Ecosystem Services (ES) provided by coffee plantations may differ according to agricultural practices (monoculture or agroforestry). The leaf area index (LAI) is assumed to be a good parameter to model ecosystem ES. We tested a method to measure leaf area index of coffee agroforestry plantations that display moderate shade or no shade, being valid for a large spatio-temporal scale, relying on a vegetation index (NDVI) derived from satellite imagery (Very High Resolution THRS and medium resolution MODIS). Time series (2001-2009) of LAI were obtained following two steps of calibration: a first calibration between the actual LAI measured in the field and the NDVI THRS image (R ² = 0.80) in order to obtain maps of LAI at high resolution; the second step between the high-resolution LAI map and NDVI from MODIS (R ² = 0.66). We studied the effects of environmental and agronomic factors on LAI. The seasonal dynamics of LAI (2.5 and 4 in the dry season in the wet season) was connected with variations in air humidity and solar radiation. We found that the LAI of some key months was correlated with coffee yield. The dynamic of LAI was also influenced by agricultural practices such as the pruning of coffee. This work is a first step towards spatial modeling of ecosystem services and production.