基于降雨类型直方图分析的降雨站点相似性研究

为研究降雨观测站点间的相似性,提出了基于单场降雨类型直方图分析的降雨站点相似性比较模型。首先提取降雨时间序列中的雨量项,经过数据预处理,选择和提取单场降雨特征,并将这些特征统计量进行数据标准化处理;采用基于Davies-Bouldin指数聚类有效性指数的K-means方法进行聚类,最后对聚类结果进行基于单场降雨的降雨类型直方图相似性分析。实验结果符合相关区域的站点相似性实际情况,表明该相似性模型是可靠的,具有较好的应用价值。

Study on Similarity Model of Precipitation Series Based on Precipitation Type Histogram

Similarity model for precipitation stations based on analysis of the single precipitation type histogram was proposed to study the similarity between the precipitation stations. Firstly, the precipitation item was extracted from the precipitation time series, the features of single rainfall were chosen and extracted after data preprocessing. Then, the data standardization was conducted on the feature vectors. Further more, the standardized feature data were clustered using K -means clustering method based on the Davies -Bouldin index cluster validity index. Finally, the clustering results were analyzed using the single precipitation type histogram similarity model. The experimental results show that the similarity for the stations in the related regions is consistent with the underlying fact, which verifies that the proposed model is of great value for water resources management.