Potential assessment of the support vector regression technique in rainfall forecasting

Forecasting and monitoring of rainfall values are increasingly important for decreasing economic loss caused by flash floods. Based on statistical learning theory, support vector regression (SVR) has been used to deal with forecasting problems. Performing structural risk minimization rather than minimizing the training errors, SVR algorithms have better generalization ability than the conventional artificial neural networks. The objective of this investigation is to examine the feasibility and applicability of SVR in forecasting volumes of rainfall during typhoon seasons. In addition, Simulated Annealing (SA) algorithms are employed to choose parameters of the SVR model. Subsequently, rainfall values during typhoon periods in Taiwan's Wu–Tu watershed are used to demonstrate the forecasting performance of the proposed model. The simulation results show that the proposed SVRSA model is a promising alternative in forecasting amounts of rainfall during typhoon seasons. Copyright Springer Science+Business Media B.V. 2007