ABSTRACT
This paper briefly describes a simple statistical modeling technique, known as Chi square distribution, applied for pattern recognition of associated or potential landslide failures from the frequency distribution of topographical contour spacing. It is an effective way to recognize or distinguish landslide regions by simple and direct measurement with statistical interpretation from topographical features. The technique involves enlargement of topographical maps in some suitable scale and measurements of the contour interval, cross slope distances etc., and then plotting of frequency distribution histograms. The histograms of topographical data can be distinguished into basically two fundamental patterns (single peak or multiple peaks), when analyzed by χ2 method in such frequency distribution plots. The application of this approach reveals that the potential landslide areas and non-landslide areas can be as accurately discriminated as unto 0.1 at 1 % risk level. The map analysis independently discovered most of the hazardous field areas that were later verified on the site, which demonstrates the applicability of this technique in preliminary evaluation of the potential landslide or unstable zones.
