A new logic-based model to predict nitrates in groundwater using Artificial Neural Network (ANN)

In this study, a soft computing technique using Artificial Neural Network (ANN) has been used to model the nitrate concentration which is an important indicator of the groundwater quality. Using an algorithm construct in the MATLAB modelling environment, a final ANN model was developed by methodical screening of the different ANN architectures produced for the cases under study. Further, the model was validated using the data on the groundwater quality of Ayapakkam, in Thiruvallur near Chennai City, Tamilnadu, India for the prediction of nitrate concentration. The correlation between the observed and predicted values of the nitrate concentrations were found to be very high (r = 0.998). The results of the model exhibit an accuracy of 99% (RMSE = 0.053 and Index of agreement = 0.992) in predicting the final nitrate concentration which underscores the efficiency of the model for the current study.