. 129– 138.

Ragab, R., 2009. Final Version and Documented Models Available, Deliverable 3.2 of the Safir Project. Available on D3 2 (accessed on 11.10 2010).

Refsgaard, A., Styczen, M., 2006. Irrigation with low quality water: model approach.

In: Lorenzini, G., Brebbia, C.A. (Eds.), Sustainable Irrigation Management, Technologies and Policies. Transaction: Ecology and the Environment, vol. 96, pp.

257–267.

Scott, C.A., Zarazúa, J.A., Levine, G., 2000. Urban-Wastewater Reuse for Crop Production in the Water-Short Guanajuato River Basin, Mexico. Colombo, Sri Lanka.

Research Report 41. International Water Management Institute, Colombo, Sri Lanka, 34 p.

Smith, M., 1992. CROPWAT—A Computer Program for Irrigation Planning and Management, Irrigation and Drainage Pap. 46, Food and Agric. Org. of the U.N., Rome, 126 pp.

Styczen, M., Poulsen, R., Abrahamsen, P., Kloppmann, W., 2009. Decision Support System for Irrigation with Low Quality Water: System, Underlying Models and Tests. Deliverable 7.1 & 2 of the SAFIR project. Available at: www.safir4eu.org (accessed on 11.10 2010). DHI, Denmark, 123 p.

Villholth, K.G., Jarvis, N.J., Jacobsen, O.H., de Jonge, H., 2000. Field investigations and modeling of particle-facilitated pesticide transport in macroporous soil. J. Environ. Qual. 29, 1298–1309.

WHO, 2006. Guidelines for the Safe Use of Wastewater in Agriculture. Volume 2, Wastewater Use in Agriculture. WHO, Geneva, Switzerland, 196 p.