Pesticide exposure assessment in rice paddies in Europe: a comparative study of existing mathematical models.

A comparative test was undertaken in order to identify the potential of existing mathematical models, including the rice water quality (RICEWQ) 1.6.4v model, the pesticide concentration in paddy field (PCPF-1) model and the surface water and groundwater (SWAGW) model, for calculating pesticide dissipation and exposure in rice paddies in Europe. Previous versions of RICEWQ and PCPF-1 models had been validated under European and Japanese conditions respectively, unlike the SWAGW model which was only recently developed as a tier-2 modelling tool. Two datasets, derived from field dissipation studies undertaken in northern Italy with the herbicides cinosulfuron and pretilachlor, were used for the modelling exercise. All models were parameterized according to field experimentations, as far as possible, considering their individual deficiencies. Models were not calibrated against field data in order to remove bias in the comparison of the results. RICEWQ 1.6.4v provided the highest agreement between measured and predicted pesticide concentrations in both paddy water and paddy soil, with modelling efficiency (EF) values ranging from 0.78 to 0.93. PCPF-1 simulated well the dissipation of herbicides in paddy water, but significantly underestimated the concentrations of pretilachlor, a chemical with high affinity for soil sorption, in paddy soil. SWAGW simulated relatively well the dissipation of both herbicides in paddy water, and especially pretilachlor, but failed to predict closely the pesticide dissipation in paddy soil. Both RICEWQ and SWAGW provided low groundwater (GW) predicted environmental concentrations (PECs), suggesting a low risk of GW contamination for the two herbicides. Overall, this modelling exercise suggested that RICEWQ 1.6.4v is currently the most reliable model for higher-tier exposure assessment in rice paddies in Europe. PCPF-1 and SWAGW showed promising results, but further adjustments are required before these models can be considered as strong candidates for inclusion in the higher-tier pesticide regulatory scheme.