Application of Support Vector Regression for Mid- and Long-term Runoff Forecasting in “Yellow River Headwater” Region

Abstract

Accurate mid- and long-term runoff forecasting is of great significance for rational development and utilization of water resources in “Yellow River Headwater” region, where runoff in the headwater region contributes to nearly 35% of the total amount of the Yellow River basin. In this paper, the monthly runoff data of Tangnaihai station in “Yellow River Headwater” region are analyzed as case studies. This paper presents support vector regression model for mid- and long-term runoff forecasting, and analyzes the influence of support vector regression model's parameters on the runoff forecasting accuracy, and finally compared with Auto Regressive model (AR) and Radial basis function neural network (RBFNN). The results indicate that SVR showed the best performance and is proved to be competitive with the AR and RBFNN models in both stations. SVR methods provide a promising reliable methods of mid- and long-term runoff forecasting.