Abstract
Recently, large-sized landslides causing a river blockage due to high magnitude earthquake or torrential rainfall occur and form a number of landslide dams. The water blocked by the dam may provoke floods which spread out the down stream area, causing damages to human activities and communication lines. The establishment of prediction method of breaching landslide dam is very important to prevent these disasters. Therefore, we examined the prediction method of breaching landslide dam utilizing by pattern classification analysis based on the fuzzy if-then rules using of data of the past landslide dams. In this analysis, new two indexes, where are designated as the ratio of landslide dam length/height (SFI) and the ratio of reservoir volume/catchment area (CAI) , are introduced for the prediction of breaching landslide dam. The result that improved of the precision of more than 10% of prediction accuracy is obtained. In addition, the influence degrees to the breaching landslide dam by the attribute factor are clarified from the fuzzy if-then rules obtained from the analysis.
