Annual Runoff Analysis And Forecast Of Watershed Above Douhe Reservoir

Annual runoff forecast is an important part of hydrological prediction.It is of great significance for reservoir flood dispatching and utilizable regulation.Due to the annual runoff forecast predicted period is longer,and subject to the conditions of measurement data and the forecast method,and the influence of many uncertain factors,the accuracy of annual runoff is still cannot meet the needs of the production department.Therefore,it is a hot spot for hydrology forecasting to establish a model of annual runoff forecast.Based on the natural annual runoff DouHe Reservoir as the research object,the DouHe Reservoir are calculated by water balance method is used to restore river basin from 1953 to 2013 a total of 61 annual runoff.This paper considers the effect of human activities on the consistency of annual runoff time series.Then using the correlation test in SPSS software to select the corresponding prediction factors.According to the applicability of the forecast method,the appropriate time series length is chosen,and the model of the annual runoff forecast is established.The main content is as follows:(1)This paper investigates the evaporation,seepage,the water of industrial life,the irrigation water of agriculture,the drainage of water in the basin and the annual storage of the reservoir.In this paper,the annual runoff of the DouHe Reservoir was calculated,and the calculation results were analyzed.(2)This paper respectively using rainfall anomaly analysis,sliding correlation coefficient method,the double mass curve method and the method of rainfall runoff correlation diagram to analyze the consistency of the annual runoff series of the DouHe Reservoir.Then,this paper use the SPSS software to select the forecast factor for annual runoff.(3)The regression model was established based on the annual runoff time sequence of the DouHe Reservoir in 1976 ~ 1999.(4)Based on the sequence of annual runoff in the DouHe Reservoir basin in 1953 ~ 2013,the paper builds the self-regression and moving average(ARMA)model.(5)The BP neural network model was established based on the annual runoff time sequence of the DouHe Reservoir in 1976 ~ 1999.This study to establish the model of calculation is convenient,simple application,taking into account the time sequence consistency,according to different data sequence with different prediction models,and comparison,we will further improve the after runoff forecast can be widely used in the region.