Environmental Project No. 1178 2007 – Evaluation of Pesticide Scenarios for the Registration Procedure – Appendix E 1 Present Use of Model Scenarios

Evaluation of Pesticide Scenarios for the Registration Procedure

Appendix E 1. Present Use of Model Scenarios

This appendix has been written by Steen Marcher and Claus Hansen from the Danish EPA.

The appendix presents the evolution from the first registration procedure including model scenarios in Denmark up to the present procedure. It also explains the background for using 1 m below ground surface (b.g.s) as reference for decisions on pesticide approval when applying mathematical models.

1.1 Original registration procedure and changes to original procedure

(Written by the Danish EPA)

A decade ago the Danish Environmental Protection Agency (hereafter Danish EPA) did not accept mathematical modelling concepts as adequate documentation on leaching of pesticides to ground water because the current models were not satisfactorily validated and were not specific for Danish conditions regarding soil and climate. By experience from national research and by joining the FOCUS group under the European Commission (FOCUS (1995), the Danish EPA elaborated the first guidance papers on how to evaluate model studies regarding leaching of pesticides and their metabolites to ground water.

The guidance paper was published November 28, 1997 (cited below).The guidance paper was – and still is – a dynamic paper, which will be updated as the administrative process is refined, caused by progress on the modelling and scenario area.

__________________________________________________________________

The first Guidance Paper for Using Mathematical Models in an Assessment of Pesticide Mobility, released November 28, 1997 (Miljøstyrelsen, 1997) - Guidance Paper for Using Mathematical Models in an Assessment of Pesticide Mobility

Object

This paper is intended to describe how the Danish Environmental Protection Agency will use results from mathematical models in the assessment of whether or not the use of a pesticide presents a risk of groundwater contamination and to describe the Agency’s requirements to models, scenarios and input data.

Background

In a number of cases, ordinary documentation in the form of laboratory tests is not sufficient to evaluate the risk of groundwater contamination, nor does supplemental documentation in the form of lysimeter studies always provide a definitive description of the actual risk in the field.

Using models to describe transport and the degradation of pesticides in soil allows testing of many more combinations of factors such as soil type, climatic conditions, application times, anticipated soil exposure, etc. than is otherwise financially and technically possible in lysimeter and field testing.

Modelling is increasingly being used as supplemental documentation. For example, modelling is typically included in EU monographies, and several member countries use models on a national level. The Agency has also in some cases commissioned model runs concerning mobility of pesticides and degradation products.

Since no validated regional EU models for groundwater studies exist as yet, this paper specifies the guidance for the Agency’s requirements to modelling and the evaluation of modelling, which are to apply until common guidelines have been set up for the use of models in the EU.

General Status of Modelling of Pesticide Leaching Today

Constructing and running models of water balances and the transport of conservative compounds is often satisfactory in the case of thoroughly studied scenarios. There are at present no validated models for pesticide transport in Danish conditions, and work on more detailed descriptions of various model parameters - especially the variability in soil conditions - should continue with respect to choice of scenarios. Modelling does not provide exact results, but the results can be indicative and used for comparisons among various compounds and scenarios.

To ensure that model results mirror reality as closely as possible, it is necessary at least to calibrate/validate all desired combinations of soil/climate scenarios against water balances and conservative transport measurement data. The optimal solution would be to use the validation procedure described in the FOCUS report “Leaching models and EU registration”, which includes both lysimeter and field testing. Since such data sets do not currently exist, it might in the short term be possible to validate against existing lysimeter testing data. This would at least ensure that the physical description of soil type is realistic. Realistic boundary conditions such as fluctuating groundwater tables and draining could then be transferred after a calibration against what we know from experience are realistic water balances. Validation against lysimeter data only should, however, in the longer term be supplemented with validation against field data.

There are many uncertainties, e.g. choice of localities (combination of soil type and climate data), choice of substance-specific parameters and, of course, the model’s handling of the various processes of the substance in question. Utilisation of a wide variety of set-up’s and input data will reduce uncertainty - and contain worst-case scenarios too.

Agency Guidance

The Agency has selected two actual localities in Denmark: Langvad and Karup. These localities were chosen, because in connection with the NPO project they were well-described with respect to climate and soil types. Data also exists in the form of field measurements of water balances. On this basis, a regional computer model (MIKE-SHE) was set up which has been validated as far as water balance is concerned. This model was then used to validate the pesticide leaching model MACRO’s water balance at the two localities.

The two scenarios, Karup (light soil with coarse sand) and Langvad (loam featuring many macropores and a rapid downward water transport), are also soils considered to represent a greater risk of leaching than the average of the two soil types found most often in Denmark: sandy soil and loam. Also, both localities are in areas receiving more precipitation than is average for either sandy soil or loam. Thus Karup and Langvad are considered to represent realistic worst-case situations.

