Multi-model Super-ensemble Forecasts for the Circulation in August 2010
-
摘要: 利用多模式超级集合预报法,以欧洲中期天气预报中心、日本气象厅、德国气象局、中国气象局和中国空军气象中心共5个决定性7 d预报产品为集合成员,对2010年8月500 hPa高度场和850 hPa温度场分别进行固定训练期和滑动训练期超级集合预报。采用均方根误差和相关系数对超级集合预报、单一模式预报和简单集合平均预报进行对比检验,同时对各预报结果的均方根误差空间分布进行对比分析。结果表明:超级集合预报在所有预报结果中最佳,且滑动集合预报对8月后期时段预报要略好于固定集合预报,两者预报效果均好于参与集合预报的各模式,也好于集合平均预报。但随着预报时效的延长,集合平均预报的优势也随之提升。从预报结果均方根误差的空间分布可知,多模式超级集合预报相比于单一模式预报效果提高的区域,500 hPa位势高度场主要位于印度半岛、印度洋、青藏高原及以西地区,而850 hPa温度场则主要位于蒙古、青藏高原、中国新疆及以西地区。
-
关键词:
- 超级集合预报;
- 2010年8月环流形势;
- 对比检验
Abstract: Based on 24—168-hour model forecast of ECMWF, JMA, German Bureau of Meteorology, CMA and China Air Force, the multi-model super-ensemble and the multi-model ensemble mean forecasts of 500 hPa geo-potential height and 850 hPa temperature from 8 August to 31 August in 2010 are conducted using fixed training period and running training period. The root mean square error (RMSE) and correlation coefficient are utilized to evaluate forecasts of the super-ensemble, the multi-model ensemble mean, and anyone of five models. Meanwhile, the distribution of RMSE about different forecasts is analyzed, respectively. The results show that ECMWF model performs the best in 500 hPa geo-potential height forecast and JMA model performs best in 850 hPa temperature forecast among five models from 24 hours to 168 hours, and the skill of super-ensemble forecast is the best of all. The super-ensemble forecast skill is improved not only in fixed training period but also in running training period, and the super-ensemble result of running training is slightly better than the result of fixed training at the end of the forecast period, and both of them are better than the result of any single model and the simple ensemble mean. But the result of simple ensemble mean has its advantage along with the forecast time. The forecast result of the whole August shows that the value of RMSE in super-ensemble forecast is the least not only in 500 hPa geo-potential height but also in 850 hPa temperature, which is the best of all forecast results, and the result of running training is better than the fixed training. But the value of the correlation coefficient is similar between the forecasts, and the skill of the simple ensemble mean is the highest of all, which means the simple ensemble mean has its advantage. The RMSE distribution of super-ensemble, multi-model ensemble mean, and any single models is quite different over various regions, the skill of super-ensemble is the best of all. The areas of skill improvement by the super-ensemble forecast in 500 hPa geo-potential height are mainly located in Indian Peninsula, Indian Ocean, the Tibet Plateau and its west. The areas of skill improvement by the super-ensemble forecast in 850 hPa temperature are mainly in Mongolia, the Tibet Plateau and Xinjiang of China and its west. -
图 5 预报期内单一模式与集合平均、固定和滑动训练期超级集合预报的均方根误差和相关系数
(a)500 hPa位势高度场平均均方根误差,(b)500 hPa位势高度场相关系数,(c)850 hPa温度场平均均方根误差,(d)850 hPa温度场相关系数
Fig. 5 The mean RMSE and correlation coefficients of the best single model, the multi-model ensemble mean, unchanged superensemble, the changed superensemble during the forecasting
