A computer-assisted method for deriving soil maps of Virginia counties

Abstract

Procedures were developed for a computer-based geographic information system to map the probable soils of a county in southwestern Virginia having no soil map. To facilitate the association of known soil characteristics and interpretations with derived soil units, the procedures were based on a soil map available for a similar, mapped county in the region. Using a grid-cell size of 1/81 km² (1.2 ha or 3.0 acres), data were collected for geology, slope, aspect, topographic shape, distance to ridge, and distance to stream in both counties, and for soils in the county having a soil map. A hierarchical, monothetic, divisive classification program was written to combine these data and produce a dichotomous key to soil types. About 36,000 cells belonging to 26 soil types were classified with 55 percent accuracy using data available for the entire study area. Twenty- three of the 26 types could not be distinguished in the classification. Using more detailed data available for an 11 km² (4.3 sq mi) subarea, 900 cells belonging to 19 soil types were classified with 69 percent accuracy. Eleven soil types in the subarea could not be distinguished in the classification. High variability within some of the larger soil types mapped in the study area and inaccuracies in the topographic data vere the most probable causes of the apparent inability to classify many of the soil types, and the results were not applied to the mapped county.