Climatic variation is to be handled by using long consecutive time series with annual applications. Critical precipitation events are included by simulated dosing every single day of the time period in which the applicant specifies the pesticide is to be used.

The variation in substance-specific parameters - degradation and sorption - can be described using sensitivity analysis (sensitivity runs of the model) or, alternatively, by choosing the least favourable combination of degradation rate and sorption conditions.

After discussion with, among others, GEUS (Geological Survey of Denmark and Greenland) and the Danish Hydraulic Institute, and taking into account results and recommendations from, among others, the EU’s modelling work group FOCUS and the Agency’s Pesticide Research Program, the following guidance for models, scenarios, input data and interpretation of results is set up.

Modelling and Scenarios

Models: A model code should be used in which it is possible to incorporate preferential transport, including macropore and capillary flow. The model must also be applicable to Danish conditions, which at this time means using the MACRO or MIKE-SHE models.
If other model codes - or new versions of existing model codes - are used, then reporting must document that the calibrated water balance corresponds to the scenarios previously run.
Soil types and localities: The soils/localities specified by the Agency must be used, which at this time means two typical Danish soils representing sandy soil (Karup) and till with preferential flow (Langvad).
Climate data: Long series of time must be used, i.e. 30 and 24 years respectively for the two above-mentioned localities.
Substance-specific parameters: A realistic worst-case combination of degradation rate and sorption conditions is to be chosen, e.g. on the basis of a sensitivity analysis.
Application: Application of the highest dose specified by the applicant should be used in the model. Separate model runs must be made for each day of the period of time during which the applicant specifies the pesticide is to be used. If this time period is very long, however, dividing it into segments may be considered.
Crops: If several crops are used, the worst-case crop with respect to plant coverage, root development, etc. should be selected if possible. Alternatively, runs can be made for all crops.
Results must be stated as annual or seasonal averages. Peak values should also be listed.
If other values/input data than those specified by the Agency/default values are used, then an explanation of why must be included.

Interpretation of Results

The assessment will be based on the average amount, which leaches down below the root zone (at a depth of about one metre) per year or season in the individual runs.
The number of instances in which amount leached exceeds the maximum residue limit is to be compared to the total number of runs. If the number of such instances exceeds a certain specified total number of instances, then the use applied for cannot be approved on the basis of the modelling performed.

The above-mentioned specifications are guiding only. An ad hoc evaluation must always be made for each pesticide on the basis of all the material available: laboratory, lysimeter and field testing as well as monitoring, modelling, etc.

The guidance stated here will be modified as new data and know-how become available.

User guidance

The Agency will set up certain requirements, including a detailed description of scenarios (soil types, water balances, climate data, etc.) and accompanying documentation of the calibrated water balances used previously for Karup and Langvad. These water balances must be reproducible and are to be documented by reporting model results.

The Agency offered this data on CD-ROM under certain conditions by the end of 1997.

End of Guidance Paper

Some comments can be made on the first Guidance paper:

The scenarios Karup and Langvad were chosen simply because it was the best described scenarios, probably the only useable, and because they at that time was assessed as the most representative realistic worst case scenarios for Danish conditions.

The reason to choose 24 years and 30 years for the two scenarios, respectively, was that this was the existing time periods.

By experience and progress in the model concept the guidance paper from 1997 was refined and clarified in the Danish EPA frameworks for assessment of pesticides 28 May 1999 (cited below).

The major progresses were as follows:

It was now no longer necessary to model application on every single day for long application periods. Instead representative sub periods could be accepted.

The results should alone be reported as annual averages. The Danish EPA no longer found reasons for requiring seasonal averages and peak values as only the annual averages should be evaluated.

A more explicit formulation of the trigger for safe use regarding ground water was presented: To support approval for the proposed application the limit value of 0.1 μg/L must not exceed 5 % of the occasions meaning that the 95^th percentile regarding the output data should be used.

__________________________________________________________________

The second Guidance Paper for Using Mathematical Models in an Assessment of Pesticide Mobility, released May 28, 1999 - Evaluation of mathematical modelling of risk of ground water pollution (Miljøstyrelsen, 1999)

The leaching of active ingredients and metabolites will usually be assessed with respect to the substance’s intrinsic properties, lysimeter tests or field studies. Unless the results irrefutably show that no unacceptable leaching will take place under Danish conditions, mathematical modelling must be carried out and included in the overall assessment.

As validated regional ground water models still not excist in the EU the Danish Environmental Protection Agency has made a guidance paper for mathematical modelling of the risk of ground water pollution from plant protection products. To use the modelling in the evaluation the mentioned guidance instructions should be followed. Alternatively, deviations from the guidance instructions should be justified and/or “worst case” situations should be used.

