Hydrology Characterization Of The TMR Basin And The Study Of Mid-long Term Runoff Forecasting

In recent years,the Tumen River basin(the TMR basin)in the southeast of Jilin province occurred floods which influenced by the "Lion Rock" typhoon in 2016.In this basin,it occurred strong precipitation over the hundred years and caused the rise of the TMR water level.In that context,the hydrological characteristics of the TMR Basin and the long-term runoff forecast are studied in this paper.The hydrological characteristics of the TMR basin and the study of the medium and long term runoff forecast are based on the precipitation data from 1957 to 2015 and the runoff data from 1959 to 2015,which upstream of Kaishantun Station,the midstream of Hedong Station and the downstream of Quan River Station.By analyzing the statistics of the anomaly percentage and the fluctuation frequency of the difference product curve,the inter annual changes of precipitation,abundant runoff,flat and dry year of hydrologic station are summarized.The precipitation in three hydrological stations is in accordance with the trend of runoff,flat and dry change.Among them,Kaishan Village Hydrological Station continued to have a high water supply year of 13 years.The precipitation and runoff trend of hydrologic station were analyzed by 5-years moving average method and Kendall rank correlation test method.The research shows that the precipitation and annual runoff of Kaishan Village Hydrological Station are more obvious.The runoff test method of sliding T mutation analysis and variance analysis of the runoff cycle F test,three hydrological stations exist mutation year annual runoff,respectively in 1974,1985,1985.And there is the test cycle phenomenon in 1981 to 2015 period.This paper carrying out the mid-term and long-term runoff forecasting research of TMR Basin on the basis of the feature analyzes of precipitation and runoff TMR Basin.Tying in runoff data of TMR Basin features.The method of artificial neural network and multiple regression analysis are used to forecast runoff,which including monthly flow forecast and annual runoff total forecast.Based on two forecasting models of the comprehensive comparison and the evaluation of the results the total qualified rate of the two methods is over 80% and over 70% respectively,the artificial neural network model is relatively better,which can adapt to the runoff characteristics of the basin.The research work has reference value for the macro-control of water resources and disaster prevention in TMR basin.