Monitoring crop growth or calibrating crop models with satellite remote sensing require too frequent data to rely on high space resolution satellites only. Coarse resolution satellites provide more frequent data, but their space resolution is too poor to allocate radiometric data to specific crops, specially when field size is comparatively small, as it is the case in West Europe agricultural systems. This paper proposes a scheme of use for such coarse data in the frame of the Preparatory Programme of VEGETATION sensor to be launched. A map of land use is first derived from high resolution data (from SPOT satellite for instance) with all wavebands at few dates (generally three) and provides the percent cover of each crop in coarser pixels. Because the radiometric value in a coarse pixel is a weighted mean of the radiometric response of individual components, the reflecta nce of a specific crop inside each pixel is predicted by a model which unmixes coarse resolution values and reflectance of each crop is thus predicted at each available date of satellite observation. A time profile of Normalized Difference Vegetation Inde x (NDVI) is thus established for each crop. NDVIs being transformed into Leaf Area Indices (LAI), parameters of a simple model of time evolution of each crop can be estimated in each pixel with non linear regression. A synthesis of crop evolution can then be obtained when projecting values of estimated parameters on a map of the observed area. Knowledge of crop parameter distribution and space pattern at a regional scale is a first step to analyse the extension and behaviour of agricultural systems.