Evaluation of the danger of debris-flow torrents in view of the effect value of measure constructions using Support Vector Machine

Abstract

Reducing the potential hazard of sediment-related disasters is a raging issue in Japan. In reality the counter-measures against the disasters are not progressing fast enough as expected. Therefore, the most important matter is to find a way to measure that can attain maximum effect with minimum expense by making precise evaluation on the level of potential danger of each dangerous spot and then comparing the measured effects with one and another. Takemoto et al. set the level of potential danger of debris flow, using the rough set theory based on desk work data and applied it to the debris-flow torrent in Nagasaki caused by heavy rain in 1982. However, their evaluation of the level of potential danger using the rough set theory has its own limitations -- only applicable to the data when the data of debris flow falls under the definition of the occurrence or non-occurrence. Furthermore, the effect of the measures was not mentioned. In this study, using a Support Vector Machine, accurate evaluation of potential danger on debris flow, which includes the effects of present measures, is carried out.