Prediction of landslide hazard area using GIS and probability models

초록

Landslide hazard area was predicted using conditional probability models under climate change scenarios. The primary predictors for landslides were forest type slope aspect slope gradient and rainfall. By using these factors a landslide area was predicted by both the Direct and Bayes models. We tested two partitioning approaches half-portion partitioning and systematic grid partitioning in constructing the prediction models. In each approach the study area was partitioned into two groups for training and validation and then reversed to verify the partitioning approach. Bayes model had high accuracy and consistent results in both half-portion partitioning and systematic grid partitioning. Thus the Bayes model was a better option for landslide hazard prediction of the study area. Considering the climate change scenario A1B the landslide hazard map based on Bayes model estimated that the landslide occurrence rate remained high in the northern part of study area while that of the southern part decreased over time thereby creating a polarization between the two regions.

키워드