摘要
以逐步回归、均生函数和多层递阶作为月平均降水量的3种子预报方法,进一步采用二次规划计算方法,通过计算各子方法的最优非负权系数进行最优组合预测建模研究。实例计算结果表明,这种组合预测模型的预报精度优于各子方法,并具有较好的理论依据,可在实际业务预报中进行推广应用。
An optimal combined forecasting model for monthly mean rainfall is developed by using the step regression model and the mean generating function model and the multi-level recursive model. In the meantime, the quadratic programming is used to calculate the best non-negative weights. Simulation results show that the combined forecasting model based on theory is superior in the precision of prediction as compared to the submodels and worthy of application in the operational forecasting.
出处
《热带气象学报》
CSCD
北大核心
2004年第6期704-712,共9页
Journal of Tropical Meteorology
基金
国家自然科学基金资助项目(40075021)
关键词
月平均降水量
非负权重
二次规划
组合预测
monthly mean rainfall
non-negative weights
quadratic program
combination forecasting
