토양함수지수를 이용한 국가 산사태 예보체계 구축 및 운영

Abstract

The objective of this study was to describe the background and methodology of the current operational national landslide forecast system in Republic of Korea and analyze its operational results in the past two years. Soil water index and tank model were employed as the national landslide forecast system with a goal of pre-1-hour forecast for landslide likelihood in a national scale. The tank model was parameterized with the past rainfall data inducing landslides in each divided eleven region after considering rainfall and geological characteristics in the whole country. The national landslide forecast system have provided local administrations with the information of predicted warnings calculated with observed and predicted rainfall data by Korea Weather Administration in a hour since 2013, and they issue landslide warnings including landslide watch and alert based on the information. As results of its two-year-operation, 420 and 248 predicted warnings were calculated in 2013 and 2014, respectively. But, 33 of 420 in 2013 and 17 of 248 in 2014 were not finally issued in reality, accounting for 7.9% and 6.9%, respectively. Also, predicted warnings were not calculated for 36 of 72 landslides occurred in 2014 compared to only one out of 212 landslides in 2013. Overall, improvement of input rainfall data accuracy, regular revision of tank model parameters, and relevant adjustment of landslide warning issuing level were suggested to improve the efficiency of the national landslide forecast system.