Main authors: Peter Schipper, Yanjiao Mi-Gegotek, Erik van den Berg
FAIRWAYiS Editor: Jane Brandt
Source document: »Rudolf, J. et al. (2021) Recommendations of the most promising package(s) of measures, policies, governance models and tools at national and EU level. FAIRWAY Project Deliverable 7.3 76 pp


In this article  we assess the effect of most promising activities to prevent and reduce pesticide pollution at national level using the integrated assessment tool GeoPEARL (Tiktak 2002, Tiktak 2004).
Contents table
1. Background
2. Methodology
3. Results
4. Discussion
5. Conclusions
6. Recommendations

1. Background

Safe drinking water is vital for the health and wellbeing of all. In Europe, groundwater is the most important source (50%) for the production of drinking water (EU 2016). Since groundwater moves slowly through the subsurface, the impact of anthropogenic activities may last for a relatively long time, which means that pollution that occurred some decades ago — whether from agriculture, industry or other human activities — may still be threatening groundwater quality today and, in some cases, will continue to do so for several generations to come. For this reason, the WFD requires measures to prevent or limit the input of pollutants into groundwater, and the GWD emphasises that upward pollution trends must be identified and reversed. However, many groundwater bodies including those that are used to produce drinking water in Europe are polluted by pesticides and agriculture is a major driver of failure of good chemical status to EU groundwater and surface waters (EU 2019).

In the Netherlands, 60% of the public drinking water supply is extracted from groundwater. Recent evaluations of the quality of the public drinking water resources in the Netherlands (Driezum 2020; Kools 2019) show that traces of pesticides or degradation products were found at least once in 70 of the 99 phreatic groundwater abstractions for drinking water (71%). Half of these extraction sites included one or more exceedances of the quality standard (0.1 µg/l) for one or two substances. Traces of pesticides or degradation products were also found in the extraction wells of 19% of the non-phreatic groundwater abstractions (abstraction wells under a clay layer), in 9% of which above 0.1 µg/l. A recent interim evaluation of the crop protection in the Netherlands (Tiktak 2019) concluded that progress has been made in many areas (less residues in products and water), thanks to the efforts of the agricultural sector, customers and governments, but the interim policy goals have not been achieved, including the targets for reduction of exceedances in water and drinking water sources.

Monitoring of the Dutch groundwater quality in the shallow aquifers, at approximately 10 and 25 m below ground level, show that residues of pesticides were found in the majority (62%) of the monitoring screens in the periode 2015-2019 (Loon 2020). In 34% of the monitoring wells, concentrations exceed the quality standard of 0.1 µg/l if the human-relevant metabolites are also included (such as ampa). In 7% of the measuring screens the sum of the substances found exceeded the sum standard of 0.5 µg/l. The European groundwater quality standard for active substances in pesticides (EU 2006) is 0.1 µg/l. For the groundwater protection areas in the Netherlands, the first and second stage of the assessment uses a safety factor of 10 on the standard; the admission criterion is then 0.01 µg/L instead of 0.1 µg/l.

The distributed model GeoPEARL uses agricultural and pedo-climatic data to calculate emission of pesticides to the environment and can be used to assess the effectivity of measures to decrease pesticide pollution of groundwater at national scale (Verschoor 2019). These assessments will deliver input for the recommendations of the most promising approaches to prevent and limit diffuse pesticide pollution of groundwater resources that are used for drinking water production.

2. Methodology

2.1 Selection of areas, crops, substances and scenario

The effectiveness of measures that can reduce leaching of pesticides (residues of plant protection products) was assesed with the spatially distributed model GeoPEARL (Tiktak 2002). The calculations are made for five crops (potatoes, asparagus, maize, grass and leek) and groundwater protection zones of seven drinking water locations in the south of the Netherlands . These seven areas (see figure 8) are part of the Fairway Case “Schoon water Brabant”, and according to the actual Drinking Water Protection files of the province Brabant and evaluation of actual situation of the Dutch drinking water sources (Kools 2019), these are most vulnerable for diffuse pesticide pollution by agriculture.

