Research chemist, U.S. Bureau of Reclamation, Denver, CO
Soils from a geotechnical project in Oklahoma may be classified as dispersive or nondispersive with a high degree of confidence by substituting selected pore-water data (weight percent H2O at saturation, saturation extract pH, log base 10 of the pore-water conductivity, and log base 10 of the pore-water sodium concentration) into derived discriminant classification functions. The discriminant functions were derived using discriminant analysis, a multivariate statistical classification technique, performed on a 28-sample data set. This statistical procedure was utilized after problems were noted with the currently accepted porewater chemical test for dispersion, suggested by Sherard et al. Suggestions for using the derived discriminant function as the basis of a simple chemical field test and precautions regarding data interpretation are presented.
Paper ID: GTJ10609J