DERF产品在贵州降水主分量逐步回归预报模型中的释用  

Application of DERF Products to Forecast Precipitation Based on PCA Regression Forecast Model in Guizhou

作　　者：白慧[1,2] 吴战平[1] 李忠燕[1] 周涛[1] 

机构地区：[1]贵州省气候中心,贵州贵阳550002 [2]贵州省山地气候与资源重点实验室,贵州贵阳550002

出　　处：《干旱气象》2015年第2期344-348,共5页Journal of Arid Meteorology

基　　金：贵州省科学技术基金(黔科合J字[2010]2058号);贵州省气象局青年基金(黔气科合QN[2014]05号)共同资助

摘　　要：从短期气候预测业务面临的实际问题出发,针对月尺度气候预测,通过多元逐步回归统计方法对1981~2010年贵州省降水主分量与NCEP/NCAR再分析资料的500 h Pa高度场建立预报模型,进而将该预报模型分别应用到2007~2013年5月16日、21日、26日起报的DERF产品500 h Pa高度场中,以检验不同起报时间DERF产品的预报能力。结果表明：贵州83站6月降水量EOF前5个模态的方差累积贡献率均达到了65%以上,PC1与同期6月降水量标准化序列的相关系数达0.955。贵州6月降水量与副热带高压位置以及中高纬度大气环流密切相关。释用DERF产品发现,该产品对贵州6月降水主分量具有一定的可预报性,其中5月26日起报的预报效果较好,其与同期NCEP/NCAR拟合值和实况值的相关系数分别为0.76和0.53。对2007~2013年5月26日起报6月的DERF产品释用预报结果进行Acc和Ps评分检验,发现除2008年预报评分较低外,其余6 a预报评分较好,Acc平均为0.45,Ps平均为82.0,尤其是2007年和2009年DERF产品释用的预报Ps评分分别达83.9和83.5,表明DERF产品在贵州降水主分量逐步回归预报模型中的释用具有较好的预报能力。Based on the practical problem, and aiming at the monthly - scale climate prediction, the forecast model was established between the PC1 of monthly precipitation at 83 meteorological stations in Guizhou in June during 1981 - 2010 and geopotential height fields on 500 hPa from NCEP/NCAR by using stepwise regression analysis method, firstly. And then the 500 hPa geopotential height fields of monthly dynamic extended range forecast（DERF） product on 16, 21 and 26 May during 2007 -2013 from National Climate Center in China were applied to predict precipitation by using the forecast model, and compared with the observed precipitation, the prediction capabilities of DERF for precipitation in June in Guizhou were tested. The results showed that the cumulative contribution rates of variance for the first five modes of EOF were more than 65% , and the correlation coefficient between the PC1 and normalized precipitation from observation stations was O. 955. The precipitation in June in Guizbou was closely related with the position of subtropi- cal high and the atmospheric circulation in mid - high latitudes. Compared with the observed precipitation and fitted precipitation of NCEP/NCAR, the DERF products at three different time had a certain predictability to the PC1 of precipitation in June, which was better on May 26, and the correlation coefficient with fitted precipitation of NCEP/NCAR and observed value respectively reached 0.76 and 0.53. The Acc and Ps test to the application of DERF products on 26 May during 2007 - 2013 showed that the average values of Ace and Ps were respectively 0. 45 and 82.0 except for 2008, especially the Ps value in 2007 and 2009 reached 83.9 and 83.5, and the 500 hPa height fields of DERF applied to PCA regression forecast model had better prediction capability to precipitation in Guizhou.

关 键 词：DERF产品 6月降水预报模型 释用 贵州 

分 类 号：P456.3[天文地球—大气科学及气象学]