Study On The Regularity Of Drainage Water Volume Of Qingtongxia Irrigation Area Of NingXia

QingTongXia Irrigation Area, one of the most historic and largest irrigation Area in China, has been draining 40-50% water irrigated. It is theoretically and practically important to the water dispatching and sustainable development of the Yellow River that forecast the drainage water volume properly. Based on reviewing numerous precious national and international research of related to the water resource, the paper focus on the transformation of four types of water groundwater environments drainage water volume change regularity and the drainage water influence factor to study the drainage water volume model by adopting field investigation technical way of experiment and theory analysis. The main conclusions are as follows:1. The depth groundwater table is an important influential factor to the drainage water volume. The depth groundwater table, affected by irrigation greatly, changes seasonally and periodic fluctuation between tiptop and lowest groundwater table regularly. After analysing the influential factor to the drainage water volume with the help of Grey Incidence Analysis, the results show that the ranking of the factors according to the degree of effectiveness in descending order is as follows: drainage water volume, irrigation water volume, evapotranspiration and precipitation. Finally the neural network model of groundwater table forecast in the east Irrigation Area of Qingtongxia is presented in this paper, which is agreed well with measured data.2. Drainage water volume is closely associated with irrigation water, the depth groundwater table, evapotranspiration and precipitation. By analyzing the series data of drainage water volume and the influential factor in the in the east Irrigation Area of Qingtongxia, drainage water volume is markedly positive related to irrigation water and precipitation, and this connection can be described as a linear line. Drainage water volume connected negatively to groundwater table and evapotranspiration. The relationship between drainage water volume and groundwater table agreed with by power function, drainage water volume and evapotranspiration, a linear function. Further study on the influential factor to the drainage water volume shows that the sequence is irrigation water volume > groundwater table > precipitation > evapotranspiration by introducing the method of grey correlative degree analysis.3. Stepwise regression and error back propagation of artificial neural net work (ANN-BP) theory are first introduced to establish the month drainage water volume forecast model in the east irrigated area of Qingtongxia irrigation area. Through stepwise regression study, the forecast model precision of using three variables (irrigation area water diversion, the depth groundwater table and precipitation) is apparently more accurate than the model of two variables (water diversion and depth groundwater table), and a little more accurate than the model of four variables (irrigation area water diversion, groundwater depth, precipitation and evaporation). In order to overcome the defect of slow astringency and local minimum of steepest decent method in ANN-BP during setting up the ANN-BP model, a method that uses GN-BFGS algorithm to train the neural net work is presented in the irrigated period (form April to November), and a monthly model of prediction of drainage water volume in the east irrigated area of Qingtongxia irrigation area by ANN-BP with GN-BFGS algorithm are presented in this paper. Results of prediction indicate that the model not only can improve predicting ability and the accuracy efficiently, but also are good at robustness and extension capacity. Lastly, the paper analyzed the two models and validated the models by real observation values. Compared with Stepwise regression model,the forecast precision of ANN-BP model is the higher, and it can be used to forecast the drainage water volume in real-time water regulation.4. To overcome the shortage of the original data in the non-irrigated period, a monthly forecast GM forecast model and the cumulative residu...