다중선형회귀분석을 이용한 대설피해액 추정에 관한 연구 = Estimation of the snow damages using multiple-linear regression analysis

다국어 초록 (Multilingual Abstract)

Recently, the damage caused by natural disasters has been increased over the world and global warming and climate change is considered as one of the most accelerating factors. 7.6% of the property damage 10 years from 1993 to 2002 in South Korea is caused by heavy snow. For example, there were substantial property damages induced by Heavy snow in Chungcheong region (March 2004) and Yeongdong·Yeongnam region (March 2005). More severe damage is also expected in the future. However, it is difficult to estimate accurate damage due to the lack of data with respect to the snow disaster. Therefore, more reliable damage estimation is required for the more effective disaster response and management. In this study, the snow damage was estimated using multiple linear-regression analysis with the regional and climate data. Historical damage was collected from the Disaster Annual Report(1994~2013) published by Ministry of Public Safety(formerly National Emergency Management). Four most crucial impact factors(Daily maximum snow depth, Relative humidity, Minimum temperature, Population density) for the snow damage were selected as the independent variables. Multiple linear-regression analysis with enter and stepwise methods were applied to estimate snow damages. As the result, adjusted R-square is above 0.8 in some sub-regions and shows the good applicability although the extreme values are not predicted well. The proposed method might be applied for the disaster response and management to prepare the damage and mitigate the property losses.