The study on runoff series prediction in Middle Yellow River

Taking into account the severe impact of human activities, complex utilization requires and sensitive and fragile eco-environment, the Middle Yellow River is selected for runoff prediction object. Based on analysis of runoff characteristics and influencing factors, Threshold Auto-Regressive Model, improved Weng Model and Mixture Regression Model are established for the runoff series prediction of control station on the middle Yellow River, Xiaolangdi station. Through the evaluation of the prediction effect of models, Mixture Regression model´s superiority and rationality are proved. Finally, research trends and implication prospects of annual runoff prediction are reviewed.