Artificial Neural Network Model for Prediction of Groundwater Levels:Case Study

There are many environmental concerns to the quantity of surface water and groundwater in the hydrological system. It is very important to estimate the groundwater levels by using readily available data for managing the water resources of the natural environment. As a case study in Sindappalli Uppodai Subbasin of Vaippar River basin in Tamilnadu an Artificial Neural Network (ANN) methodology was applied to estimate the groundwater levels as function of monthly precipitation, Evapotranspiration, lake water levels and roundwater level. Among the different robust tools available, the Back-Propagation (BPNN) Artificial Neural Network model is commonly used to empirically forecast hydrological variables. The simulations results indicated that BP is accurate in reproducing (fitting) and forecasting the groundwater levels time series based on the R2 are 0.99 and 0.88, respectively. The RMSE, MAE for BP model in the predicting stage are 0.085, 0.076, respectively. It is evident that the BPNN is able to predict the groundwater levels reasonable well.