Validation of the pesticide leaching model PELMO using lysimeter studies performed for registration

Abstract

The performance of simulation models to predict the amount of leached water, pesticide concentrations in soil and pesticide concentrations in the leachate from 39 lysimeters columns with undisturbed soil cores is described in this paper. The behaviour of 14 pesticides plus 1 metabolite was investigated mainly on loamy sand soils at four different locations in Germany. In three lysimeter studies a silty loam soil was used. The annual amount of precipitation ranged from 706 mm to 1393 mm (rainfall plus additional irrigation), 85 % of the experiments received a minimum precipitation of 800 mm. The K oc values of the pesticides were between 5 and 4000, the mean DT 50 values in soil ranged from 5 to 120 days. The model calculations were performed with three different computer models: PRZM-1, PELMO 1.5 and PELMO 2.0. The model PELMO 2.0 predicted amounts of cumulated leachate with a mean absolute deviation of 134 mm (relative deviation factor of 1.37) more exactly than PRZM-1 (absolute value 194 mm, relative value 1.59) and PELMO 1.5 (absolute value 205 mm, relative value 1.51). The mean variability within the replicates of the lysimeter collective was absolute 66 mm (relative factor of 1.14) over the two years period of the experiments. PELMO 2.0 predicted the amounts of cumulated leachate from the lysimeters with good accuracy, PRZM-1 was less precise, probably because crop specific scenarios were not included in this model. A statistical analysis showed that the amount of cumulated leachate depends on climatic parameters and on the production of biomass. Both models showed a slight tendency to underestimate residues in soil. PELMO 2.0 predictions of the mean annual pesticide concentrations in the leachate differed absolutely by 0.02 μg/L and relatively by a factor of 1.5. PRZM-1 predictions showed an absolute deviation of 0.01 μg/L and a relative one of a factor of 1.7 from the lysimeters. The variability within the lysimeter collective itself was 0.02 μg/L (relative value of 1.3). The 90 percentile of PELMO 2.0 predictions differed absolutely by 0.08 μg/L (relative factor of 6.4) and PRZM-1 predictions by 0.05 gg/L (relative 7.9) for the lysimeters. The variability within the lysimeter collective was absolute 0.08 μg/L (relative for a factor of 3.1). 90 % of the PELMO 2.0 and PRZM-1 simulations had a maximum deviation of not more than a factor of 10 (one order of magnitude) from the measured results. The variability within the lysimeter collective was in the same order of magnitude as the deviation of the model predictions from the mean of the lysimeter 90 percentile. Pesticides were measured in 30 % of the lysimeters, 2 % showed a mean annual concentration above 0.1 gg/L. From the agreement of observed and predicted values and the variability within these data groups it can be concluded that PELMO 2.0 is appropriate to predict leachate volumes and concentrations of pesticides in leachate water and to assess the leaching potential of pesticides for regulatory purposes within the aove mentioned ranges and limits and for the environmental conditions of the studies evaluated.