Simulation of pesticide leaching in a cracking clay soil with the PEARL model.

Testing of pesticide leaching models is important to increase confidence in their use in pesticide registration procedures world-wide. The chromatographic PEARL model was tested against the results of a field leaching study on a cracking clay soil with a tracer (bromide), a mobile pesticide (bentazone) and a moderately sorbing, persistent pesticide (imidacloprid). Input parameters for water flow and solute transport were obtained from site-specific measurements and from literature. The model was tested using a stepwise approach in which each sub-model was sequentially and separately tested. Uncalibrated simulations for the water flow resulted in moisture profiles that were too wet. Calibration of the hydraulic relationships resulted in a good description of the moisture profiles. Calibration of the dispersion length was necessary to obtain a good description of bromide leaching. The calibrated dispersion length was 61 cm, which is very long and indicates a large non-uniformity of solute transport. The half-life of bentazone had to be calibrated to obtain a good description of its field persistence. The calibrated half-life was 2.5 times shorter than the half-life derived from the laboratory studies. Concentrations of bentazone in drain water and groundwater were described reasonably well by PEARL. Although measured and simulated persistence of imidacloprid in soil corresponded well, the bulk of the imidacloprid movement was overestimated by PEARL. However, imidacloprid concentrations in drain water were underestimated. In spite of the extensive calibration of water flow and tracer movement, the behaviour of the moderately sorbing pesticide imidacloprid could not be simulated. This indicates that the convection-dispersion equation cannot be used for accurate simulation of pesticide transport in cracking clay soils (even if extremely long dispersion length is used). Comparison of the model results from a poorly sorbed chemical (bentazone) and a moderately sorbed chemical (imidacloprid) were useful in defining the limitations of using a chromatographic flow model to simulate the effects of preferential flow.