Abstract
The objective of this study is to improve the accuracy of extracting areas in danger of the landslide by using the SSE model (Slope Stability Evaluation model) developed by ourselves in 1990. In this SSE model, we use not only Geographical Information (GI) but also satellite data. As a new spatial information for the satellite data, a “Fractal feature” calculated with Box Counting Method was defined. To evaluate the accuracy of SSE Model, three examination cases were executed as follows: CASE-1) Using GI. NVI (Nomalized Vegetation Index) and Fractal Features, CASE-2) Using GI and Fractal features, CASE-3) Using GI and NVI. As a result, in case of using Fractal features (CASE-1 and CASE-2), it was shown that the best accuracy for discriminating dangerous areas on the landslide could be achieved through analyses with the Quantification method TYPE II and III. Furthermore, a subtracted image extracted different pixels between CASE-1 and CASE-3 was proposed. This subtracted image is very useful for supporting the judgement whether the dangerous or not in the field investigation and the countermeasure against landslide.