The following requirements for modelling and scenarios must be satisfied:

§ Models: A model code, which can indicate preferential transport mechanisms, including macropore flow and capillary rise, must be used. The model shall be usable for Danish conditions. This means the “MACRO” and “MIKE-SHE” models. If another model code is used, the report must document the way in which the calibrated water balance corresponds to the Danish scenarios.

§ Soil types and localities: The soils/localities specified by the Danish EPA are used – at present, two typical Danish soils, representing sandy soil (Karup) and moraine clay with preferential flow (Langvad).

§ Climate data: Time series over 30 and 24 years, respectively, for Karup and Langvad must be used.

§ Substance-specific parameters: A realistic “worst case” combination of degradation rate and sorption, ex. based on sensitivity analyse, must be used.

§ Crop: Where several crops are involved, the worst-case crop (with respect to vegetation mantle, root development, etc.) must be used where possible. Alternatively, all crops must be modelled.

§ Application: Application of the highest dose for which approval is sought must be modelled. Separate model runs should be presented for every single day in the application period. Alternatively, if the application period is very long, representative sub periods can be accepted.

§ The results must be reported as annual averages.

§ All values/input other than those set by the Danish EPA/default values must be justified.

The appraisal is done on the basis of the average annual leaching to below the root zone (a depth of about one meter). The number of occasions when leaching exceeds the limit values are accounted against the total number of runs. If the limit is exceeded on more than a specified proportion of the occasions (5%, as the point of departure), the model runs cannot be used to support approval for the proposed application.

End of Guidance Paper

The following years the practical experience of the Danish EPA in assessing approvals on pesticides showed that practically no pesticides or metabolites could be forced to leach through the Langvad moraine clay scenario by preferential flow when using the above mentioned model set-up. This modelling result was in contrast to monitoring results showing that pesticides and their metabolites in some cases depending on compound and uses could be detected under moraine soils. Moreover, GEUS concluded that the Langvad scenario was not a worst case scenario regarding Danish moraine clay soils. Therefore, the Danish EPA found that it did not make sense still to require modelling on the Langvad scenario. For this reason the Danish EPA considered to accept modelling alone on the Karup scenario.

Another experience was that the Karup sandy soil scenario was not as conservative as expected comparing the model results with the FOCUS PELMO Hamburg scenario. Actually, the results showed a comparable or even a little higher leaching in Hamburg than in Karup. At the same time GEUS [står dette i den første VAP rapport?] confirmed that the Karup scenario was not a worst case scenario for Danish conditions. In this light it seemed reasonable to accept modelling on the PELMO Hamburg scenario in stead of using the two Danish scenarios: Langvad and Karup.

Moreover, again learning by experience it seemed to be too conservative to require a worst case combination of degradation rate and sorption parameters. In practise, the possibility that the highest degradation rate would coincide with med the lowest adsorption coefficient was very low. Therefore, the requirement was changed to use 80^th percentile values for degradation rate and sorption conditions (including 1/n) derived from studies representative for Danish conditions. Compared with the recommendation from EU FOCUS ground water modelling of using 50^th percentile values for the two parameters the Danish guidance was still relatively conservative.

Finally, the Danish EPA found that it was not necessary to require modelling for every single day in the theoretical application period.

On this background the Danish EPA of practical reasons refined the requirement as described in an updated annex to the frameworks for assessment of pesticides June 21, 2005.

__________________________________________________________________

The third Guidance Paper for Using Mathematical Models in an Assessment of Pesticide Mobility, released June 21, 2005 - Framework for the assessment of plant protection products (Miljøstyrelsen, 2005)

Appraisal of mathematical modelling on risk of ground water pollution

The following requirements for modelling and scenarios must be satisfied:

§ Climate data: Time series over 30 and 24 years, respectively, for Karup and Langvad must be used and 20 years (+ 6 years calibration) for Hamburg.

§ Substance-specific parameters: 80^th percentile values must be used for degradation rate and sorption conditions (including 1/n). These values must be derived from studies that are relevant/representative for Danish conditions.

§ Crop: Where several crops are involved, the worst-case crop (with respect to vegetation mantle, root development, etc.) must be used where possible. Alternatively, all crops must be modelled.

§ Application: Application of the highest dose for which approval is sought must be modelled. To illustrate the sensitivity to changing the date of application separate model runs should be presented for different dates during the period in which use of the product is proposed.

§ The results must be reported as annual averages.

§ All values/input other than those set by the Danish EPA/default values must be justified.

End of Guidance Paper

Caused by this revised Guidance paper it was now possible for the applicants when applying for approval of pesticides in Denmark to use the results from the EU registration procedure (the FOCUS PELMO on the Hamburg scenario) with only two corrections: Use of the 80^th percentile in stead of the geometric mean for degradation rate and sorption constant, and use of the 95^th percentile in stead of the 80^th percentile regarding the output data (average leaching in the 1 metres reading point).

