Land surface parameterization schemes such as the Simple Biosphere Model (SiB2) have found considerable use in climate simulation models, where they provide lower boundary conditions in the form of surface sensible and latent heat fluxes. A methodology is described to apply models of this type at high resolution, using data from the Department of Energy's Cloud and Radiation Testbed in Oklahoma and Kansas, to determine the spatial variations of heat fluxes over the domain and to determine area-weighted flux averages for use in single-column model studies. Data from a dense array of meteorological instruments are interpolated to provide the wind, temperature, vapor pressure, radiation, and precipitation values needed by SiB2. The state of the vegetation is characterized through the use of the normalized difference vegetation index determined from satellites. The performance of the SiB2 model is evaluated by comparing its predictions with flux data from seven Bowen ratio stations over a 6-month period. No tuning of the model parameters for individual sites was allowed during the simulation period. Although there is significant scatter in the results, the performance of the model was generally good, accounting for over 60% of the variance in sensible heat fluxes and over 80% of the variance in latent heat fluxes. The model was therefore used to prepare flux maps for the study area. These maps show large contrasts in sensible and latent heat fluxes associated primarily with differences in vegetation cover and soil moisture over the site. The differences in vegetation, in turn, result from the planting of large areas with winter wheat, which leaves some regions nearly devoid of actively growing vegetation in midsummer, while other areas are still covered with thriving crops or naturally occurring vegetation. Implications for determining large-area averages of fluxes from a limited number of measurement sites are discussed. ¿ 1998 American Geophysical Union