Coffee (Coffea arabica L.) is one of the most important global commodities and employs 125 million people, mostly in the tropics. Coffee production systems vary widely, from monocultures to diversified agroforests, managed at a range of intensification levels (from organic coffee with low inputs up to very intensive systems). These different systems are not equivalent in terms of profitability and provision of ecosystem services (ES) (production, carbon sequestration, erosion control...). In order to facilitate the remuneration of ES provided by these plantations, it is recommended to evaluate the provision of services either directly or indirectly using indicators or models (MEA, 2005). The leaf area index or LAI is defined as the leaf area per unit of soil area. Our working hypothesis is to consider that plantation LAI is an integrated indicator of ES: leaf area determines light interception, gas exchange and hence primary productivity, and LAI can be related to the filling of fruit (yield), to the annual input of litter to soil (C cycle and soil OM) and to rain interception (erosion control). The LAI of coffee is affected by the quantity of light it receives (Righi et al., 2007) and varies significantly between different levels of shading in agroforestry systems. Coffee is pruned annually following different methods (per line, selective ...). Coffee LAI varies between dry and wet seasons (Mendez et al., 2009). Seasonal variations of LAI are explained by intrinsic phenology variations, soil moisture (Siles et al. 2010) or by the agricultural calendar. Most previous studies of agroforestry coffee LAI were performed at the experimental plot scale. In contrast, we propose to explore the plant, plot and farm scales, using a combination of satellite images with high and low spatial and temporal resolution, and to study temporal series in contrasting conditions (climate, degree of shade or management). On these images, we recorded the variation of a vegetation index (No