Abstract: Weights of evidence (wofe) modeling and its application based on GIS are introduced. Wofe is applied to the zonation of karst collapse. Sensitivity of karst collapse around West Lake of Hangzhou is analysed. The cell size used in the analysis is 10 m×10 m. Spatial data of strata, structure and overburden is extracted from the study area. Spatial association between evidences and karst collapse is analysed, and contrasts are calculated. The studentized contrasts, which serve as a guide to statistical significance of spatial association, give cut off values to convert multiclass evidential data into binary pattern. Five evidences appropriate to the model are finally selected and converted into binary patterns, such as the strata unit, strata boundary of Gufeng formation - Qixia formation and of Huanglong formation - Laohudong formation, proximity to EW-NW trending structure features, proximity to NE trending structure features. Posterior probability of each cell which reflects degrees of karst potential collapse is calculated. The ratio of numbers of known collapses to the numbers of predicted collapses is 0.864, which indicates conditional independence among evidences with respect to karst collapse. The special application of wofe shows that the model is adapted to zonation of karst collapse.

Key words: weights of evidence, GIS, karst collapse, zonation, prediction

摘要： 介绍了证据权模型及 GIS 应用原理, 并引入到岩溶塌陷危险性区划中。在杭州西湖周边岩溶塌陷危险性分析中, 将研究区划分为 10 m ×10 m 的网格单元, 并提取地层、构造等证据图层, 分析各证据与岩溶塌陷发生的空间相关性, 计算各证据对塌陷发生的对比值, 剔除显著性水平不满足要求的证据, 并且确定了对具有分级特征的证据进行二值化时的切值, 进而选取了地层单元、孤峰组-栖霞组地层界线、黄龙组-老虎洞组地层界线、距 EW-NW 向构造线的距离、距 NE 向构造线的距离等5个证据, 求得每个单元网格的后验概率, 并计算塌陷点已知与预测的个数之比为 0. 864, 认为符合条件独立性检验, 最后的应用表明, 这种数据驱动的方法同样适用于岩溶塌陷区划问题的研究。

关键词: 证据权, GIS, 岩溶塌陷, 区划, 预测