基于逻辑回归法预测黄河口尾闾河道出汊概率

Branch probability prediction on the Yellow River estuary based on logistic regression method

研究黄河口尾闾河道出汊的孕育过程及触发机制,对维持黄河口流路稳定和促进河口区域生态经济发展具有重要意义。以黄河口概化实体模型试验数据和2004年实测数据为基础,结合前人的研究成果,选取流量、含沙量、潮流动力、沙嘴形态参数、河道比降和断面河相系数等6个参数作为评价指标,采用逻辑回归法建立黄河口尾闾河道出汊的概率预测模型,并对模型进行了交叉验证分析。结果表明:基于分类阈值为50%的逻辑回归模型出汊试验样本预测的准确率为70%,对未出汊试验样本预测的准确率为94.62%,对所有试验样本预测的准确率为90.27%。最后,利用该模型对2007年黄河口尾闾河道的出汊过程进行了概率预测,其结果与实际情况基本吻合。该模型主要用于评估行水流路的稳定性,可以为黄河口流路规划提供技术支撑。

Based on model experimental data and measured data in 2004 of the Yellow River estuary,six variables including flow,sediment concentration,tide power,sand shape parameter,stream gradient and fluvial coefficient were selected as indexes.A prediction model of the Yellow River estuary is established by logistic regression method.The model is cross-validated.Result shows that the accuracy of prediction of the branching experimental samples with a classification threshold of 50%is 70%,the accuracy of prediction of the experimental samples without branching is 94.62%,and the accuracy of prediction of all the experimental samples is 90.27%.In addition,the probability of the branch for the Yellow River estuary in 2007 is predicted by this model,and the result is consistent with the actual situation.The model is mainly used to evaluate the stability of the flow path and can provide technical support for the flow path planning of the Yellow River estuary.