As groundwater is a critical source of water for both drinking and agriculture in Jilin Province, China, it is important to investigate and understand groundwater level dynamics in this area. Time-series analysis and artificial neural networks (ANN) are commonly used for analysing and forecasting groundwater levels. The integrated time-series (ITS) and auto-regressive integrated moving average (ARIMA) are the most commonly used models for time-series analysis. Among ANN approaches, the radial basis function neural network (RBFNN) is a widely used model for making empirical forecasts of hydrological variables. There are no previous reports comparing the ITS, ARIMA and RBFNN models together in groundwater-level dynamics literature. An attempt has been made in this study to investigate the applicability of these three models for the prediction of groundwater levels based on root mean squared error, the Nash-Sutcliffe coefficient and mean absolute error. The results indicated that all three models reproduced groundwater levels accurately. In addition, the RBFNN model was more reliable than ITS and ARIMA. This provides a choice in the selection of models for analysis and prediction of groundwater levels. The predicted results also provide a basis for rational exploitation and sustainable utilization of groundwater resources.

You do not currently have access to this content.