泥石流危险度评价的距离判别模型及应用

根据距离判别分析法原理, 建立了距离判别模型对泥石流危险度进行评价。综合已有的文献资料, 选择20条典型的泥石流沟作为距离判别分析的训练样本, 以影响泥石流形成的地形、地质、水文气象以及人类活动等10个主要因素为评价指标, 并对样本指标数据进行了归一化处理, 同时采用基于离差平方和最大的最优组合赋权方法对样本评价指标数据进行优化改善, 消除了主观的随意性和客观的奇异性, 使得样本指标数据更加科学合理。最后建立了泥石流危险度评价的线性判别函数, 并采用误差率回代估计法进行检验, 检验结果表明该模型判别能力强。将该模型应用于工程实例中, 对云贵高原地区6条泥石流沟危险度进行评价, 研究结果表明: 这6条泥石流沟危险度与实际相吻合, 并且与其它文献中提供的方法评价结果完全一致。

Model of Distance Discriminant and Its Application in Debris Flow Hazard Assessment

According to the distance discriminant analysis theory, a model of distance discriminant was established for debris flow hazard assessment. Based on the available documents, 20 typical debris flow gullies were chosen as samples for the model, and 10 main factors connected with the formation of debris flow were selected as indexes, including topography, geology, hydrology, meteorology and human activities and so on. In order to eliminate the subjective arbitrariness and objective singularity, sample data were normalized and improved with an optimal weights combination method based on sum of squares of deviations, resulting in data indexes more scientific and rational. At last, linear discriminant functions were established for debris flow hazard assessment, and a method of error rate estimation was used for results verification, improving the well-establishment of the model. This model has been applied into engineering practice for hazard assessment of 6 debris flow gullies in the Yunnan-Guizhou plateau areas. Results showed the assessment was consistent with actual conditions, as well as the evaluation results obtained by methods from other literature.