Test of the LEACHP model for predicting atrazine movement in three Minnesota soils

Leaching Estimation and Chemistry Model-pesticide (LEACHP) was evaluated for predicting atrazine (2chloro-4-ethylamino-6-iso-propylamino-s-triazine) movement in sandy loam (sandy, Typic Hapludoll), silt loam (fine-silty, Typic Hapludoll), and clay loam (fine-loamy, Typic Haplaquoll) soils under moldboard and chisel plow tillage practices. Atrazine distribution in the root zone was measured by analyzing soil cores. Model predictions were compared with observed atrazine concentrations in the soil profiles on various dates after application (two to three growing seasons). Model performance was evaluated by testing whether model predictions fall within a specified factor of true values and using Goodness-of-Fit tests (maximum error, root mean squared error, coefficient of determination, modelling efficiency, coefficient of residual mass). LEACHP predicted depth of peak atrazine concentration in the soil profile accurately in all three soils (under both moldboard [MB] and chisel plow [CP] treatments and at all sampling dates). It predicted depth of atrazine movement and atrazine concentrations in the soil profile reasonably well in the Estherville soil (sandy, mixed, mesic Typic Hapludoll) except some disparities between measured and predicted concentrations in 1988 because of the difficulty in recovering representative samples. In Port Byron (fine-silty, mixed, mesic Typic Hapludoll) and Webster soils (fine-loamy, mixed, mesic Typic Haplaquoll), depth of atrazine movement and atrazine concentrations in the soil profile were predicted reasonably well during relatively dryer years (1988 and 1989). In a relatively wet year (1987), however, the model did not predict significant amounts of atrazine at depths >45 cm, whereas, 2.5 to 27.5 μg kg -1 of atrazine were detected in soil samples during earlier sampling dates (43, 73 DAA in Port Byron and 14, 43 DAA in Webster soils). Overall performance of the model was similar in both tillage treatments (MP or CP). Considering the broad range in soil properties and climatic conditions used in testing, the model performed well.