D73 fig08
Figure 8

For a baseline (reference) scenario, leaching from agricultural sites is calculated for the most (top 5) commonly used agents per crop in the Netherlands, according to a national survey in 2016 (statistic Netherlands, and sales figures of permitted crop protection products. The selected crops are potatoes, asparagus, grass, maize and leek. The 23 selected pesticide substances are listed in Table 4.

Based on these calculations, the common application practices and experiences gained in the project “Schoon Water Brabant” (Hoogendoorn 2020), we have further narrowed down the list of the selected agents for every crop: glyphosate for potatoes, metribuzin and glyphosate for asparagus, mecoprop-P, 2,4-D and glyphosate for grass, bentazone and glyphosate for maize, and oxamyl for leek. The following scenarios with measures are calculated, including reducing application doses, reducing application frequencies, and replacing certain agents with cleaner options:

  • Potatoes: Glyphosate, dose of 50%
  • Asparagus: Glyphosate, dose of 50%
  • Asparagus: Glyphosate, one in stead of two applications
  • Asparagus: Metribuzin, two in stead of three applications
  • Grassland: Glyphosate, dose 50%
  • Grassland: Mechanical treatment, no pesticide application
  • Maize: Glyphosate, dose of 50%
  • Leek: Oxamyl, dose of 50%
  • Leek: Azoxystrobin, dose of 50%
  • Leek: Azoxystrobin, one in stead of two applications These calculations also give an indication of the effectiveness of promosing measures like buffer stripes, biological and mechanical crop protection methods that are assiciated with less chemical applications.

2.2 GeoPEARL

GeoPEARL is based on the spatial schematization of the integrated modeling system STONE (Wolf et al., 2003) for calculating nutrient emissions from agriculture in the Netherlands and the parameterization of its hydrological model SWAP (Kroes, 2009). This schematization contains 6405 unique different plots with respect to land use, hydrogeology (soil type, soil profile, groundwater level, seepage) and meteorological region. Based on these conditions, 382 unique STONE-plots have been selected. These plots also cover areas outside the seven groundwater protection zones. For GeoPEARL applications, the plots are combined with maps of 24 crops in Dutch agriculture.

The leaching model PEARL (Van den Berg, 2016) is run for a period of 20 years, with the plots contributing to the area of use of a pesticide. Details on the calculation of the area of the GeoPEARL crops are included in Table 3. Once the median leaching concentration over a period of 20 years is calculated for each hypothetical substance and all the plots of the STONE schematization, the spatial 90th percentile can be calculated for any crop or combination of crops, using the relative contribution of each plot to the area of use. GeoPEARL version 3.3.3 was used for this study.

Table 3. Groundwater protection and water abstraction areas of the selected drinking water locations, and the areas of the representative STONE-units used for the GeoPEARL calculations.

Name of drinking water locations Groundwater protection zone (ha) Water abstraction area (ha) Representative STONE-units in the Netherlands (ha) selected for GeoPEARL
arable grass maize total
Gilze 131 17 36 8 51  96
Gilzerbaan 551 320 914 2813  2184  5911
Helmond 696 91 1224 3305  2164  6693
Helvoirt 651 19 30 22  19  72
Lith 505 2 7 71  188  266
Nuland 652 108 376 5436  2228  8040
Vessem 1822 58 1664 3891  2940  8494
Total 5009 615 4251 15547  9773  29571
Total number of selected STONE-plots (calculation units) 121 132  129  382

3. Results

3.1 Baseline scenario

The major input parameters of the selected substances and their application timing and doses, and the leaching concentrations calculated by GeoPEARL are listed out in Table 4.

The (maximum) dose and timing of application is derived from the authorised uses and regulation as indicated on the lables of the permitted products. Properties of the active ingredients (molar mass, saturated vapour pressure, solubility in water) are taken from the compound properties in the Dutch National Pesticide Risk Indicator NMI version 4. The values for half-life degradation values of the substance in the topsoil system (DT50), and the coefficient for sorption on organic matter (Kom) are taken from the end-point values in the European Food Safety Authority (EFSA) peer reviewed journals.

