Evolution Analysis And Multi Model Prediction Of Runoff In Shaanxi Section Of Hanjiang River

Water resources are indispensable elements for human life and social development,the formation and evolution of runoff profoundly affect the water resources system.Because runoff is affected by many factors such as meteorology,underlying surface and human activities,and the changing process often shows a highly nonlinear characteristics.In this paper,runoff sequence data of WuHou,HanZhong,YangXian and ShiQuan stations in Shaanxi section of Hanjiang River are used to analyzed the evolution law of runoff in this area.At the same time,different runoff prediction models were used to predict the annual runoff of YangXian station.And based on the integrated platform,the simulation system of runoff evolution and prediction in the Shaanxi section of the Hanjiang River is built.The main research results in this paper are as follows:(1)Analysis of change trend of runoff series in Shaanxi section of Hanjiang River.The trend of annual runoff series and flood season runoff series of the WuHou,HanZhong,YangXian and ShiQuan stations is analyzed by Mann-Kendall test,linear tendencies estimation,cumulative anomaly,the results showed that the runoff sequences of the four stations showed a no significant decreasing trend.Among them,the test statistics of HanZhong station were the smallest,the change rate was the smallest,and the trend change was the least obvious,the test statistics of ShiQuan station were the largest,the change rate was the largest,and the trend was the most obvious.The accumulative anomaly results of the four stations indicate that the runoff sequence has changed one or two times from rise to decline,but the overall trend is still downward.(2)Mutation point diagnosis of runoff sequence and persistence analysis of change trend in Shaanxi section of Hanjiang River.Using Pettitt and sliding T mutation test method,the results show that the mutation year results of the annual runoff series and flood season runoff series of the four stations are all around 1990.1n addition,Pettitt mutation test results also showed that the mutation point of WuHou and HanZhong station is not significant,the mutation points of YangXian and ShiQuan station passed a certain level of significance test.And the Hurst index of each station was calculated by R/S analysis to analyze the continuity of runoff change the results show that the runoff sequence of the four stations will keep decreasing in the future period.(3)Study on multi model annual runoff prediction in Shaanxi section of Hanjiang River.The ARIMA model,the MGF model the grey dynamic and the DenseNet model were used to predict the annual runoff of YangXian station.From the prediction accuracy and performance of the model,The ARIMA model is not ideal,the MGF model is better,the grey dynamic prediction model is good and the DenseNet model is the best,the relative error of one prediction is recorded as qualified within 20%plus or minus,the prediction rate of the model was 51.1%,72.9%,86.4%and 100%respectively.(4)Simulation and implementation of runoff evolution and prediction based on integrated platform.The simulation system for runoff evolution and prediction in the Shaanxi section of the Hanjiang River Based on integrated platform,the application example shows that the system has considerable adaptability and flexibility,it can respond well to changes in demand.The system intuitively analyzed the simulation results by charts and other forms,the results are visible and believable,it is also easy to promote and maintain.