Reservoir Water Quality Modeling for COD using Artificial Neural Network

Major sources of water are surface and subsurface; surface water includes River, Reservoir, Creek, Streams, etc. In this study, Jayakwadi Reservoir, INDIA considered as a study area with monthly observed data from 2001–2012. Study area provides a multiple services such as drinking, energy production, irrigation, industrial benefits and others. Two different ANN networks, that is, the Feedforward Neural Network (FFNN) and Cascade Correlation Feedforward (CCFF) networks were developed to estimate COD using various combinations of monthly input parameters; those covered maximum water quality, simple field and continuously measured such as BOD, Temperature, pH, TDS, and DO. RMSE, MAE, FA1.1, IA and R² statistics were used for the performance criteria. Comparison of the results indicated that the CCFF model, with four inputs BOD, Temp., DO and TDS has performed slightly better than the FFNN in estimating COD. However, it can be conclude that these techniques provide similar accuracy in estimating COD concentration.