NUMERICAL APPROACH TO SOIL MATERIAL CLASSIFICATION : II. Sample Allocation

Multiple discriminant analysis was applied to allocating soil profiles into one of the predefined soil material classes. Fifty-six profiles in four areas, differing with respect to soil material, in Kaya Township, Kyoto Prefecture, were examined in this study. Numerical values, which were assigned to each profile by Hayashi 3 analysis and which reflect similarity between any pair of profiles, were used as the discriminating variables. Representative profiles, which were employed for the derivation of discriminant functions for each soil material class, were selected so that they would fall within the 90% probability ellipses of the respective soil material classes computed from the scores assigned to samples through Hayashi 3 analysis. Representative profiles were tested for the adequacy of the class assignment, and the remainder were allocated into appropriate soil material class using the derived discriminant functions. Fifty-two out of fifty-six profiles were allocated into the same soil material class...