Assessment of Landslide Susceptibility Using GIS-Based Models in Jinbu Area

서지정보

· 발행기관 : 한국지형학회
· 수록지 정보 : 한국지형학회지 / 23권 / 4호 / 153 ~ 167페이지
· 저자명 : Sara Khodadad, 조상희, 장동호

초록

This research focuses on specifying and zoning the Jinbu area based on the potential risk of a landslide happening by developing different models. For this purpose, a landslide inventory map representing landslide locations has been produced through web based aerial photograph (offered by Daum corporation) interpretation and several field works. A total 210 landslides were detected at the study area. Seven landslide contributing factors including slope, aspect, curvature, geology, forest type, forest timber diameter, and soil texture were chosen as the factors that influence the slope stability over the Jinbu area. To reveal the relationship between the landslide causative factors and occurred landslides, factors distribution and the Frequency Ratio of all factors classes were calculated in GIS environment. Three modeling methods of Fuzzy, Bayesian analysis, and Logistic regression have been applied to create landslide susceptibility maps. Then, in order to verify the practicality and success rate of resultant models, they were evaluated by a cross validation technique. Based on the Success Rate Curve plotted for the models, the Bayesian predictive analysis with an 87.03% success rate, turned out to be the most precise model compared to the other models. Fuzzy model and Logistic regression resulted in 85.07% and 83.89% Success Rates respectively. However, the produced models were just slightly different and provide acceptable and reliable models that can be beneficial for risk assessment for land managers or planners.

참고자료

· 없음