The results are listed in Table 4. The results of this baseline scenario show that for most substances (21 of the 23) the 90th percentile leaching concentrations are below 0,01 µg/l. Only for the assumed application, sorption en decay time of 2,4-D and Terbythylazin, the modeled concentrations were higher than 0.01 µg/l, i.e. 0.29 µg/l and 0.033 µg/l respectively. It should be noted that specific restrictions are prescribed for the use of 2,4D in the protected areas for drinking water abstractions.

These results depend largely on the input parameters, especially with respect to the half-life degradation values of the substance in the soil system (DT50), and the coefficient for sorption on organic matter (Kom). Both are the most sensitive parameters in pesticide leaching models (GeoPEARL and PEARL). The values for these parameters that are mentioned in the EFSA peer review journals differ largely. For instance, in our final calculation for Glyphosate in GeoPEARL, we have chosen the value of 3106 and 40.9 for Kom and DT50, respectively. For potatoes, when a Kom value of 9031 is used in stead of 3106, a 90 procentile of 0.001 is calulated instead of 0.005; if the DT50 value of 40.9 is replaced by 500.3, a 90 procentile of 0.009 is then generated.

To estimate the effect of measures, further calculations for the baseline scenario are made for the applications of glyphosate, metribuzin, Mecoprop-P, 2,4-D, Bentazone, Oxamyl and Azoxystrobin. The model input and results are listed in Table 5. Compared to the calculations for the baseline scenario, less favorable values have been used for sorption and degradation. For this scenario, the lowest Kom and highest DT50 values for sandy soils have been selected from the EFSA peer reviews.  

Table 4. Baseline scenario GeoPEARL; input of sorption and decay values, application data and doses of the most commonly used pesticide substances per crop in the Netherlands and the resulting calculated 90th percentile leaching concentrations baseline scenario.

Crops Substance Kom (L/kg) DT50 (days) Application Date Dose (kg/ha) P90 (µg/l)
Potato        Mancozeb 573.79 0.43 01-May, 15-May, 01-Jun, 15-Jun, 01-Jul, 15-Jul, 01-Aug and 15-Aug 1.4 0.00
Propamocarb 360.96 26.48 01-May, 15-May, 01-Jun, 15-Jun 01-Jul and 15-Jul 1.0 0.00
Prosulfocarb 1693 15.27 01-Mar 4.0  0.00
Maleic hydrazide 26.6 1.83 01-Jul 3.0 0.00
Glyphosate 13050 16.99 01-Mar 2.16 0.00
Difenoconazole 92 3760 01-Jul, 15-Jul 01-Aug and 15-Aug 0.125 0.00
Clomazone 128.3 27.3 01-Apr 0.09 0.00
Rimsulfuron 47 10.8 01-May and 01-Jun 0.01 0.00
Asparagus     Mancozeb 573.79 0.43 01-Jul, 15-Jul 01-Aug and 15-Aug 2.0 0.00
Glyphosate 13050 16.99 01-Mar 2.16 0.00
Pyridate 360.96 26.48 01-Jun and 01-Jul 0.45 0.00
Metribuzin  1693 15.27 01-Apr, 15-Apr and 01-May  0.21 0.00013
Isoxaben 26.6 1.83 01-May  0.25 0.000038
Grass      Glyphosate 13050  16.99 01-Mar 2.16 0.00
MCPA 74 25 01-Mar 1.8 0.000039
2,4-D 24 29 01-Mar 1.0 0.29
Fluroxypyr-meptyl 19550 0.32 01-Mar 0.216 0.00
Mecoprop-P 59.8 21 01-Mar 1.2 0.000014
Leek      Oxamyl 11.67 11.85 01-Mar 0.2 0.0063
Pendimethalin 6658 146.71 01-Dec 0.8 0.000025
Pyridate 7.1 1.59 01-Mar 0.9 0.00
Prothioconazole 2556 0.82 01-Jun, 01-Jul and 01-Aug 0.192 0.00
Ametoctradin 2335 1.8 01-Jan and 01-Feb 0.21 0.00
Azoxystrobin 423 180.7 01-Mar 0.25 0.0048
Corn     Dimethenamid-P 133.4  25.63 01-Apr 1.02 0.000004
Terbuthylazin 130.22 104.8 01-May and 01-Jun 0.165 0.033
 S-metolachlor 132.7 19.89 01-May 0.864 0.000001
Glyfosate 13050 16.99 01-May 2.16 0.00
Fluroxypyr-meptyl 19550 0.32 01-Apr 0.288 0.00

