STATISTICAL EVALUATION OF UNCERTAINTY IN SMALL-SCALE GIS-BASED SOIL PROPERTY DATABASES

The limited availability of spatial soil data captured at a large scale requires that soil data captured at a small scale be used in regional environmental modeling. Defining the uncertainty associated with small-scale soil data compilation is important for evaluating the effectiveness of environmental modeling techniques that are strongly dependent on these data. These spatial soil data are being captured digitally by the USDA-NRCS and published at two separate scales in Geographic Information Systems (GIS) format. The state soil geographic (STATSGO) databases were compiled at a small scale (state-level) and are available for the entire United States. Soil survey geographic (SSURGO) databases are being compiled at a large scale (county-level) and are currently available for approximately 20% of the counties in the United States.

The purpose of this study is to evaluate and statistically quantify the variation in spatial soil properties between the STATSGO and SSURGO databases. Weld County, Colorado, and Rio Grande County, Colorado were chosen as test cases, because both databases are available for these counties and modeling methods currently used in Colorado rely on the STATSGO database. Selected properties of the uppermost soil layer, from both databases, were spatially and statistically analyzed to assess the representativeness of the STATSGO scale data. The results of this study are intended to provide a general estimate of the uncertainty associated with the STATSGO scale data in other locations. Preliminary results indicate that the soil properties contained in the STATSGO database are not spatially and statistically similar to the soil properties contained in the SSURGO database. These results suggest that the techniques for generalizing large-scale non-digital data to form the small-scale digital database need to be redesigned to more accurately capture the spatial distribution of soil properties.