GIS is a powerful tool for modeling environmental problems, but data availability issues often hinder applications. In the development of a 1:24,000-scale erosion model to estimate erosion rates in the 1950s and 1980s for the Yadkin River basin, a 5,896 km2 watershed, complete watershed coverages were digitally available for all model variables except soils and land cover. To avoid the time and cost constraints of digitizing the soil surveys for 12 counties and thousands of aerial photos, sample areas were selected for digitizing using a stratified random sample based on elevation, slope, and distance from a stream.