The paper presents highlights of a five-years investigation on the use of airborne and space observations to study the heat and water balance of heterogeneous land surfaces. Three algorithms to estimate heat fluxes using satellite observations were developed and evaluated against ground measurements collected during the EFEDA (Spain), HAPEX-Sahel (Niger) and HEIFE (China) International Field Experiments. A local area meteorological model (HIRLAM) was applied to obtain spatial fields of heat fluxes atmesoscale. Comparison of model calculations with satellite-based estimates of albedo and of evaporative fraction indicated that significant improvements of model performance could be achieved through assimilation of multispectral satellite observations. A preliminary case-study was performed on the assimilation of satellite estimates of evaporative fraction to improve the soil moisture field used for initialization of forecast runs. The model RACMO was used for the case-study. Results indicate that assimilation improved model performance.