Runoff Series Spatial And Temporal Variation And Forecast Of Four Tributaries In Middle Reaches Of The Yellow River

Runoff is a physical quantity and would change over time. There will be a lot of runofftime sequence sample after a long time observation. Establishing a reasonable predictionmodel on the basis of temporal and spatial variation of the stream runoff series analysis andresearch is a basic work in the process of understanding and exploring the water resourcesdevelopment, and it is also the focus of foreign experts.Base on the construction of northern Shaanxi Yulin Energy and Chemical Base and localwater demand, this paper combines modern technology theory and traditional stream runoffanalysis methods and takes runoff data of the controlling hydrological stations of the Sihewatershed (Kuye river, Tuwei river, Wuding river and Gushan river) from1956to2009asstudying object. Mathematical statistics, Mann-Kendall rank test and wavelet analysis areused to analyze the temporal and spatial variation of the stream runoff series in Sihewatershed, and GIS spatial interpolation function is used to study its spatial distributionpattern in differt time. Then the amended combination model of river mid-long term runoffforecast is brought forward and applied to the yearly runoff forecast in controllinghydrological station of Sihe watershed. This paper is mainly focus on the following points:(1)Tributaries runoffs show a volatility reduction in the studying period. TheMann-Kendall rank test analysis gives out that runoffs abundance, flat and dry seasons changesignificantly order1979and1996for boundary; yearly distribution inhomogeneities graduallydecreases and then increases in the studying period and have a relatively gentle trend after the1980s.(2) Precipitation and human activities are the main factors which lead to the Sihewatershed runoffs changes. The analysis of the precipitation-runoff relationship before themutation shows that after70s the human activity is an apparent factor and plays a leading rolein the past20years.(3) Wavelet analysis can be drawn to the stream runoff series of the Sihe watershed andfind there are cycles of25,7and3years with obvious mutations.(4) GIS spatial interpolation analysis shows that annual spatial distribution runoff decreases from southeast to northwest and there are significant differences between east andwest. In the studying areas, the annual runoff distribution of each tributary is different:Wuding River desreases obviously from south to north, and from east to west shows the sametrend; Tuwei River has its high value in the middle and lower reaches, and the whole basinshows a decreasing trend from south to north; The downstream of Gushan River is relativelyhigh, the whole basin decreases from southeast to northwest; The middle reaches of KuyeRiver is small, much lower than other regions.(5) Amended combination model is brought out for forecast, this model is combined byGrey and neural network and amended with markov. The combination model by the markovamended is applied to the mid-long term stream runoff forecast. The result shows it is superiorto single models and other combined models without amended and has better simulation andforecasting precision. So the amended combined model is more effective for runoff forecast.