Table 5. GeoPEARL results for Glyphosate, Metribguzin, Mecoprop-P, 2,4D, Bentazone, Oxamyl and Azoxystrobin using the lowest input values for Kom and highest DT50 values mentioned in the EFSA peer review journals.

Crops  Substance Kom (L/kg) DT50 (days) Application Date Dose (kg/ha) P90 (µg/l) Number of calculation units
Model input Range EFSA1 Model input Range EFSA1
Potatoes Glyphosate  3106  503.8-34200 40.9 40.9 01-Mar 2.16 0.005 115
Asparagus Glyphosate  3106  503.8-34200 40.9  40.9  01-Mar 01-Apr  2.16  0.002  94 
Metribuzin 21.6 Not available 16.8 10.2-17.3 01-Apr 15-Apr 01-May 0.21 0.019 94
Grass Glyphosate 3106 503.8-34200 40.9 40.9 01-Mar 2.16 0.0000010 143
Mecoprop-P 6.84 Not available 8.05 Not available 01-Mar 1.2 0.012 143
2,4-D 38 32.8-73.5 26 Not available 01-Mar 1.0 0.012 143
Maize Glyphosate 3106 503.8-34200 40.9 40.9 01-May 2.16 0.00073 136
Bentazone 7.71 1.7-45 8.9 8.9 01-Apr 0.96 0.031 136
Leek Oxamyl 5 4.6-22.6 9.3 9.3 01-Mar 0.2 0.021 99
Azoxystrobin 248 173.3-412.7 180.7 Not available 01-Mar 11-Apr 0.25 0.049 99

3.2 Measures

For mitigation scenarios, the input parameters application (date and dose), and resulting leaching concentrations calculated by GeoPEARL are listed out in Table 6. For comparising, the leaching concentrations calculated for the baseline scenario (derived from Table 4) are also uptaken in the last column of Table 6. Figure 9 shows the spatial distribution of the calculated 50 percentile leaching concentrations. The mitigation scenarios show that leaching to groundwater can be reduced by a large extend: 50 % less dose can reduces leaching concentrations with more then 50 % (up to more then 80%) for the use of glyphosate and azoxystrobin. Also no late application can reduce leaching significantly.

4. Discussion

Decreasing the amount and the frequency of pesticide application, largely decreased the pesticide concentration in leached water. Also alteration of pesticides with less harmful products or mechanical methods can reduce leaching to groundwater and thus protect drinking water resources to a large extent.

Model simulations for the baseline scenario show larges differences if different model input values are used for sorption and degradation. From field experiments and information gathered from the EFSA peer review journals (Europan Food Safety Authority) and the Pesticide Properties DataBase (PPBD, University of Hertfordshire, UK), it can be concluded that the uncertainty margins of these values are very large. For instance, from a field experiment to determine the movement of bentazone in the south of Brabant (Boesten and Van der Pas 2000), a half live (DT50) of 206 days at 5 °C was derived and sortpion coefficent (KL value) of 0.105 dm3 kg-1, while from the ESFA a DT50 of 8.9 and a Kom range of 1.7 to 45 is mentioned. This illustrates the large uncertainties of these values wich strongly determine the risks for leaching to groundwater.

To calculate valid 90 percentile values, a minimum of 250 calculation plots in the GeoPEARL is required. This precondition is not met, which means that larger margins of uncertainty must be taken into account for the interpretation of the 90 percentile values.

