The objective of this study was to explore the potential of very high spatial resolution imagery to contribute to a better understanding of the Rift Valley Fever (RVF) transmission at the local scale in the Ferlo Region in the northern Senegal. We propose a landscape approach to map the favourable mosquitos? biotopes and to test for associations between landscape variables and RVF incidence rates around the village of Barkedji, Ferlo region, Senegal. A very high spatial resolution satellite image (2.4 m /pixel resolution) provided by the Quickbird sensor was used to detect and characterize the temporary ponds, which are the breeding sites for Aedes Vexans and Culex Poicilipes, the two main mosquito vectors of the RVF virus. We applied object-based image-processing techniques, which exploit both spectral and textural information, to provide a detailed pond map, a vegetation map around the ponds, and a general land use map. Then, we derived from these maps five landscape variables, based on bibliographic knowledge of the vector ecology: - a landscape closure index, - an index of water vegetation coverage, - and a pond density index, - the location of the pond, - the surface area of the pond. To test the relations between these landscape variables and RVF incidence, we used a beta-binomial regression model. The Akaike's information criterion allowed selecting the best fitted models. The 500-m landscape closure index was significantly correlated with higher serologic incidence (p<0.05) showing the influence of the vegetation density on the RVF incidence rates in small ruminants herds. (Texte intégral)