Performance of a pesticide fate model for predicting multi-year surface runoff contamination in a Mediterranean...

Characterization of the risk of water contamination by pesticide runoff is an essential issue in the assessment of the environmental impact of pesticide use in agriculture. Models are valuable tools for this purpose. However, the use of pesticide runoff models implies that they can simulate the intra- and interannual variations in runoff contamination with a sufficient accuracy. In this study, we aimed to assess the performance of a modelling approach coupling Richards and convection-dispersion equations with the uniform mixing cell concept and an overland flow routine to predict the between- and within-year variability of the pesticide concentration in runoff water at the outlet of an agricultural field within a Mediterranean context. The model performance was evaluated by calibrating and validating model predictions to observed concentrations of two herbicides, diuron and glyphosate. The temporal structure of the model performance was analysed over 10 and 9 years for diuron and glyphosate, respectively. The results showed a favourable average model performance for runoff pesticide concentration prediction over approximately ten years. However, the annual performance largely varied, suggesting that at least 3 years of data are required to obtain reasonably precise runoff pesticides estimates in terms of the Nash-Sutcliffe efficiency in a Mediterranean climate. Although limited differences in the prediction performance were found between the two compounds, the within-year fluctuation of the predictions was more accurate for glyphosate than for diuron. This is interpreted by ageing effects that affect diuron sorption but not glyphosate sorption. Furthermore, the study results indicated that the parameters critical to precise estimation depend on the simulation objectives. The main parameter values to be determined include the pesticide half-life time when estimating chronic exposure to pesticide runoff and the sorption coefficient and mixing depth when predicting the maximum observed concentration during a given season.