Table 6. GeoPEARL inputs of different pesticides, and their leaching concentrations (mitigation scenarios)

Crops  Substance Measure Application Date Dose (kg/ha) P90 measure (µg/l) P90 baseline (µg/l) Effect measure (decrease)
Potatoes Glyphosate Dose 50% 01-March 1.08 0.00094 0.005 81%

Glyphosate Dose 50% 01-March 01-April 1.08 0.00031 0.002 85%
Glyphosate 1 in stead of 2 applications 01-March 2.16 0.00058 0.002 71%
Metribuzin 2 in stead of 3 applications 01-Apr 01-May 0.21 0.012 0.019 37%
Grass Glyphosate Dose 50% 01-March 1.08  <0.00005 0.000001 >90%
Mechanical No application / / 0   (100%)
Maize Glyphosate Dose 50% 01-May 1.08 0.00012 0.00073 84%

Oxamyl Dose 50% 01-March 0.1 0.0092 0.021 56%
Azoxystrobin Dose 50% 01-March 11-April 0.125 0.0086 0.049 82%
Azoxystrobin 1 in stead of 2 applications 01-March 0.25  0.0080 0.049 84%

D73 fig09
Figure 9

The GeoPEARL model is not designed to calculate concentrations of pesticides in groundwater at specific locations. Validation with measurements from national and regional groundwater quality monitoring wells is not possible, taken into account the uncertainties of the (legal) applications in practice, the local variable conditons that determine the behaviour and movement in the soil and groundwater and the decay between the time of infiltration and sampling in the monitoring well. In addition, the neccecary input data for sorption and degradation (DT50) derived from field study results vary a lot and observed concentrations in groundwater can origin from other (not agriculture) sources, such as infiltrating surface waters or applications in urban areas.

5. Conclusions

The pesticide leaching to groundwater in Dutch soils where the leaching conditions are representative for the safequard zones of seven drinking water areas in the south of the Netherlands were calculated with the spatially distributed model GeoPEARL. Further conceptual mitigation scenarios were used to explore the consequences of different management strategies.

The mitigation scenarios show that leaching to groundwater can be reduced by a large extend: 50 % less dose reduced leaching concentration with more then 50 % (up to 80% or even 100 %). Also no late application can reduce leaching.

The model results also show that reliable and representative parameter values for substances are very important. Therefore, more field and laboratory experiments are necessary for the improvement of the model performance. Additionally, the development of good methods to interpret data from monitoring studies could also benifit the understanding of pesticide behavior.

6. Recommendations

Rational use of pesticide is important for drinking water quality control. Pesticide application dose should be sufficient but no greater than the level required for best results, and the timing of application is another important factor to consider. The adoption of the appropriate application methods by the growers should also be taken good care of.

The risks of leaching of the pesticides strongly depend on the sorption and degradation characteristics, but the values for these parameters mentioned in literature differ largely. This emphasyis the importance of field studies to underpin these characteristics and thus the permission of pesticides.

Alternative control practices, such as the use of cleaner pesticides and mechanical suppress of pest, should be taken into consideration.

To sum up, the following recommendations on most promising measures can be derived from the results of GeoPEARL calculations:

  • Decrease input of pesticides: decreasing the amount of applied pesticides with 50% reduces the concentration of pesticides in groundwater with more than 50%.
  • Divide total application quantity over more application times: dividing the dose of pesticides in more dressings strongly (37-84%) reduces the concentration of pesticides in groundwater
  • Alteration of pesticides with less harmful products or mechanical methods reduce leaching to groundwater and thus protect drinking water resources to a large extent.


We are grateful to our partner CLM in the FAIRWAY project who have contributed to provide key information about pesticide use in the Netherlands and experiences in the Dutch case study to stimulate farm measures to reduce pesticide leaching to the groundwater bodies that are used for the abstraction of public drinking water in the south of the Netherlands.


Note: For full references to papers quoted in this article see

» References

Go To Top