长期天气预报中有关预报因子的若干探讨

目前行之有效的长期天气预报方法都是数理统计方法。本文以江淮流域夏季旱涝的长期预报为例，对有关预报因子的几个问题作了探索。挑选有实际天气学意义、各自独立、单相关系数较高的物理因子作预报因子，更有利于提高预报准确率，预报因子的时间尺度以月、季平均值最为适宜，而以三个旬滑动平均值的效果最佳，建立各种预报量的初选、精选因子库，更适用于计算机技术制作长期预报。

SOME INVESTIGATIONS ON PREDICTORS INLONG-RANGE WEATHER FORECASTING

Up to now,the most effective long-range weather forecasting methodis mainly mathematic statistical method. In this paper,taking the long-rangeforecasting for summer flood-drought in the Yangtze-Huaihe Rivers as anexample,several predictor problems are studied. It is found as follows:The meaningful predictors and these with high single correlation as well asmatually ndependent ones are selected.As a result,the monthly or seasonally mean predictors are most suitable;among them,the three-ten-deysrunning mean value would be the best.The establishment of a preliminaryand superior factor banks for every predictant is extremelly useful for production of long-range forecasting by use of computer technique.