A Support Vector Machine for Landslide Susceptibility Mapping in Gangwon Province, Korea

摘要: In this study, the support vector machine (SVM) was applied and validated by using geographic information system (GIS) in order to map landslide susceptibility. test usefulness effectiveness of SVM, two study areas were carefully selected: PyeongChang Inje Gangwon Province, Korea. This is because, not only did many landslides (2098 2580 Inje) occur 2006 as a result heavy rainfall, but 2018 Winter Olympics will be held these areas. A variety spatial data, including landslides, geology, topography, forest, soil, land cover, identified collected Following this, data compiled GIS-based database through use aerial photographs. Using database, 18 factors relating forest use, extracted SVM. Next, detected randomly divided into sets; one for training other validation model. Furthermore, specifically type data-mining classification model, radial basis function kernels. Finally, estimated susceptibility maps validated. validate maps, sensitivity analyses carried out area-under-the-curve analysis. The achieved accuracies from SVM approximately 81.36% 77.49% areas, respectively. Moreover, assessment performed. It found that all factors, except soil drainage, material, texture, timber diameter, age, density area, had relatively positive effects on maps. These results indicate SVMs can useful effective