Research On Runoff Evolution In Mountainous Area Under Climate Change In Manas River

The dramatic changes in the climate caused the variation in the sequence of the Manas River,and the hypothesis of consistency has been questioned.Based on the hydrometeorological data of the outcoming pass of the upper reaches of Manas River in 1956-2014 a,this paper analyzes the statistical characteristics of hydrometeorological data,and combines the historical survey data to diagnose the comprehensive variation of hydrological series,and constructs the mixed distribution model,the conditional probability distribution model and the time-varying moment model with time and climate factors as covariables are established based on GAMLSS theory,and compared with the traditional P-? distribution,the univariate frequency analysis of runoff series is carried out.Determining the marginal distribution of runoff characteristic variables,combined with Copula connection functions,a nonuniform multivariate joint distribution model is finally constructed,and joint frequency analysis is carried out.The contents and results of the study are as follows:(1)Statistical analysis of runoff series.The non-uniform coefficient of annual runoff series and the complete regulation coefficient of annual distribution show a decreasing trend,and the distribution of runoff gradually tends to be uniform.The concentration degree of distribution is about 66.86% during one year.With the passage of time,the concentration period of river basin is mainly in early July,and the concentration period of different years shows an earlier trend.The evolution of annual runoff series has the oscillation periods of 30 ~ 60 years,10 ~ 30 years and 2 ~ 6 years,respectively.Among them,45 years and16 years were the first,and the second main period controlled the variation of runoff series in the whole time period.(2)Runoff sequence variation diagnosis.The Hurst index analysis showed that there was a huge variation in the annual runoff sequence,and the sliding t test showed that there were two significant mutations in the annual runoff sequence,of which 1995 is a catastrophic year.Mann-Kendall test showed that the variation trend of the whole runoff sequence of Manas River was not significant,but the change trend of sub-sequence was significant before and after the variation point.The diagnosis showed that 1995 was a reliable year for the variation of Manas River runoff sequence.Combined with the physical origin of variation and field investigation and analysis,the diagnosis showed that 1995 was a reliable year for the variation of Manas River runoff sequence.(3)Analysis of annual runoff based on the hydrological statistical model.Through the goodness of fit test and observation of the distribution fitting diagram,the mixed distribution model is better than the P-?distribution model and the conditional probability distribution model.Because the traditional P-?distribution parameters are constant,the quantile grayscale map is horizontal linear,and it has a poor fitting with empirical points.The GAMLSS model with ATD exponent as covariable can explain the dynamic process of annual runoff series and the mechanism of physical runoff more effectively.Under different design guarantee rates,the design annual runoff of the GAMLSS model with ATD as a covariable is larger than the traditional P-? distribution by 3.08% ~ 16.10%,and is larger than that of the conditional probability distribution and the mixed distribution model by 8%~14%.The inconsistency effects of climate change are significant.(4)Multivariable Joint probability Analysis based on the Copula function.When the multivariable connection function is Gumbel-Copula,the multivariable joint distribution of runoff characteristics is best fitted.The R-P-ATD sequence is divided into two sub-sequences in the year of variation,1995,however,the statistics have not passed the significance test.Therefore,the subsequences are consistent with the consistency hypothesis.The uncertainty of Copula modeling is quantified by MCMC simulation,and the optimal values of MCMC estimation of Copula function parameters before and after the variation point are1.35 and 1.38 respectively,and the uncertainty ranges are [1.273,1.460] and [1.221,1.599] respectively.It is found that the design runoff with a higher design guarantee rate can be achieved with only a small amount of rainfall after the R-P-ATD three-variable joint probability analysis and the calculation of the combined recurrence period show that the variation point needs only a small amount of rainfall.This is due to the increasing impact of climate on runoff in the Manas River basin.The combined probability analysis of R-P-ATD and the calculation of the combined recurrence period show that the influence of climate change on runoff in Manas River basin is becoming more and more serious,because the unique geographical environment and runoff generation mechanism make runoff more and more dependent on climate change.