A landslide survey of more than 1,600 slides was performed at scales of 1:50,000 and 1:5,000 in Gyeonggi province, Korea. Although most of the landslides are classified as debris flows, the slides are translational at their triggering positions, On the basis of the field survey and laboratory tests, the 198 transla· tional slides selected for this study were analyzed for triggering factors, and quantitative weighting values for these factors were determined by means of geostatistics. The 198 landslides occurred under the same rainfall condition; as much as 250 to 300mm for two days in 1998. GIS techniques acquired data for topographical factors in the whole study area, Soil tests determined ten physical properties per sample. Then, seven landslide triggering factors were selected using geostatistical methods. With the selected influential factors, logistic re· gression analysis assigned weighting values to the factors and established a formula for predicting the probability of landslide.
There have been many studies to characterize landslides for several decades throughout the world. The ultimate goal of these studies was to predict and mitigate landslide hazards. Since landslides are caused by many complex factors, it is not a simple problem to solve.