方差分析周期外推法在春季降水量预报中的应用

选取西安市1990-2009年共20年春季（3～5月）降水量资料，采用方差分析法，按不同长度周期进行排列，求出F值并进行检验，得出春季总降水量变化：以6年为主周期，8年、7年为校正周期，将所得的3个周期的稳定位相值叠加，再用外推法对西安2010年、2011年春季降水量进行预报．结果表明：应用方差分析制作的春季降水量趋势预报方法，历史拟合率为95％，其中1996年、2002年和2003年定量预报偏差较大，为20～30mm；在2010年、2011年的试报中，定性预报效果较好，2010年定量预报偏差较大，为37．3mm；用此方法作预报时若配合其他预报方法使用，效果会更好．该方法对实际业务预报春季降水量具有一定的参考价值．

Application of Variance Analysis and Extrapolation Method to Forecasts of Spring Precipitation

Talcing as a sample the precipitation data of Xi’an in 20 a springs(from march to may) in the period of 1990~2009 and arranqing them in order according to the different length cycles,we finally got the value of F by means of the variance analysis.It was found that in the change of total precipitatoen in springs,6 a was the dominant cycle and 8 a and 7 a were calibrated cycles.Superposing the stable phase values obtained in three cycles and the extrapolation method,we could effectively make forecasts of the spring precipitations of Xi’an in 2010 and 2011.Results showed that by applying the variance analysis to precipitation forecasts,the historical fitting rate was 95 percent.In the 1996,2002 and 2003 years the quantitative forecast deflections were larger from 20 to 30 mm.In 2010 and 2011 years,the qualitative forecast effect was better,but the 2010 quantitative prediction deviation was 37.3 mm.This method could achieve better results when combined with other forecast method.