论文部分内容阅读
针对地震滑坡规律复杂,预测难度大,而传统地震滑坡敏感性分析采样策略研究不足的问题,分析不同采样策略对地震滑坡敏感性分析的影响。以多边形缓冲区作为采样区,并将滑坡的发生区域及前两者总和2个采样策略用于对比实验。选取4·20芦山地震中受灾严重的宝盛乡作为研究对象,根据采样策略对滑坡各个评价因子进行采样,并构建支持向量机模型定量计算敏感性指数,再利用自然断点法生成敏感性区划图。通过多种方法对结果进行分析。其中,以多边形缓冲区为采样策略得到的滑坡正确率94.44%,受试者工作特征曲线下面积99.1%,种子单元面积指数综合评价效果最佳,3项结果均占优势。结果证明该采样策略是有效可行的,可为后续防灾减灾工作提供依据。
In view of the complicated law of landslide and the difficulty of prediction, the traditional method of analyzing seismic sampling sensitivity is not enough, and the influence of different sampling strategies on the sensitivity analysis of earthquake landslide is analyzed. A polygon buffer was used as a sampling area, and the landslide occurrence area and the sum of the two first two sampling strategies were used for comparison experiments. Taking BaoSheng township of 4.20 Lushan earthquake as the research object, the landslide sampling factors were sampled according to the sampling strategy, and the support vector machine model was used to quantitatively calculate the sensitivity index, then the natural breakpoint method was used to generate the sensitive regionalization Figure The results are analyzed in a variety of ways. Among them, the correct rate of landslide obtained by taking polygon buffer as sampling strategy was 94.44%, the area under the working characteristic curve of subjects was 99.1%, the comprehensive evaluation of seed unit area index was the best, and all the three results were dominant. The results show that the sampling strategy is effective and feasible, which can provide the basis for the subsequent disaster prevention and mitigation work.