Tetsuya Kubota (Depart. Environmental Science, Tottori University, Japan)

Topographical and geological factors have been used in order to find out landslide susceptible area. In this study, Fractal dimension of landslides is discussed as a topographical factor to express landslides complex topography for investigation of susceptible areas. Considered the self-affinity of landslide landform, Fractal dimension of landslides "D" is found, by multi-discriminant analysis, to be an index of landslides susceptibility. In this situation,"D" is applied to neural network computer system(NNW) with geological condition(formation etc.) and relief energy "RE" which was demonstrated as a susceptible factor. Then, landslide susceptibility can be judged automatically by NNW as the value of "Destimated", using "RE" and geological point "G" that is given to each geological group (or formation) decided by its average "D". The result of NNW judgment for example area is successful.