Application of Geostatisical Techniques for Predicting Continuous Soil Properties From Soil Survey Polygons On Glaciated Landscapes.

Frequently, the soils information requested by users goes beyond the scope of the readily available nationally distributed soil survey, available through the Soil Survey Geographic (SSURGO) database. Originally conducted at scales ranging from 1:15,840-1:63,000, U.S. soil survey captured soils as polygon classes having uniform properties within classes and discontinuous boundaries between classes. Much of the inherent soil variability information, required for applications such as modeling, is lost in the uniform, discrete polygon map units. Therefore, a method is needed to generate informative, continuous soil property information from SSURGO maps at the various raster scales required for modeling and other mapping purposes. The research objective of this project was to take this class information, in conjunction with terrain attribute data, and develop quantitative 10 m raster soil property predictions, focused specifically on soil organic carbon. In the Little Pine Creek watershed in north central Indiana, point coordinates were extracted from the center of each SSURGO map unit polygon and assigned the corresponding SSURGO property values. Central locations were selected assuming that the polygon center is the best representation of a mapping unit and therefore most accurately representative of centroid soil properties. Soil mapping units were stratified based on parent material and historic vegetation. Total organic carbon estimates for the upper 50 cm were extracted from SSURGO and regressional relationships were developed between soil carbon at the centroids and various terrain attributes found to covary. Regression Kriging of the centroids was then preformed to produce continuous soil carbon estimates for the approximately 5,331 hectare watershed.