Estimates of forest land area are derived both from national-level forest inventories and satellite image-based map products. These estimates can differ substantially within sub-regional extents, e.g. states or provinces, primarily due to differences in definitions of forest land between inventory- and image-based approaches. We present a geospatial modeling approach for redefining satellite image-based pixels to meet inventory definitions. We compare resulting estimates of forest land area for six test states - Arizona, Minnesota, Montana, New York, North Carolina, and Oregon - using image estimates based on Moderate Resolution Imaging Spectroradiometer (MODIS) and inventory estimates from the U.S. Department of Agriculture Forest Service’s Forest Inventory and Analysis Program (FIA). Our geospatial model utilizes several ancillary geospatial datasets to simulate conditions required by FIA’s definition of forest land, including minimum forest patch area and width, minimum tree stocking or canopy cover, and exclusion of lands not used primarily as forest land.