The applicants were very satisfied by the possibility to use the FOCUS PELMO on the Hamburg scenario. Actually, the Danish EPA have not received modelling on the Danish scenarios Karup and Langvad after releasing the above mentioned revised Guidance Paper.

__________________________________________________________________

1.2 The Professional Administrative Rationale for using Pesticide Leaching at 1m in the Registration Procedure

(Written by the Danish EPA)

1.2.1 The Danish Approach

The selection of 1 metre as a reading point when assessing potential ground water pollution with pesticides and their metabolites has both practical and historical reasons.

One and a half decade ago the Danish EPA (Environmental Protection Agency) decided to set up criteria for the acceptance of pesticides. A working group with delegates from the Pesticide Advisory Board and the Danish EPA was formed with the objective to set up criteria ensuring protection of the ground water from the point of view that ground water was the main source of drinking water.

The decision criteria were thus based on the Drinking Water Directive and the National Danish Drikkevandsbekendtgørelsen with the limit value specified as 0.1 mg/L for pesticides and pesticide related compounds (for each single compound) and 0.5 mg/L for the sum of compounds.

The fulfilment of these criteria could in principle be assessed in different depths of the soil. In reality, the criteria should be met in the ground water, but it was not possible to make this operational for the registration purpose. For practical reasons it was therefore decided to base the assessment on the annual volume of ground water formed under one field. This definition was considered relevant and in line with other investigations at that time i.e. field and especially lysimeter studies, where an annual volume of leachate was collected and analysed for pesticide and metabolites.

Since lysimeters almost universally were 1 metre deep, it was quite natural to identify 1 metre as a reading point. This was further supported by the fact that field investigations of pesticide movement usually included analysis of soil samples down to approximately 1 metre for quantification of pesticide movement.

At that time, it is important to mention that the conceptual understanding of pesticides leaching was based on matrix transport, only. It was known that leaching could be facilitated by macro pore flow, but this was not included in the assessment as the knowledge about macro pore flow was very sparse especially in relation to the quantitative significance.

Another reason for the selection of 1 metre as a scientifically defendable reading point was that if compounds had leached to a depth of 1 metre further dissipation caused by sorption and degradation was not likely to occur. The assessment was that the concentration of a pesticide would not decline significantly having reached a depth of 1 metre. One argument for this was that the concentration of organic matter, which was thought to be the sole energy resource for the micro-organism, declined strongly with depth meaning that there would be practically no activity under 1 metre. Moreover, the depletion of organic matter was also taken as an indication that adsorption of the compounds would be negligible, as organic matter normally determines the adsorption of pesticides.

It is important to stress that the above-mentioned criteria were only established with regard to the approval system, which by nature is prospective (i.e. it is targeted for pesticides that are not yet on the market). The 1-metre reading point was not meant as a criterion for the assessment of monitoring results. The focus of the approval system regarding the pesticide content in ground water was not a defined depth but one that was definitely larger than 1 meter.

Recently in setting up criteria based on computer simulations of leaching it was a natural extension to use 1 metre as the reading point for the output of simulations. This was necessitated by the wish to be able to make comparisons with lysimeter results. Thus, the development of criteria was an on going parallel process in many countries and consequently the 1 metre reading point was adopted in EU legislation.

Thus, the 1 metre reading point is for practical and protective purposes and do not represent a depth, at which the ground water should comply with the 0.1 mg/L limit.

1.2.2 The European Approach

Principles and interpretations comparable to the Danish approach can be seen in the EU guidance paper on relevant metabolites (European Commission, 2003.), chapter 2: Context and general approach:

“As noted above, this document is intended to provide guidance for the inclusion of active substances in Annex I of Directive 91/414/EEC. According to Art 5 of the Directive “… an active substance shall be included in Annex I … if it may be expected that plant protection products containing the active substance … do not have any harmful effects … on groundwater…”. This possibility of potential groundwater - or drinking water - contamination is investigated generally on the basis of the convention that a soil layer of approximately 1 m is used to represent the “groundwater” aquifer. Such an assumption is far from representative for all regions of Europe but it is considered to provide a realistic worst case on the European scale, in compliance with Art 5 of the Directive. Should, at a future stage, more realistic assessment schemes and models become available for refined assessments at the European scale (e.g. probabilistic assessments based on real groundwater distribution data), this Guidance document will be revised to reflect such a progress.”

Another example on the fact that the 1 metre is a reading point only, can be seen in the report “Generic guidance for FOCUS groundwater scenarios, Version 1.1, April 2002” (FOCUS, 2002), where the executive summary states that “The models all report concentrations at 1m depth for comparative purposes, but this does not represent groundwater”.