PREDICTION AND FORECASTING OF GROUNDWATER LEVEL FLUCTUATION BY ANN TECHNIQUE

Groundwater is one of the most important natural resource in order to full fill the water requirements from irrigation, domestic, industrial and research needs. The present study is carries out to find the optimum model of ANN technique through various pre-processing technique in the process of groundwater level prediction in Lower Bhavani River Basin (LBRB). Hybrid Con-ANN (Conventional Artificial Neural Network) model with WT (Wavelet Transform), CWT (Continuous Wavelet) and CWFT (Continuous Wavelet Fast Fourier Transform) preprocessing techniques are performed. Model development is carried out through calibration, validation process by three input variable and one output variable for the period of 2009-2015 on monthly basis. Based on the prediction performance, CWTFT-ANN is found to be the best model by comparing with the other models through statistical indices RMSE, R-Squared, COC, MSE and MAS measurement. Further the optimum model is used to forecast the groundwater fluctuation to next one time stress period