Surface evapotranspiration is a major component of the water cycle. The parameterized subgrid scale (PASS) model has the ability to couple routine, spatially sparse surface meteorological data with satellite remote sensing data to calculate the evaporative loss of water from the surface over extended, heterogeneous areas. The extent to which heteorogeneities in surface conditions are spatially resolved is determined primarily by the satellite data pixel size. Because PASS is observationally driven, relies on considerable parameterization of surface properties, and does not rely on a mesoscale meteorological model, it is relatively efficient computationally. Past work has shown that estimates of latent heat fluxes from PASS agree well with surface-based and aircraft-based eddy covariance measurements at the Walnut River Watershed in southern Kansas. Recent work has focused on modeling and evaluating the amount of evapotranspiration over the Watershed, an area of about 5000 square kilometers, for the five-year period 1996-2000. Surface vegetation conditions were updated with biweekly, composite, 1-km-resolution NDVI (normalized difference vegetation index) data products. Root-zone soil moisture content for each pixel was updated continuously by using modeled evapotranspiration values from PASS and radar-derived estimates of precipitation. The soil moisture content field was adjusted for times when reliable data in thermal satellite channels were available for cloudless skies. Estimates of accumulated soil moisture loss by evapotranspiration agreed reasonably well on average from year to year with the soil moisture loss inferred roughly from precipitation and stream gauge measurements.