摘要: |
【目的】分析不同Copula函数参数估计法对函数类型选择及对河流年径流丰枯遭遇的影响,为西安水资源开发、供水调度管理提供理论依据。【方法】利用黑河黑峪口站、沣河秦渡镇站、灞河马渡王站3个水文站55年的年径流数据,基于Kendall系数估计法和极大似然估计法分别估计Copula函数的参数,选择拟合程度好的GH Copula函数,对比分析两种估计法推求的黑河、沣河、灞河的年径流丰枯遭遇频率。【结果】两种参数估计法均表明GH Copula函数是拟合程度最好的函数。极大似然估计法的拟合结果略优于Kendall系数估计法,但两者之间相关系数的绝对差≤0.02%。两种方法估计的河流间两两丰枯同步和丰枯异步的概率≤3%,沣河、黑河、灞河3条河流两两丰枯同步的概率远远大于丰枯异步的概率;丰枯同步的概率中,两条河流同平的概率几乎是同丰和同枯的概率和,同枯的概率略小于同丰的概率;在丰枯异步的概率中,2条河流一丰一枯的概率最小,其中灞河和沣河一丰一枯的概率只有0.048%~0.138%。【结论】在西安供水河流年径流丰枯遭遇分析时,极大似然估计法的拟合结果略优于Kendall系数估计法,但两种参数估计法都是可行的。 |
关键词: 径流 丰枯遭遇 Copula函数 Kendall系数 极大似然估计 |
DOI: |
分类号: |
基金项目:教育部国家外专局高等学校学科创新引智计划(111)项目(B08039);中国工程院咨询研究项目(2014-07-XZ-002);中央高校基础研究项目(310829161010);中央高校基本科研业务专项(310812161007) |
|
Analysis on coincidence of high and low stream flows for water supplying rivers in Xi’an based on Copula function |
YOU Qiying,ZHOU Yanlong,LIU Yan,et al
|
Abstract: |
【Objective】This paper analyzed the influence of different parameter estimation methods of Copula function on selection of function type and high and low stream flows of river annual runoffs to provide theoretical basis for exploiting water resource as well as dispatching and managing water supply.【Method】The Kendall coefficient estimation method and maximum likelihood estimation method were applied to estimate the parameter of Copula function.The GH Copula function with fine goodness fit was selected to analyze the frequencies of high and low stream flows of water supplying rivers in Xi’an based on 55 years annual runoff data in three hydrologic stations including Heiyukou Station of Hei River,Qindu Station at Feng River and Maduwang Station at Ba River.【Result】Both methods showed that GH Copula function had the best goodness fit.The fit result of maximum likelihood estimation method was slightly better than Kendall coefficient estimation method.However,the absolute difference in correlation coefficients between fit results was ≤0.02%.The synchronous and asynchronous frequencies in pairs of rivers estimated by both methods were ≤3%.The synchronous frequency was much greater than the asynchronous frequency among Feng River,Hei River and Ba River.The synchronous frequency of normal flow for two rivers almost equaled to the sum of high flow and low flow,and that of the low flow was slightly less.The asynchronous frequency of two rivers with high and low flows was the lowest,with the minimum value of 0.048%-0.138% between Ba River and Feng River.【Conclusion】The fit result of maximum likelihood estimation method was slightly better than the Kendall coefficient estimation method and both methods were feasible to analyze high and low stream flows among water supplying rivers in Xi’an. |
Key words: river runoff high and low stream flows Copula function Kendall coefficient Maximum likelihood estimation |