摘要
将滑动窗口与Copula函数相结合,提出滑动Copula函数的降水和径流的关系变异诊断新方法。以渭河流域华县站1951-2011年的实测月降水量和月径流量序列为例,画出不同滑动窗口的序列相依关系图,在欧式距离评判标准下,拟合降水和径流关系变异前后的最佳Copula函数。结果表明,华县站降水和径流关系变异年份为1972年和1984年,降水和径流关系在1951-1972年和1985-2011年服从T-Copula函数分布,1973-1984年服从Clayton-Copula函数分布。Copula函数模拟二元变量的诸多优点,有助于进一步认识降水和径流的尾部相关关系,为随机变量关系的变异研究提供一种新的途径。
A diagnostic method of abrupt changes combining sliding window and Copula function is presented. This method draws dependency diagram in different sliding windows using monthly series of precipitation and runoff at the Huaxian station of the Wei River, and it fits best Copula functions of precipitation and runoff based on an Euclidean distance criterion. Results show that in their abrupt change periods of in 1972 and 1984, precipitation and runoff followed a relationship of T-Copula function in 1951-1972 and 1985-2011 and of Clayton-Copula function in 1973-1984. Advantages of the Copula function are helpful in simulation of bivariate relationship and deeper understanding of the tail dependence of precipitation-runoff relationship, which would provide a new approach to variables relationship for random abrupt changes.
出处
《水力发电学报》
EI
CSCD
北大核心
2014年第6期20-24,60,共6页
Journal of Hydroelectric Engineering
基金
国家自然科学基金重大项目(51190093)
国家自然科学基金(51179148)
国家重大基础研究973(2011CB403306)
