Using precise data sets on farming and pesticide properties to verify a diffuse pollution hydrological model for predicting pesticide concentration

Verification of a diffuse pollution model involves comparing results actually observed with those predicted by precise model inputs. Acquisition of precise model inputs is, however, problematic. In particular, when the target catchment is large and substantial estimation uncertainty exists, not only model verification but also prediction is difficult. Therefore, in this study, rice-farming data were collected for all paddy fields from all farmers in a catchment and pesticide adsorption and degradation rates in paddy field soil samples were measured to obtain precise model inputs. The model inputs successfully verified the model's capability to predict pesticide concentrations in river water. Sensitivity analyses of the model inputs elucidated the processes significantly affecting pesticide runoff from rice farms. Pesticide adsorption and degradation rates of the soil did not significantly affect pesticide concentrations, although pesticide discharge to river water accounted for less than 50% of the total quantity of pesticide applied to fields, possibly owing to pesticide adsorption and degradation. The timing of increases in pesticide concentrations in river water was affected mostly by the farming schedule, including the time of pesticide application and irrigation, and secondarily by rainfall events.