Estimation of Hazard Area for Rock Mass Failure Debris using Multivariate statistical Analysis

Abstract

Rock mass failures are one of the serious natural disasters, and the failures can affect the considerable damage to the region as suffered at the Toyohama Tunnel Collapse, northern Japan, in 1996. Therefore, it is very important to identify the extent of the hazard area precisely and effectively from the viewpoint of making hazard maps. In this paper, we analyzed the factors and their contributions that affect the extent of the hazard area on the rock mass failures, dealing with angles of elevation and spread using the quantification theory type I that is one of the multivariate analysis. And we discussed potential availability of probabilistic prediction on the hazard map.
As a result, we recognized that the angle of elevation depends largely on the volume of the rock mass and the type of the failure, and slightly on the slope condition of the lower part of the failure. The angle of spread depends largely on the elevation and the width of the failure source respectively. The coefficients of determination on the prediction of the hazard area are approximately 0.55 using the multiple linear regression analysis of the angles of elevation and spread, it seems possible to predict the hazard area by the analysis. Besides, we could assess the probabilistic prediction of the hazard area by means of the probability distribution of the residual value between actual value and predicted value on the hazard area. These quantitative predictions of the hazard area should become useful data with the object of making GIS hazard maps on the rock mass failures.