Infiltration Loss Affects Toward Groundwater Fluctuation Through CANFIS in Arid Watershed: A Case Study

Abstract

The focal objective of this study is to predict variation of water table depth in an arid watershed. Co-adaptive neuro-fuzzy inference systems (CANFIS) and recurrent neural network (RNN) are employed to validate model performance of proposed watershed. Two scenarios are considered to measure the efficiency of model and found that Scenario 2 shows prominent performance than Scenario 1 for both methods. For Scenario 2, the value of coefficient of determination (R2) is 0.9287 and 0.9537 for RNN and CANFIS in testing phase, respectively, while Scenario 1 gives 0.9011 and 0.9365 for RNN and CANFIS, respectively. It is found that premeditated losses because of infiltration loss are reasonably not much as at time of high precipitation in Sambalpur region. Results of R2 recommend that insertion of infiltration loss in this situation develops the model efficacy to predict depth of water table.