西藏喜马拉雅山地区冰湖溃决的预测模型及其应用研究

以西藏喜马拉雅山地区的冰湖为研究对象，基于现有的冰湖溃决预测方法，提出了建立冰湖溃决预测方法的关键点，即选取的指标必须能够体现冰湖的动态变化特征.在定量分析的建模过程中应该采用不确定性的数学理论，对于冰湖溃决可能性的等级划分需要进行合理性及实用性验证.选取坝顶宽度、湖水面距坝顶高度与坝高之比、冰湖面积和补给冰川面积为预测指标，通过对西藏喜马拉雅山地区29个冰湖样本进行逻辑回归分析，建立了冰湖溃决的预测模型，并用所有样本进行了交叉验证.结果表明：该模型能够在分类应用中取得较好效果，根据溃决冰湖累积百分数随冰湖溃决可能性大小的变化曲线，将冰湖溃决的可能性划分为四个等级.以黄湖为例，把湖水面距坝顶高度与坝高之比作为冰湖溃决的诱变指标，分析了冰湖溃决可能性大小的变化规律.结合现有的冰湖溃决预测的定性方法，讨论了所建立的冰湖溃决预测模型的优点和缺点.

Prediction model and its application for glacial lake outburst in the Himalayas area,Tibet

Taking the glacial lakes as research object in the Himalayas area,Tibet,this paper firstly extracts three critical points in the process of making a glacial lake outburst prediction model,that is,the prediction in-dex must be able to reflect the dynamic characteristics of the glacial lakes,the uncertainty mathematics theory should be employed to model in the quantitative analysis,and the grade classification of the probability of the glacial lake outburst need to be validated. After that,we choose the prediction variables as following,lake crest width,ratio of the lake freeboard to moraine crest height,lake area,and glacier area,and propose a prediction model based on the logistic regression analysis of 29 glacial lake samples. we found the model is effective with the cross-validation. According to the curve of the cumulative percentage of drained lakes versus outburst proba-bility,the glacial lake outburst probability is divided into four grades. In the case of the Yellow Lake,we ana-lyzed the variation of the glacial lake outburst probability when the ratio of the lake freeboard to moraine crest height is considered as the precipitating factor. Finally,in view of the existing qualitative methods,we have dis-cussed the advantages and disadvantages of our model.