Testing PESTLA using two modellers for bentazone and ethoprophos in a sandy soil

Two modellers tested the PESTLA model (version 2.3.1) against results of a field study on bentazone and ethoprophos behaviour in a sandy soil. Both modellers achieved an acceptable description of the measured moisture profiles but only after calibration of the soil hydraulic properties. Both could describe the bromide-ion concentration profiles measured at the end of the first winter reasonably well. However, both predicted that practically all bromide had leached out of the top 50 cm of the soil at the end of the second winter, whereas about 10% of the bromide dose remained in this layer. This is attributable to a systematic deviation of bromide transport from the concept assumed in the convection/dispersion equation and/or to the release of bromide from dead root remnants. Both modellers derived pesticide transformation and sorption parameters from laboratory studies with soil from the field. Both described bentazone movement reasonably well. Modeller 1 described the concentration profiles reasonably well, whereas Modeller 2 strongly overestimated the concentrations at the end of the study. This difference was mainly attributable to a difference in interpretation of the temperature dependence of the transformation rate of bentazone. Only Modeller 2 simulated ethoprophos behaviour. He simulated the persistence of ethoprophos in the top 20 cm very well during the first 200 days. However, thereafter the transformation in the field proceeded much faster than simulated. This is probably caused by accelerated transformation resulting from exposure of the top soil layer to about 1 mg kg−1 of ethoprophos over 200 days. Simulated penetration of ethoprophos was deeper than measured. By including accelerated transformation (admittedly on an ad-hoc basis) within the simulations, good agreement was achieved between measured and simulated penetration of ethoprophos. Calculations showed that the effect of calibrating water flow was substantial for bentazone but small for ethoprophos. However, the effect of calibration of water flow for bentazone was much smaller than the effect of the difference between the transformation rate parameters derived by the two modellers. We recommend that the guidance for deriving pesticide–soil input parameters be improved in order to reduce differences between modellers because a large influence of the person of the modeller on the outcome of model tests is unacceptable for methodological reasons.