水库泥沙来源定量判别方法的判别能力对比

Comparative analysis of discriminant ability of quantitative discriminant models for reservoir sediment sources

水库淤积问题是干旱、半干旱沙区水库面临的主要环境问题之一。物源定量判别法是识别水库泥沙来源的新兴手段,而筛选某一区域泥沙来源的最优判别方法是精准识别物源区的重要前提。选取党河水库作为研究对象,基于野外调查、室内分析、模型模拟等手段,对比分析了多组复合指纹法、最优复合指纹法和距离法在党河水库泥沙来源判别过程中的适用性。结果表明:在该研究区多组复合指纹法和最优复合指纹法判别结果较为理想,且判别结果接近,而距离法表现不理想,有待进一步检验。增加复合指纹因子组合数(19组)能使95%的置信区间变窄,理论上具有统计意义。从统计学角度分析,多组复合指纹法表现更好一些,将在一定程度上提高定量判别结果精度。

Reservoir siltation is one of the main environmental problems in arid and semi-arid sandy areas.The sediment source fingerprinting is a new method to identify reservoir sediment sources,and screening an optimal discriminant model for sediment sources in a certain area is an important prerequisite for accurate identification of sediment source.In this study,Danghe Reservoir was selected as the research object.Based on field investigation,indoor analysis,model simulation and other means,the applicability of multiple composite fingerprints,optimal composite fingerprint,and distance weight method in the identification process of sediment source were compared and analyzed.The results showed that both multiple composite fingerprints and optimal composite fingerprint could obtain reasonable contributions and were relatively close,which were suitable for the quantitative identification of sediment sources in this watershed.However,the distance weighting method was not satisfactory and needed to be further tested.Increasing the number of composite fingerprints(19)could narrow the 95%confidence interval,which has statistical significance.From a statistical point of view.Theoretically the multiple composite fingerprints perform better,which will improve the accuracy of quantitative discrimination results to a certain extent.