This paper addresses vegetation monitoring at European scale using satellite data. Such a study requires the simultaneous use of different kinds of sensors: high temporal resolution sensors, such as NOAA-AVHRR, provide daily images making it possible to study vegetation growth on large areas, whilst high spatial resolution sensors (such as SPOT) provide local access to accurate ground measures. The final objective is the estimation of vegetation at both high spatial and high temporal resolutions. The fusion process can be summarized into three steps: learning of land covers reflectances on NOAA data; computing high quality ground data using a SPOT sequence on a given area; on-line estimation of vegetation evolution on an unknown site in the vicinity of the learning site. The first two points are learning phases, that can be applied on a set of well chosen learning sites. The last step is the applicative phase of the process: vegetation is analyzed with incoming satellite data.