浙江省磐安县滑坡灾害空间预测

结合浙江省磐安县滑坡灾害发生的环境和基本特征,分析了影响灾害发生的各因素与灾害发生之间的相关性,提取出影响灾害发生的主要因素:地形、坡度、岩性、构造、水系、公路。利用GIS的空间分析功能将各因素图件栅格化为250 m×250 m的单元共19765个。采用二元逻辑回归方法对研究区滑坡灾害进行了定量的概率预测,为解决滑坡灾害定量风险预测研究提供参考方法,同时也做了危险性定性空间预测分区图,将预测结果与该区历史滑坡灾害发生情况进行对比发现,预测精度为75．8％。2004年8月浙江省遭遇“云娜”台风期间,浙江省突发性地质灾害预警预报系统采用该方法进行了试运行发现,预测范围内新发生的灾害点均落在了高、中易发区范围内。

LANDSLIDE HAZARDS SPATIAL PREDICTION IN PAN' AN COUNTY, ZHEJIANG PROVINCE, CHINA

Based on the environment and characters of landslide in Pan'an, Zhejiang Province, China, the importance of each factor-inducing landslide was analyzed. And it is concluded that landform, gradient, lithology, texture, river and excavation of road are the main factors. Using the spatial analysis technology of GIS, each map of factors is divided into 19765 units measured at 250 m×250 m. A statistic method called binary logistic regression has been used to create a landslide probability prediction map for Pan'an, which can be used as a reference to landslide risk prediction on a quantitative basis. A spatial prediction map with 4 classes (high, mid, low, non) was obtained at the same time. Contrasting to the occurrence of historical landslide hazards, the precision can be up to 75. 8% . The results are validated in typhoon Rananim, in August 2004, in Zhejiang Province, China. And it is found that almost all of the hazards happening during typhoon Rananim are all in the high and middle risk zones. And the prediction map can provide scientific references for land layout and geo-hazard controlling in Pan'an County.