A coupled vegetation growth and soil-vegetation-atmosphere transfer (SVAT) model is used in conjunction with data collected in the course of the SALSA program during the 1997-1999 growing seasons in Mexico. The objective is to provide insights on the interactions between grassland dynamics and water and energy budgets. These three years exhibit drastically different precipitation regimes and thus different vegetation growth. The result of the coupled model showed that for the 3 years, the observed seasonal variation of plant biomass, leaf area index (LAI) are well reproduced by the model. It is also shown that the model simulations of soil moisture, radiative surface temperature and surface fluxes compared fairly well with the observations. Reflectance data in the red, near infrared, and short wave infrared (SWIR, 1600 nm) bands measured by the VEGETATION sensor onboard SPOT-4 were corrected from atmospheric and directional effects and compared to the observed biomass and LAI during the 1998-1999 seasons. The results of this 'ground to satellite' approach established that the biomass and LAI are linearly related to the satellite reflectances (RED and SWIR), and to vegetation indices (NDVI and SWVI, which is a SWIR-based NDVI). The SWIR and SWVI sensitivity to the amount of plant tissues were similar to the classical RED and NDVI sensitivity, for LAI ranging from 0 and 0.8 m2 m-2 and biomass ranging from 0 to 120 g DM m-2. Finally, LAI values simulated by the vegetation model were fed into a canopy radiative transfer scheme (a 'model to satellite' approach). Using two leaf optical properties datasets, the computed RED, NIR and SWIR reflectances and vegetation indices (NDVI and SWVI) compared reasonably well with the VEGETATION observations in 1998 and 1999, except for the NIR band and during the senescence period, when the leaf optical properties present a larger uncertainty. We conclude that a physically-sound linkage between the vegetation model and the satellite c