Evaluation of Fuzzy-logic Based Soil Inference Modeling for Soil Survey Updates in Pennsylvania.
Rick L. Day1, ED White2, Gary W. Petersen3, and John Chibirka2. (1) Pennsylvania State Univ, 116 ASI Building, University Park, PA 16802, (2) USDA-NRCS, One Credit Union Place, Harrisburg, PA 17110, (3) Pennslyvania State Univ, 116 ASI Bldg, University Park, PA 16802
Soil inference modeling, based on fuzzy-logic theory, was applied to Soil Survey mapping updates in watersheds of two Pennsylvania Counties. SoLIM software, developed by the University of Wisconsin, was used to conduct the modeling. Geographic information system (GIS) data were analyzed to characterize landscape topographic and environmental parameters and were combined with expert knowledge to predict spatial distributions of soils and associated probabilities of occurrence. Expert knowledge data acquisition included interviews with field soil scientists and “data mining” of prior field mapping from published Soil Survey Reports. Effectiveness of “data mining” to accelerate the modeling process was evaluated. Field validation was conducted to evaluate model success.