Combined Forecasting Method Based on Water Resources Risk Management

In water resources risk management, since water dispatching decision at the later stage lacks sufficient support, because of low accuracy, in forecasting, a combined forecasting method is put forward. The combined forecasting method constructs an angle cosine model through measurement vector, prediction vectorand weight vector; upon determining variant quotient, water resources are rated “sufficient” or “insufficient” taking into account historical water resources and collected statistics, and calibration and initialization of parameters are determined by fitting goodness and dynamic approximation, switching the forecast from a solo to dual model, meanwhile comparing implications of prediction accuracies which assure the forecast supports the coordination. As the experiment demonstrates, this methodology can effectively control risks and shortcomings of a solo model of water resources risk management. Through merging various forecasting models, forecasting accuracy is improved by 20%, lowering the risk of water dispatching decisions while providing referable statistics applied in the water resources risk management. © Springer Nature Singapore Pte Ltd. 2019.