Estimation of Aquifer Potential Using BPNN, RBFN, RNN, and ANFIS

Abstract

In this paper, aquifer potential in terms of sensitivity of a well in arid region is measured. Four techniques such as Backpropagation neural network (BPNN), Radial basis function network (RBFN), Recurrent neural network (RNN), and Adaptive Neuro-Fuzzy Inference System (ANFIS) are used to predict the aquifer potential of the well. Measured specific drawdown of the well is considered as input and evaluated aquifer loss coefficient as output for developing the efficiency of model. The complexity of aquifer characteristics of the region is measured through the sensitivity of the developed models. Results of all techniques explain the variation of aquifer characteristics in arid region. Among all proposed models, ANFIS executes best for mapping the sensitivity of aquifer. Overall results show the integrity of performances to understand the complex behavior of aquifer in decreasing order of ANFIS, RNN, BPNN, and RBFN.