三种地下水位动态预测模型在吉林西部的应用与对比

准确而可靠地预测地下水埋深对生态环境保护和水资源规划管理具有重要意义。针对吉林西部浅层地下水位动态变化的复杂性和非线性,提出了基于小波分析与人工神经网络相结合的预测方法小波神经网络(WA-ANN)模型。将研究区2002年1月2009年12月当月降水量、蒸发量、人工开采量和前月平均地下水埋深4个参数作为输入,当月平均地下水埋深作为输出,建立浅层地下水埋深预测模型,并与BP神经网络(BP-ANN)模型和自回归移动平均(ARIMA)模型进行比较,对比分析了三者的建模过程及其模拟精度。结果显示:相比两种ANN模型,ARIMA模型建模过程更为简单,计算效率更高;但WA-ANN模型的拟合精度高于BP-ANN和ARIMA模型,预测效果更好。总体来看,WA-ANN模型在浅层地下水埋深预测中具有一定的应用推广价值。

Comparison of Three Dynamic Models for Groundwater in Western Jilin and the Application

Accurate and reliable groundwater depth forecasting model is important to ecological environment protection and water resource planning and management.To minimize the interference of the nonlinear and complicated kinetic changes in the shallow water depth forecasting in western Jilin,a model for prediction is established based on the combination of wavelet analysis and artificial neural network,the wavelet network (WA-ANN)model:The parameters inputted in the models are monthly precipitation,evaporation, labor exploitation, and pre-monthly groundwater depth recorded from January 2002 to December 2009;and output was monthly groundwater depth in the study area.A comparison is made to BP artificial neural network (BP-ANN)model and autoregressive integrated moving average (ARIMA)model.The results show that ARIMA model processes more simple,but WA-ANN model predicted more accurate than both the BP-ANN and ARIMA models.In conclusion, the wavelet neural network model is more applicable for monthly average shallow groundwater depth forecasting.