图们江流域洪水灾害危险性评价实验设计研究

Experimental design and study on risk of flood disaster assessment in Tumen River Basin

为探索图们江流域洪水灾害危险性,该研究设计了基于GIS和层次分析法、信息量法和随机森林算法的洪水灾害危险性评价试验。试验主要包括选取洪水灾害危险性评价因子、构建洪水灾害危险性评价模型、危险性评价、精度对比分析和评价以及洪水灾害危险性区划制图等。洪水灾害危险性使用8个因子进行研究,包括地形地貌因子(高程、坡度、地形起伏度)、水文因子(河流缓冲区、河网密度)、土地利用因子、土壤因子和降雨因子。模型的训练和验证采用了156个已发生洪水事件和102个未发生洪水事件的数据,其中,将总数据的70%作为训练数据,30%作为验证数据。结果表明:利用层次分析法、信息量法和随机森林算法获得的洪水灾害区划图中,最危险区域的面积分别为4 311.84,3 980.16和4 311.84 km^(2),每个区域中检测到的洪水灾害点的占比分别为61%、70%和65%,其中,信息量法监测出的最危险区域的面积最小,灾害点百分比最大,说明信息量法的精度优于其它2种方法。该研究可以为受洪水影响地区的土地利用规划、防灾治灾及居民区与生命线工程保护提供数据依据。

In order to explore the flood hazard in the Tumen River basin,a flood hazard assessment experiment based on GIS,AHP,information volume method and stochastic forest algorithm was designed.The experiment mainly includes the selection of flood disaster hazard assessment factors,the construction of flood disaster hazard assessment model,hazard assessment,precision comparison analysis and evaluation,and flood disaster hazard zoning and mapping.The flood disaster hazard is studied using 8 factors,including topography(elevation,slope,topographic relief),hydrology(River buffer zone,river network density),land use,soil type and rainfall.The data of 156 flood events and 102 non flood events are used for the training and validation of the model,of which 70% is the training data and 30% is the validation data.The results show that the area of the most dangerous area in the flood disaster zoning map obtained by AHP,information method and random forest algorithm is 4 311.84 km^(2),2 3980.16 km^(2) and 4 311.84 km^(2) respectively,and the proportion of flood disaster points detected in each area is 61%,70% and 65% respectively,indicating that the accuracy of information method is superior to the other two methods.This study can provide data basis for land use planning,disaster prevention and control,and protection of residential areas and lifeline projects in flood affected areas.