Application of VARLEACH and LEACHM models to experimental data on leachingof a non-reactive tracer and three sulfonylurea herbicides

A chromatographic model (VARLEACH) and a process-based model (LEACHM) were evaluated under low and high water input regimes ('rainfall' and 'rainfall + irrigation', respectively) using experimental data for a non-reactive tra cer (KBr), and 3 sulfonylurea herbicides: triasulfuron, metsulfuron-methyl, and chlorsulfuron. Both qualitative and quantitative comparisons of predic ted and measured water contents revealed that both models performed well un der 'rainfall', but prediction was less satisfactory under the treatment of 'rainfall + irrigation'. Both models failed to simulate the bimodal distri bution of Br- and the herbicides observed under 'rainfall + irrigation' tre atment. Distribution of Br- and the 3 herbicides in the profile was over-or under-estimated at certain depths and times, probably due to the limitatio ns inherent to the models (e.g. inability to deal with the macropore flow), and to simplifications associated with some input data. Simulations for a dry (1 decile), median (5 decile), and wet year (9 decile) with LEACHM and VARLEACH models using long-term weather data for the site showed that leach ing and persistence of the herbicides are highly dependent on the amount of rainfall the site receives. The results indicate that the risk of injury t o herbicide-sensitive crops in the following year of application of herbici des is likely to be greater when the herbicides are applied in a drier rath er than a wetter year.