Forecasting of Flood Stage Using Neural Networks in the Nakdong River, South Korea

This paper describes the use of Artificial Neural Networks (ANNs) analysis as a computing device for forecasting flood stage in the Nakdong River basin, South Korea. The FR-CGBP, BFGS-QNBP, and LMBP algorithms were used in ANNs training here. The connection weights and biases, parameters that were obtained from training procedures, were used for validation of ANNs. The training pattern that involved LMBP algorithm was found to yield slightly better results by statistical analysis of forecast errors. The validation of ANNs was carried out so as to evaluate model fitness with untrained validation data set. Results indicated that ANNs approach produced forecasts of comparable accuracy to observed flood stage.