Automatic Soil Classification Using Quantitative Terrain Factors.

摘要

A success rate of 60 to 80% was achieved in identifying the soil parent materials of sample areas from among several highly similar parent materials. Only quantitative terrain factors, which were computed from digital elevation and drainage data, were used for discriminating the soil parent materials from each other. The results clearly demonstrated that a reliable method of automatic soil mapping is now technically feasible. Nine soil associations from the state of Illinois and one each from the states of Indiana, Nebraska and Kentucky were used in the study. Eleven commonly used terrain factors were modified for efficient computation using electronic computer, and standard univariate and multivariate analysis techniques were used for testing and classification. Drainage density and surface variance, which is a statistical measure of surface relief, were found to be the most efficient factors for discriminating the soil parent materials. (Author)