Study On The Evolution Characteristics Of The Relationship Between Runoff And Sediment Load Above Longyangxia Reservoir In The Source Area Of The Yellow River

The region above longyangxia in the source region of the Yellow River was taken as the main research object.The runoff and sediment load of tangnaihai hydrological station and guide hydrological station of longyangxia reservoir from1960 to 2013 were collected to study the relationship between runoff and sediment load under natural and man-made environment changes.Firstly,the trend,persistence and multi-periodicity evolution characteristics of runoff and sediment were studied by using mann-kendall method,R/S method and CEEMDAN method respectively.The multi-time scale correlation evolution characteristics of the relationship between runoff and sediment load at tangnaihai hydrological station were analyzed by the double accumulation curve method,and the long-term equilibrium relationship and short-term fluctuation relationship in the micro-scale and macro-scale were analyzed by using the co-integration theory.The ARMA model was used to study the decomposition and combination prediction of the annual runoff at tangnaihai hydrological station,and the prediction accuracy was evaluated by using the Nash efficiency coefficient.At the same time,the entropy theory was used to analyze the uncertain relationship of runoff and sediment load at multi-time scales at guide hydrological station,and the influence of longyangxia reservoir operation on the relationship between runoff and sediment load was analyzed.Finally,the variable structure co-integration relationship between runoff and sediment load at guide hydrological station was analyzed by introducing the mutation point and using the theory of variable structure co-integration.The main conclusions are as follows:(1)The runoff and sediment load of the tangnaihai hydrologiacal station and the guide hydrological station both showed a downward trend.It is predicted that in the future,the runoff at tangnaihai hydrological station will still show a downward trend,the sediment load will show an upward trend,and at guide hydrological station will still show a downward trend.There is a good relation between runoff and sediment load at multi-time scales of the tangnaihai hydrological station,and the quasi-period of each IMF component is basically the same.Before the reservoir operation,the correlation between runoff and sediment load was good and the quasi-period was basically the same.The reservoir operation changed the correlation,and the relationship between runoff and sediment load and the fluctuation period at mult-time scales were changed.(2)The double cumulative curve corresponding to the IMF1 and IMF2components of the runoff and sediment load were well fitted,and the double cumulative curve corresponding to the IMF3 and IMF4 components had obvious mutation points.Due to the large errors in the component prediction at individual scales,the relative errors of the combined regression model prediction are large,while the average relative errors of the ECM model prediction are small,and the prediction accuracy of the runoff is high.In general,the relationship between runoff and sediment load of the original sequence is determined by the changes of the relationship between runoff and sediment load at multi-time scales,and the weak relationship between runoff and sediment load under the medium-long time scale will affect the relationship between runoff and sediment load of the original sequence.(3)Simulated prediction by using ARMA model,the corresponding component simulation forecast of NSE is larger,and increases with the increase of wave period,the model simulation of the annual average relative error was 8.70%,the model to predict the annual average relative error is 9.42%,the relative error is small,shows that simulation prediction precision is higher.On the whole,the high frequency components play an important role in the mode reconstruction and combinatorial simulation prediction.(4)For each IMF component,the multi-time scale entropy is the largest for IMF2 component of runoff before the reservior operation,and the entropy of IMF1component of sediment load is the largest,and the entropy of the IMF1 components after the reservior operation are the largest,indicating that the corresponding components carry more information and contribute the most to the original sequences.For the multi-time scale structural entropy of runoff and sediment load,the IMF1 and IMF2 components before the reservoir operation were transformed into the IMF1components after the reservoir operation,which played leading roles in the entropy change,and the entropy were larger after the reservoir operation,indicating that the reservoir operation made the fluctuation period of runoff and sediment load more complicated.To sum up,it is suggested that the relevant observation and research on the runoff and sediment load in the upper reaches of the Yellow River should be carried out in a quasi-periodic period of 4-7 years and 3-4 years respectively.(5)The corresponding R~2 of the comutation model in the three cases were improved.From the ecm(1)coefficient,the average relative error and the goodness of fit R~2 together,considering the constant term,trend term and the cointegration vector of the ECM model corresponding to model 3 has the strongest explanatory power,and can more accurately reflect the long-term equilibrium and short-term fluctuation relationship between runoff and sediment load than model 1 and model 2.However,it is worth noting that the model 2 considering the constant term and trend term also has a strong explanatory ability and a high prediction accuracy.It shows that the co-integration regression equation is mainly the constant term and the trend term have changed,considering the change of the co-integration vector term,the model simulation accuracy will be affected and improved,but the improvement and adjustment force is not big.