关键词: 证据权, 危险性, 空间分析, 空间预测

Abstract: The evidence weights model based on the weights of evidence method was applied to study the zonation of the landslide geohazards in Hangzhou. The main data sources consisted of geological maps, digital elevation model(DEM), land use data and 1905 specical cases of geohazards collected from 1930 to 2009 in Hangzhou city. The evidence factors involve heavy rain, types and structure of rock, slope, slope direction and high slope, river network and Path buffer, which were derived from maps using Arcgis spatial analysis and information extraction. The weights for each evidence map were calculated using the frequency of known landslides in the classes of causative parameters; then the probability of landslide for arbitrary grid was produced by summation of assigned weight values using Arc-WofE extended module, achieving the goal of spatial prediction for potential landslide points. The accuracy verified by the spilt-sample validation method was 88.3%. The obtained results were well coincident with the occurrence of the known landslides in the study area. It is shown that the weights of evidence method could be evaluated objectively and quantitatively for multivariate analysis, as a basis of the choice of factors for landslide geohazards zonation. So the methodology is very useful and accurate, and it is worth popularizing and applying to the risk zonation and evaluation of landslide geohazards.

Key words: evidence weight, risk, spatial analysis, spatial prediction