澳门单站气象资料变化趋势的预报模型之比较

Comparison among Forecast Models of Variation Tendency in Meteorological Data of Macao Single Station

以澳门单站年和四季降水和温度两个气象要素的气象资料（1952～2001年）为主要研究对象，利用3种统计预报模型（谐波分析外推模型、开环门限自回归模型和主分量因子的逐步回归模型）进行拟合及试报，对澳门的降水及温度的变化做较深入的分析及预测，比较3种统计预报模型在研究澳门单站气象资料的变化趋势中的优劣。结果发现，3种预报模型在预报年和四季平均气温的方面都比较理想，但对降水量变化序列做拟合及试报则差异较大。总体而言，气候场的主分量逐步回归法对相同的气候资料建模做拟合及试报，结果拟合率是三者中最高的，复相关为0．821-0．911，预报效果除秋季较差外，其余都非常好。

Historical data from 1952 to 2001 of two meteorological elements, rainfall and temperature, in time series of year and four seasons of Macao single station are the prime objects of this study, being taken into three statistical forecast models, that is extrapolation of harmonic analysis, open threshold auto-regression and stepwise regression of principal component factors, to execute simulation and put into the forecast trials. Besides having more thorough analysis of the variations of rainfall and temperature in Macao and thus making prediction, the pros and cons of the three models are compared in the study of the meteorological data variation tendency of Macao single station. The results reveal that the performance of predicting the annual and seasonal average temperatures of the three fore cast models are rather satisfactory but have wider gaps in simulating and forecasting the time series of rainfall. In general, with the same data to deal with, the simulation percentage of stepwise regression of principle components of climatic fields is the highest among all the three, with correlation between 0. 821 0. 911, and it performs well except in autumn.