Prediction of rainfall - runoff relation with artificial neural networks and autoregressive moving average models

In terms of efficient use of water resources and prevention of disasters, rainfall and runoff relation is very important. Complete and consistent data increases the efficiency of planning and design processes and ensures more reliable results. In rainfall - runoff modeling, traditional methods for estimating data may be insufficient. Therefore, many alternative methods have been developed for the prediction of missing or future runoff. In this study, the data obtained from USA Waltham Massachusetts Stony Brook Reservoir was used. 731 days of rainfall, runoff and temperature data were used to generate input data in the Feed Forward Backpropagation Neural Network (FFBNN), Multiple Linear Regression (MLR) and Autoregressive Moving Average (ARMA) models. The results obtained were compared with the actual results. As a result, it was found that the ANN method is more than 92% compatibility in estimating these current data.