Modelling the spatialisation of pesticide sales to monitor environmental policies in France

Abstract

Environmental policy monitoring requires data with suitable spatial and temporal resolutions. We present a new method of spatialising pesticide sales in France to monitor environmental policies on a territorial scale. Data on pesticide sales in France are available yearly at a postal code resolution. Our spatialisation model combines parcel-scale land use data with information on pesticide types and applied doses, based on legal authorisations for pesticide use in France. We compared the performance of our new model with that of a simple surface fraction model in two distinct territories: Bourville (830 ha, Normandy), a small catchment composed of two small postal code fractions with available ground truth data about pesticide application, and Pleine-Fougères (12,007 ha, Britany), a larger area with a single postal code comprising seven municipalities with contrasting inter-municipal land uses and no ground truth information about pesticide application. For the Bourville site, due to homogeneous land use, the new model was no better than the simple model for global pesticide use, but it performed better when the results were analysed by crop type. For the Pleine-Fougères site, the new model was effective in distinguishing the use of pesticides by municipality, whereas the simple model was not. Our results demonstrate that our new model improves pesticide use quantification at a territorial level and can provide valid support for monitoring public policies regarding water quality and pesticide use over time and space.