Methods For Detecting The Variation Of Correlation Relationship Between Hydrometerological Elements And Their Application

In the recent hundred years,global climate has undergone dramatic changes.Coupling the impacts of large-scale human activities,the hydrological cycle changed remarkably,which greatly challenges the classical studies on hydrology and water resources.Disclosing the evolution mechanism of hydrological cycle under the changing environment is of great significance for water resources management development and cognizing the mechanism of hydrological cycle.Traditional studies focus on detecting the variation of the univariate and neglects the variation of the relationship between hydrometerological elements within the hydrological cycle system.Thus,they failed to present the hydrological cycle mechanism systematically and comprehensively.Therefore,this study involving the Wei River Basin proposes three methods to detect the changes of the precipitation-runoff relationship,potential evapotranspiration-runoff relationship,potential evapotranspiration-precipitation relationship and the runoff-sediment relationship in the Weihe River Basin.The three methods are running correlation coefficients,permutation entropy-based method and scaling exponents-based method.The main research results of this paper are as follows.(1)Through maximum entropy,Mann-Kendall test and rescaled range methods,this paper analyzed the evolutions of periods,trends,and abrupt changes of precipitation,runofif,air temperature and sediment in different sub-areas of Weihe River Basin.The precipitation had non-significant decreasing trends,while annual runoff and sediment showed significant decreasing trends,and air temperature persistently grew in the overall basin.Except for the upstream of Beidao,the other sub-areas of Weihe River basin had no abrupt changes.The abrupt moments of runoff and sediment series were mainly distributed at the middle of 70s and the middles of 80s at the 20th century.(2)Based on the sliding window technique,this paper proposed a correlation coefficient variation diagnosis method,which was applied to analyze the correlation coefficient variations of the relationship between precipitation,runoff,potential evapotranspiration and sediment in the Weihe River Basin and one sub-basin in the Yellow River,i.e.the Ningxia Mongolia reach.The precipitation-runoff relationship changed around 1970 and 1997.Change points of the potential evapotranspiration-runoff relationship are similar to that of the precipitation-runoff relationship.The potential evapotranspiration-precipitation relationship has altered around 1980 and 1997.Change points of the runoff-sediment relationship at the Zhangjiashan station locate in 1977 and 1989,while that at the Zhuangtou station are 1982 and 1997.The runoff-sediment relationship in Ningxia Mongolia reach has changed in 1968 and 1995.Human activity such as construction and operation of upstream large reservoirs and the climate change such as activity and flood,ENSO were two main reasons of abrupt changes of correlations between different hydrological factors.(3)The permutation entropy-based and scaling exponents-based methods for detecting the relationship between bivariate variables are utilized to investigating the relationship between hydrometerological elements in the Weihe River Basin.Compared with detection results of sliding correlation method,we found that the results present difference and similarity among the three methods.The difference of the focus of different methods leads to the difference of different methods.Here,the sliding correlation method characterizes the synchronous changes of bivariate variables,while the permutation entropy-based and scaling exponents-based methods focus on the complexity and long range correlation of the variable itself.(4)The major innovation of this paper is to propose three methods to detect the change point of the relationship between two variables and the application of these methods to the detection of the relationship between hydrometerological elements in the Wei River Basin.From the multivariate perspective,systematically and compressively revealing the variations of the hydrological cycle is our primary contribution.In addition,these methods can provide reference for other subjects tending to investigating the changes of the relationship between multivariate.