Rainfall-Runoff Simulation Using Artificial Neural Network Considering Temperature

Determination of runoff is a complex Hydrological process. Various models are useful to simulate rainfall-runoff process, but artificial neural network is better alternative to other models due to its black box nature. Artificial neural network (ANN) working as a universal predictor, it works as similar function like human brain. This study uses Feedforward back propagation artificial neural network (BPANN) in rainfall-runoff simulation for Panshet Dam catchment Pune India. Various climatic parameter effect on rainfall-runoff process, change in temperature largely affected on this process from last three decade. Accuracy of ANN models mostly depend on selection of different input parameters. In this study, ANN model used to predict runoff of one day lead time considering the inputs as runoff, runoff and rainfall and runoff, rainfall & temperature. It is found that out of three different ANN models, model considering temperature as additional input parameter perform better as compared to other models. Accuracy of ANN model considering temperature increases by 6% to 9% over other two models.