Dependent Structure Model For Runoff Forecasting Errors In Multiple Forecast Periods Based On Nested Copula Function

Under the background of global climate,when forecasting the inflow runoff,it often forecasts the runoff value in multiple forecast periods to reflect the inflow situation in the future as much as possible,so as to provide the basis for the formulation of reservoir operation plan.However,due to the inevitability of the errors in the process of runoff prediction and the complex dependence of hydrological information in time and space,it is of great theoretical and practical significance to carry out the research on the dependent structure model of runoff prediction errors in multiple prediction periods for the reservoir to maximize the utilization efficiency of water energy resources.In this paper,the runoff forecast errors of Jinxi reservoir in Yalong River Basin and Jiulong River area in different forecast periods are studied.The copula function is used to build different models to study the correlation structure of runoff forecast errors in different forecast periods.Firstly,the commonly used distribution curves,such as normal distribution,t distribution,logistic distribution,stable distribution and generalized extreme value distribution,are used to optimize the distribution of runoff forecast errors in the four forecast periods(6 h,12 h,18 h and 24 h)of Jinxi inflow and Jiulong River inter regional inflow.Secondly,according to the characteristics of binary Gumbel copula,Clayton copula and frank copula,the linear combination is carried out to get M-Copula.M-Copula function is used to construct the structure model of runoff prediction error dependence of the two prediction periods(such as Jinxi 6 h inflow and interval 6 h inflow).The M-Copula function is used for stochastic simulation and the probability of occurrence of the random combination of runoff prediction errors in the two prediction periods is calculated.Finally,on the basis of binary copula function,the dependent structure model of runoff prediction error in multi prediction period is established based on hierarchical nested copula function.The hierarchical nested copula function is used to simulate the runoff forecast error in different forecast periods,which is coupled with the forecast value to get a series of runoff processes closer to the actual inflow,so as to provide a reference for the reservoir operation plan.