Development of geographical information system for agricultural land mapping in metropolitan France and its application to estimate residential proximity to crops in the GEOCAP Agri Project

Context and objective. To meet the needs of epidemiological studies exploring the impact of pesticide exposure in the agricultural environment on health, we developed a method to create a multi- source crop map for metropolitan France. This map was then applied to the GEOCAP-Agri project on the risk of cancer in children. Method. We integrated three data sources - Corine land cover, graphic parcel register, and information from BD TOPO (R) about vineyards and orchards - into a geographic information system (GIS). The resulting multi-source map was refined by removing non-agricultural surfaces and duplicate occurrences. In the case of discordance between sources, an algorithm was used to select the most probable crop type for a given polygon. The map was intersected with the French administrative divisions as well as the buffers around the places of residence of 20,198 control children taken from the national GEOCAP study (2010-2013 period). The results of the multi-source map were then compared to the agricultural census, which is an exhaustive database that serves as a reference database but is not a detailed geographic data source. Results. For viticulture, straw cereals, oilseeds, and corn, the areas estimated with the multi-source map show a very strong correlation with those surveyed by the agricultural census, whereas this correlation is weaker for arboriculture. Nearly 90% of the French population resides in municipalities with at least one agricultural area, and seven out of 10 GEOCAP controls live within 1,000 m of an agricultural area. Conclusions. Built using a homogeneous method for the entire metropolitan territory of France, this multi- source mapping tool is evolutive. It characterizes, retrospectively and for a targeted period, the agricultural land around residential areas at different administrative scales (e.g., for ecological studies) and around geolocated places of residence (e.g., for case-control studies).