不同引水量条件下位山灌区地下水位动态研究

Study on Groundwater Table Dynamics of Weishan Irrigation District under Different Yellow River Water Diversion Conditions

为定量评价引黄水量的变化对位山灌区浅层地下水位动态的影响 ,利用地下水位观测数据和气象资料 ,建立了输入延迟神经网络模型 (IDNN)和递归神经网络模型 (RNN) ,并对模型的表现进行了比较 ,结果表明 ,RNN模型对于地下水位动态的模拟更为有效。利用RNN模型 ,在引黄水量减少条件下 ,对不同气候条件下灌区地下水位动态情况进行了定量分析 ,并提出了位山灌区适宜引水量为 10～ 14亿m3 /a。

In order to quantify the impacts of changes in Yellow River water diversion on shallow groundwater table of Weishan Irrigation District, two types of functionally different artificial neural network (ANN) models, Input Delay Neural Network (IDNN) and Recurrent Neural Network (RNN), are established using a thirteen_year_length series of groundwater table records, related meteorological data, and the performances of the two models are compared. Simulation results suggest that the RNN is more efficient for groundwater table dynamics simulation in a period of 2 years prediction than the other artificial neural network model. Furthermore, the RNN model is used to quantify changes in groundwater level in the district under conditions of decreased Yellow River water diversion and different weather conditions. The results suggest that the appropriate water diversion from Yellow River in Weishan Irrigation District be in the range of 1～1.4 billion m\+3 annually.