The Comparisons Of Forecasting Methods For Mid- To Long-Term Stream In Nenjiang River Basin

China is located in the East Asian monsoon region, where the climatical environment changes drasticly. The unique monsoon circulation in this area (or watershed) determines the complexity and variability of the hydrological cycle, which shown as the inhomogenous spatiotemporal distributing of precipitation. Compared with the other countries or regions, as a non-engineering measure which can resolve the conflict between water incoming and using, the mid- to long-term stream discharge forecasting can resolve the conflict between flood controlling and drought relief, water storing and discarding. Through overall planning, the mid- to long-term stream discharge forecasting really can ease the contradiction between water supply and demand, provide services for the flood recycling and the unified management of water resources, as well as achieve the economical and reasonable running of the hydropower engineering. Therefore, the study of the mid- to long-term stream discharge forecasting in our country has an important practical meaning, especially with the outstanding contradiction between water supply and demand, and it has paied more and more attention by the water resources administrative department in recent years.Selecting a reasonable prediction method is the premise of making accurate forecast. In this paper, with the long series hydrological and meteorological datas in the Nenjiang River Basin, we conduct the research by citing the sensitivy analysis results which the runoff in the Nenjiang River basin to the main influences. Firstly, we compare simulative accuracy and the forcasting inspective accuracy of time series analysis and regression analysis, therefore to evaluate the mathematical statistics method and the physical cause analysis method. Then, we compare the simulative accuracy and the forcasting inspective accuracy of multiple regression analysis and artificial neural network method, what's more, we also compare the the strengths and weaknesses of linear and nonlinear analysis methods in the correlative identification between dependent and independent variables based on the physical cause analysis. Through these comparative studies, the artificial neural network method based on physical causes is preliminarily determined as the best.analysis methodA reasonable physical cause analysis is a the base to improve the forecast accuracy. In this paper, artificial neural networks in selected to identify the relationship between predictive factors and project, then, after the further analysis of the influence degree of runoff in different seasons and with different forecasting factors, we gradually identify the key predictors, thus to build a middle and long term runoff predictive model with more rational and high accuracy.In addition, this paper primarily analyzes the main factors, mechanism and evolution rules of abnormal large droughts and floods, moreover, with the commensurable method, it also identifies the performed period and cataclysm nodal years of the abnormal large droughts and floods in the Nenjiang River Basin. Abnormal runoff forecasting such as droughts and floods belongs to hydrological and climatical forecasting, which mainly studies how to use month, quarter or year total precipitation in the basin, and the relationship between physical factors and evolution rule of temporal sequences of the hydrological characteristics to estimate drought and flood trends and the runoff dry and wet levels in the future. The connotation of hydrological climate is different from the middle and long term runoff forecasting which studies mounthly and seasonally changes of runoff, and it pays more attention to the abnormal runoff, namely large droughts and floods, the two extremes.By selecting forecasting methods, the furtherly selecting the predictors and continuous improving the prediction model, finally, a set of middle and long term forecasting model for the Nenjiang River Basin and a commensurable figure which can identify the cataclysm nodal years ot the large droughts and floods.