小流域山洪灾害危险性分析之降雨指标选取的初步研究

Study on Rainfall Index Selection for Hazard Analysis of Mountain Torrents Disaster of Small Watersheds

降雨是诱发山洪灾害的直接动力和激发条件,是山洪灾害危险性分析中不可或缺的重要指标。对于不同的研究区,不同频次和历时的降雨对山洪灾害的影响不同,危险性分析所选取的降雨指标也可能不同。本文基于1:5万小流域数据、历史山洪灾害空间分布数据和暴雨图集资料,以江西省婺源县小流域为例,采用相关分析法对初步选取的24个降雨指标进行相关性分析,筛选出相关性不强的降雨指标;利用GIS技术和基于局部Getis-Ord Gi*算法的优化热点分析工具对历史山洪灾害点进行空间聚类分析,获得研究区每个小流域山洪灾害危险度的估值,该值能够较好地反映小流域山洪灾害危险性的空间分布;通过地理探测器模型中的因子探测器和交互探测器对已筛选出的降雨指标和小流域山洪灾害危险度进行综合探测分析,从而获取对该研究区小流域山洪灾害危险性影响较大的降雨指标,即100年一遇最大6 h降雨和100年一遇最大24 h降雨。本文研究对小流域山洪灾害危险性分析降雨指标的定量选取具有参考和指导意义。

Mountain torrents are one of the most dangerous weather-related natural disasters in the world. As the direct driving force and inducing factor of mountain torrent disasters, rainfall can be used as an indispensable index in the hazard analysis of flash flood. In different study areas, the rainfall with different return periods and durations may affect mountain torrent disasters differently and would be therefore selected disparately for the hazard analysis. However, there is little quantitative research on the principles of rainfall index selection in the previous literatures. Based on the small watershed map（1：50,000 scale）, the rainstorm atlas and the spatial distribution of mountain torrent disasters, this paper focused on the small watersheds in Wuyuan County in Jangxi Province and used GIS（Geographic Information System）, Correlational Analysis, Spatial Cluster and Geographical Detector to obtain the rainfall indices which had great impact on the mountain torrent disasters of the small watersheds. The selected rainfall indices will take part in the hazard analysis of mountain torrent disasters in the study area. The conclusion is drawn as follows：（1） Seven uncorrelated rainfall indices have been obtained from the original 24 rainfall indices by the correlational analysis method. This means that in order to avoid the information overlap, we need to analyze the correlation between pairs of the rainfall indices before the hazard analysis of mountain torrent disaster.（2） By using the Optimized Hot Spot Analysis tool based on Local Getis-Ord Gi＊ Algorithm to get the estimated hazard values of the mountain torrent disasters of small watersheds, the calculation results are in line with the spatial distribution of mountain torrent disasters in this study area.（3）It＇s better to introduce geographical detector to quantify the relevance between rainfall indices and mountain torrent disasters of small watersheds. Through the comprehensive analysis, the final rainfall factors, which cause mountain torrent disasters of the small watersheds with a great probability, will be derived and will participate in the hazard analysis of flash flood. The finally selected rainfall factors of small watersheds in Wuyuan County are the maximum rainfall with 100-year return period and 6-hour duration and the maximum rainfall with 100-year return period and 24-hour duration. The technical method in this paper possesses an important reference value and guidance to the quantitative selection of rainfall indices in the hazard analysis of flash flood of small watersheds.