基于主成分分析的流域聚类研究

针对流域聚类研究中流域特征指标变量过多会增加计算的复杂性和指标间存在相关性使信息重叠导致计算结果失真的问题,采用主成分分析的方法将原特征指标综合为少数几个不相关的主成分,且提取出的主成分基本包含原特征指标的全部信息,并根据各流域在主成分上的得分值作为新样本进行流域聚类研究,找出相似流域,实现了无资料地区的参数移植。实例应用结果表明,该方法可行、有效。

Research on Watershed Clustering Method Based on Principal Component Analysis

In study of watershed clustering, choosing too many watershed characteristic indexes may increase computational complexity and overlapped information can lead to the distorted results. The principal component analysis method is applied to study the watershed clustering, which integrates the original indexer into few numbers of uncorrelated principal components. The extracted principal components include the whole information of original characteristic indexes. The similar watershed is found out by taking the score of the principal component on the watershed as new clustering sample, which achieves parameters transplantation of data gap area. The instance results show that the proposed method is feasible and effective.