Real time runoff forecasting by Artificial Neural Network

In this paper, Artificial Neural Network (ANN) is proposed as a tool to predict runoff for the Lighvan Chay basin which is located in the North-West part of Iran. A feed-forward artificial neural network is trained by using back-propagation algorithm. The training and testing data were collected during years 1979 to 2000. The results of ANN model are compared with Linear Regression (LR) model. Three criteria Mean Square Error (MSE) Correlation Coefficient R2 and Nash Sutcliff Coefficient NE are used. Results demonstrate that neural network is better than the linear classic model and non-linear ANN modeling is useful for snow stay basin when given the same data inputs.

کلیدواژه ها:

Runoff forecasting ، Artificial Neural Network ، Linear Regression