A technique to discriminate landslides

Abstract

This paper describes an innovative way of distinguishing landslide-prone regions by simple and direct measurement and statistical interpretation of a topographic map. For this purpose, the topographic maps are enlarged and the contour interval and cross-slope distances are measured. A frequency distribution histogram based on chi-square method is constructed from the measurements. Generally, two fundamental patterns emerge from the histogram: the landslide-prone area shows several peaks whereas the remaining area shows just a single peak.

The technique was used to study the landslide-prone areas of the western Kyushu District of Japan. Six sub-regions were categorised depending upon the scale functions and probability parameters. The landslide-prone and safe areas were accurately discriminated under 0. 1 to 1% confidence level. The analysis independently discovered most of the hazardous areas that were later verified in the field.