PREDICTION MODELS IN SPATIAL DATA ANALYSIS FOR GEOLOGIC HAZARDS

C. F. Chung (Geological Survey of Canada, 601 Booth Street. Ottawa, Ontario, Canada K1A 0E8)

To construct prediction models for geologic hazards using GIS-based spatial data, three mathematical frameworks, namely probability theory, Zadeh's fuzzy set theory and Dempster-Shafer's evidential theory, have been studied. The models have been built to incorporate both the expert's knowledge and the quantitative relationships between the past occurrences of geologic events and the supporting spatial data. In addition, several comparison and evaluation techniques have been developed to test and validate the performance of the models as prediction tools identifying vulnerable areas for the occurrence of future geologic hazardous events. A landslide hazard study in the Rio Chinchina area in Colombia is used to illustrate the techniques proposed.