多层递阶回归分析法在全国初夏雨量预测中的应用

Application of Multi-level Recursive Regression Analysis on Forecasting Early Summer Rainfall over China

在传统统计方法的基础上,引入多层递阶方法。将多层递阶方法与回归方法合二为一,作为新的预测方法尝试,以弥补常规统计方法中回归系数是常数的不足。将多层递阶回归方法应用于短期气候预测,建立多层递阶回归模型,并用全国160个气象站5月份雨量作预报试验。经过近2a的实际预报表明:该模型对华中、华东、两广大部、云贵高原大部的5月份雨量有较好的预报能力,具有一定的参考价值。

Based on traditional statistical method, the multi - level recursive regression method was used to make up the inadequate problem that regression coefficient is a constant in conventional statistical method. Using the multi -level recursive regression model, the forecasting precipitation amounts in May for 160 weather stations over China are investigated. The results show that the model can reasonably simulate the precipitation amounts over Central China, East China, most of Guangxi and Guangdong provinces, a majority of Yunnan and Guizhou provinces in the early stage of summer.