Application of data-driven evidential belief functions to landslide susceptibility mapping in Jinbu, Korea

Evidential belief function (EBF) model was applied and validated for analysis of landslide susceptibility in the Pyungchang area of Korea using geographic information system. Areas of landslide occurrence in the study area were determined from the interpretation of aerial photographs and subsequent field surveys. Landslide locations were randomly allocated for landslide susceptibility map generation (70%) and validation (30%) purposes. Maps relevant to landslide occurrence (topography, geology, soil, and forest cover) were assembled in a spatial database, from which 17 landslide-related factors were extracted. The relationships between the observed landslide locations and these factors were identified and quantified using the EBF model. Three relationships were calculated: disbelief (Dis), uncertainty (Unc), and belief (Bel). The quantified relationships between each factor and landslide locations of each factor with known landslides were then used as factor ratings in an overlay analysis to create landslide susceptibility indices and maps. The most representative of the resulting susceptibility maps (the Bel map) was validated using the landslide data reserved for validation. The landslide susceptibility map demonstrates 85.96% accuracy. Thus, the EBF model was found to be effective in terms of prediction accuracy. (C) 2012 Elsevier B.V. All rights reserved.