GIS를 이용한 산사태 발생예측도 작성과 결과 비교 : 용인, 안성지역을 중심으로
Prediction mapping for landslide occurrence using GIS and result comparison in Yongin-Ansung area
산사태 GIS 용인 안성;
- 원문 URL
Landslides caused by the intensive rainfall lead to a loss of lives and properties every year, but it is hard to recognize landslide in natural slope. Thus Korea Institute of Geoscience and Mineral Resources developed a probabilistic prediction model for debris flow occurrence. We applied this model to Yongin-Ansung area that had various sized landslides about 660 by the local downpour in 1991. The studied area have low elevation that is below 500m, and show lithology of Jurassic fine-grained granite in the east and Precambrian banded gneiss in the west. We determined the factors that affect landslide by basic statistical analysis, and calculated the coefficient of factors through logistic regression analysis. And we input coefficient to 1/5,000 digital topographic map which divided into 5×5m grid, thereby we made prediction map of landslide occurrence by overlaying layers generated by factors. The prediction map indicated that areas which have landslide occurrence probability is 75% and that is 11% of total studied area, 50㎢. When a dimensional distribution was analized by the occurrence factors, we could know the characteristics of areas that are susceptible to landslides; 3.68㎢ in elevation of 200-400m, 4.75㎢ in slope angle of 20-40°, 4.18㎢ in porosity of 70-90%. And samples of slided locations had lower dry density value and permeability than samples of non-slided locations. We compared the results with Hazard-Mapping Technique for Landslide using GIS of Korea Forest Research Institute in 2002. The results shows similar distribution of slided locations and different hazard grade. We made isopleth map to compare with prediction map and to generalize spatial distribution of landslide deposit quantitatively using inventory map of landslide deposit observed in field. And because of insufficient data, prediction map did not accord with isopleth map exactly.