顾及空间非平稳性的地质灾害易发性评价

Evaluation of geological hazard susceptibility taking into account spatial non-stationarity

传统的地质灾害易发性评价,通常将整个研究区进行全局的易发性评价,但忽略了空间非平稳性带来的局部差异。为了探究由于地理位置变化而引起的相同评价因子在不同局部区域对地质灾害发生的重要程度的变化,提高地质灾害易发性评价的精度,采用地理加权回归(Geographically Weighted Regression,GWR)模型将研究区进行区域尺度分割,以得到各评价因子空间自相关性较低的局部区域,并选取坡度、坡向、高程、岩组、距断层距离、距水系距离、距道路距离、年平均降雨量8个评价因子建立地质灾害易发性评价指标体系。采用信息量模型对全局和局部区域分别进行易发性评价,得到全局和区域尺度分割的易发性评价结果。结果显示:与全局模型相比,GWR-信息量模型具有更好的预测性能(AUC为0.896);GWR-信息量模型充分考虑了不同区域内的地质灾害影响因子对地质灾害易发性的影响,削弱了某些在全局模型中影响性较高但在局部区域影响性较低的因素对评价结果的影响。因此,运用GWR-信息量模型进行易发性评价更符合实际,更适于指导研究区风险管控与防灾减灾工作。

Traditional evaluation of the geological hazard susceptibility usually makes a global evaluation and susceptibility division of the entire study area,but ignores the differences between local areas caused by spatial non-stationarity.To research the changes in the relationship or structure between evaluation factors caused by changes in geographical location,and the importance of the same evaluation factors in different local areas for the occurrence of geological hazards,and to improve the accuracy of geological hazard susceptibility evaluation,the geographically weighted regression model(GWR) is used to split the study area at a regional scale,and seven local areas with low spatial autocorrelation of each evaluation factor were obtained,eight evaluation factors including slope,aspect,elevation,rock group,distance from faults,distance from the water system,distance from the road and annual average rainfall were selected to establish the evaluation index system of geological hazard susceptibility,and based on the neural network model,the change of the importance of each evaluation factor globally and in each local area were calculated.The global and seven local areas were evaluated for susceptibility using the information value model,and the global susceptibility evaluation results and regional-split susceptibility evaluation results were obtained and compared.The results show that the GWR-information value model has better prediction performance(AUC is 0.896).Through the comparative analysis of the results,the GWR-information value model takes into account the influence of geological hazards in different regions on the susceptibility of geological hazards,weakening the influence of evaluation factors that are more important in the global model but less influential in the local area on the evaluation results.The importance of evaluation factors in each local area changes after a regional split,which is more in line with local characteristics and improves the accuracy of geological hazard susceptibility evaluation.Therefore,the susceptibility evaluation using the GWR-information value model is more in line with the reality of the study area and is more suitable for guiding risk management and hazard prevention and mitigation.