The investigation of many issues relating to global change is dependent on spatially distributed models of land surface processes. These models may originate as point models, and the methodology for 'scaling up' to the regional level remains a significant research issue. Here, the authors describe an approach to running a carbon and water balance model (Forest -BGC) over a gridded digital elevation surface for the Columbia River Basin. Estimates were made at each 1 sq km grid point at a daily time step over a two week period (May 1-15, 1990) for climate variables required by the model including solar radiation, minimum and maximum air temperature and precipitation. The daily time series for the climate drivers was developed from interpolations of regional meteorological data. To initialize the run, surfaces for vegetation properties including biome type and leaf area index (LAI), as well as soil factors including water holding capacity and carbon content, were also prepared. The spatial distribution of vegetation type from land cover maps derived in part from satellite remote sensing.