Estimating the Threshold Value of Disaster-Causing Snowfall by Region and Its Application to Regional Classification

Abstract

In this study, the estimation of damage-causing thresholds and their application to regional classification were examined using logistic regression, critical success index, probability of detection, and a false-alarm ratio. For 10 stations where damage caused by heavy snow occurred frequently, the fresh snow depth data provided by the Korea Meteorological Administration and heavy snow-related damage data obtained by the Ministry of Interior and Safety were used. Results show that the estimated thresholds differed by region. These results are attributed to many factors such as analysis methods, regional disaster protection, the response of local residents, vulnerability of structures, and meteorological characteristics. Additionally, based on estimated thresholds, the area with the highest threshold-sensitivity and the areas with the most internal differences and similarities between calculated thresholds corresponded to the coastal region, the Gangwon region, and the inland region, respectively. Gangwon region, where heavy snow was often observed, had a long duration of snowfall, whereas inland regions faced snowfalls that were relatively weaker in intensity and shorter in duration.