Influence And Prediction Of Climatic Index On Water Resources In The Middle And Upper Reaches Of The Yellow River

As China’s economy advances swiftly and social production and people’s needs for sustenance are continually enhanced,the judicious exploitation and utilization of water resources become increasingly essential.Governments at all levels and relevant departments in the middle and upper reaches of the Yellow River not only require hydrology bureaus at all levels to provide short-term hydrology,but also hope that they can successfully and accurately predict the long-term hydrology and water resources prediction information of each basin and the medium’s dependability makes it the ideal foundation for the most advantageous apportionment of water resources,river ecological revival,flood control and energy production,irrigation and transportation.-and long-term hydrological prediction,accurate and timely prediction of possible changes in the future can help us develop effective coping strategies more quickly,realize effective utilization of water resources,and contribute to the orderly distribution and effective management of water resources.With the increasing demand for medium-and long-term hydrological forecasting,^[1].With the development of technology,how to make full use of the existing hydrological data enhancing the precision of medium and long-term runoff prognostication.has become an important research topic for hydrologists.To this end,researchers have studied a variety of models to meet different application requirements,thus bringing important practical value to the medium and long term runoff forecast^[2].As the research object,this paper examines the measured:Tangnaihai’s source station,Lanzhou’s upper station,and the middle station.The BP（Back Propagation）neural network model,LSTM（Long Short-Term Memory Long and Short Term memory network）model and Gated recurrent units（cells）GRU forecast model are analyzed comprehensively.Through the study of the monthly scale runoff forecast model,we found some more effective methods,which can help us find more suitable models for medium and long term hydrological forecast,and offer a valuable reference for enhancing forecast precision.Its main research findings and accomplishments are as follows:（1）To uncover the evolution law and variation features of long-and extended runoff time series,M-K mutation test,Morlet wavelet transform and Hurst index are utilized to contrast and evaluate the periodic fluctuation nonlinear trend change and mutation characteristics of runoff at Tangnahai hydrological Station,Lanzhou hydrological station and Tongguan hydrological station.The results of the study demonstrate that runoff in the three areas has a distinct periodic alteration pattern,and the runoff at the three sites is downwardly inclined.Thus,mastering the characteristics of runoff time series is essential.Throughthcomprehensive analysis of the three characteristics o wavelet ref runoff change,we can predict the future runoff more accurately.（2）Using WPS,M-K mutation test method and Morlet wavelet transform,the trend chart,M-K mutation test chart,al part coefficient contour map and wavelet square difference chart of11 climate indexes in the whole time series were drawn respectively,and the trend analysis,mutation analysis and period analysis were carried out.（3）Conduct correlation analysis on meteorological factors of hydrology station（represented by the mean value of meteorological factors of surrounding meteorological stations）,11 climatic indexes,climate index,distance from planet to Earth and distance from Sun to Earth with monthly runoff of Tangshan Naihai Hydrology Station,Lanzhou Hydrology Station and Tongguan Hydrology Station,and take the variables with the maximum correlation coefficient as follows:rhu,NINOSSTD3.4-（1+2）,Sun-earth distance.Tangnaihai Hydrology Station:rhu,NINOSSTD3.4-（1+2）,distance between the Sun and the Earth.Tongguan:rhu,NINO1+2,distance between Sun and Earth.（4）For the same hydrological station and the same model,the forecasting effect of the third group of independent variable combination（sun-earth distance,climate index,meteorological factor and historical runoff data）is better than that of the second group of independent variable combination（climate index,meteorological factor and historical runoff data）.According to the correlation analysis between the solar Earth distance wavelet decomposition and the climate index wavelet decomposition,the correlation coefficient of the solar Earth distance d3 wave is the highest,therefore,the d3 wave is used as a link to extend the climate index.In the case of different model prediction effect,the the d3 wave is used as a link to extend the climate index has different degree of improvement on the runoff prediction effect of each hydrologic station studied.In the BP,GRU and LSTM prediction models,the prediction effect of the d3 wave is used as a link to extend the climate index on Tangnaihai Hydrological Station increased by 13.4%,8.6%and 10.6%respectively,and the prediction effect on Lanzhou Hydrological station increased by 7.85%,16.2%and 16.3%respectively.The forecast effect of Tongguan hydrology station is increased by 12.5%,6.95%and 7.45%respectively.The d3 wave is used as a link to extend the climate index are essentially the sun driving the seasonal changes of Earth’s day and night,which in turn have an impact on water resources.