【摘要】 本文对1972—1978年七年春播期(3月27日—5月10日)500mb亚欧环流形势,应用变K-均值聚类法进行客观分型,共得九型。对它们和春播期四种主要中期天气过程——连阴雨和低温连阴雨、过程性天气、连晴,用对应分析方法数量化。以环流形势与同期的北半球500mb波谱分析中选出的物理量作为因子,对天气过程类型作出了逐步回归和AID聚类的统计推断。 对1979、80年春播期作了预报检验,效果尚好。更多还原

【Abstract】 In this paper, an objective classification of Asia-Europe circulation pattern at 500 mb during the spring sowing periods (Mar. 27 to May 10) in 1972-1978 has been done by K-means clustering algorithm, and nine circulation patterns have been obtained. In addition, there are four major medium range weather patterns in the spring: the persistent rainy weather, the persistent rainy weather with lower temperature, the weather with alternative processes of rainy and sunny days and the persistent sunny days. The corresponding analysis has been made for these weather patterns. The weather patterns are considered as dependent variables, while the circulation patterns and the selected factors in spectral analysis of the angular momentum transfer and kinetic energy variability at 500 mb are considered as independent variables. The statistical inference is made by using the stepwise regression analysis and automatic interactive detection (AID) clustering algorithm.The method was used in operational forecasting during the spring sowing period in 1979 and 1980, and the accuracy was rather satisfactory.更多还原