An Application of ANN on Groundwater Level Prediction of the Fractured Aquifers in the Nhue—Day River Basin

Groundwater level (GWL) varies periodically or non-periodically with various factors including rainfall (RF), evaporation (ET), number of sunny hours (NS), humidity (HM), air temperature (AT), and topographic elevation (TE). This study presents an implementation of an Artificial Neural Network (ANN) to predict groundwater level in the fractured aquifers of Nhue—Day River basin, Vietnam. In this regard, the monthly historical time series climatological data (rainfall, temperature, humidity, and evaporation) during 2018–2019 and hydrogeological parameters at seven observation wells have been used as input variables to estimate GWL. The statistical performance of the developed Levenberg-Marquardt back-propagation artificial neural network (ANN) models was evaluated using three criteria: Mean Square Error (MSE), Mean Absolute Error (MAE), and Coefficient of Determination (R2). Results showed that the ANN model predicted groundwater level with reasonable errors. The average root mean square prediction error was 1.77 m for groundwater level prediction at seven observation wells. The training results revealed that monthly precipitation and evaporation are important variables that have a strong influence on groundwater level prediction. Based on these values, the seasonal groundwater level and fluctuation were mapped using the geostatistical toolbox in ArcGIS 10.8. Finally, the present study as a pioneer approach provides significant contributions to groundwater management and development of Nhue—Day River basin.