(a) the mean RMSE of 500 hPa geopotential height, (b) correlation coefficients of 500 hPa geopotential height, (c) the mean RMSE of 850 hPa temperature, (d) correlation cofficients of 850 hPa temperature
图 6 预报期500 hPa位势高度场和850 hPa温度场24 h预报的均方根误差空间分布
(a) 欧洲模式500 hPa位势高度场,(b) 滑动训练期500 hPa位势高度场,(c) 日本模式850 hPa温度场,(d) 滑动训练期850 hPa温度场
Fig. 6 Distribution of RMSE of 24-hour forecast at 500 hPa geopotential height and 850 hPa temperature
(a)500 hPa geopotential height of ECMWF model, (b)500 hPa geopotential height of the changed superensemble, (c)850 hPa tempertature of Japan model, (d)850 hPa tempertature of the changed superensemble
-
[1] Lorenz E N.A study of the predictability of 28-variable atmosphere model.Tellus, 1965, 17:321-333. [2] Epstein E S.Stochastic dynamic prediction.Tellus, 1969, 21:739-759. [3] Leith C E.Theoretical skill of Monte Carlo forecasts.Mon Wea Rev, 1974, 102:409-418. doi: 10.1175/1520-0493(1974)102<0409:TSOMCF>2.0.CO;2 [4] Toth Z, Kalney E.Ensemble forecasting at NMC:the generation of perturbations.Bull Amer Meteor Soc, 1993, 74(12):2317-2330. doi: 10.1175/1520-0477(1993)074<2317:EFANTG>2.0.CO;2 [5] Toth Z, Kalney E.Ensemble forecasting at NCEP and the breed method.Mon Wea Rev, 1997, 125(12):3297-3319. doi: 10.1175/1520-0493(1997)125<3297:EFANAT>2.0.CO;2 [6] Hamill T M, Snyder M.A comparison of probabilistic forecasts from Bred, singular-vector, and perturbed observation ensembles.Mon Wea Rev, 2000, 128(6):1835-1851. doi: 10.1175/1520-0493(2000)128<1835:ACOPFF>2.0.CO;2 [7] 刘金达.集合预报开创了业务数值预报的新纪元.气象, 2000, 26(6):21-25. doi: 10.7519/j.issn.1000-0526.2000.06.005 [8] Toth Z.Ensemble forecasting in WRF.Bull Amer Meteor Soc, 2001, 82(4):695-697. doi: 10.1175/1520-0477(2001)082<0695:MSEFIW>2.3.CO;2 [9] 陈静, 陈德辉, 颜宏.集合数值预报发展与研究进展.应用气象学报, 2002, 13(4):497-507. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20020465&flag=1 [10] 李泽椿, 陈德辉.国家气象中心集合预报数值业务系统的发展及应用.应用气象学报, 2002, 13(1):1-15. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20020101&flag=1 [11] Krishnamurti T N, Kishtawal C M, LaRow T, et al.Improved weather and seasonal climate foreeasts from multimodel superensemble.Science, 1999, 285:1548-1550. doi: 10.1126/science.285.5433.1548 [12] Krislmamurti T N, Kishtawal C M, Zhang Z, et al.Multi-model Superensemble Foreeasts for Weather and Seasonal Climate.FSU Report 9928, 1999. [13] Evans R E, Walters M K.Calibrated Probabilistic quantitative Preeipitation foreeasts based on the MRF ensemble.Wea Forecasting, 1998, 13:1132-1147. doi: 10.1175/1520-0434(1998)013<1132:CPQPFB>2.0.CO;2 [14] Mylne K R, Clark R T, Evans R E.Quasi-operational Multi-model Multi-analysis Ensembles on Medium range Time seales.AMS 13th Conf on Numerieal weather Predietion, Denver, Colorado, 1999:204-209. [15] Ebert E E.Ability of a Poor man's ensemble to prediet the probability and distribution of preeipitation.Mon Wea Rev, 2001, 129:2461-2479. doi: 10.1175/1520-0493(2001)129<2461:AOAPMS>2.0.CO;2 [16] 周兵, 赵翠光, 赵声蓉.多模式集合预报技术及其分析与检验.应用气象学报, 2006, 17(增刊):104-109. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20120207&flag=1 [17] 陈丽娟, 许力, 王永光.超级集合思想在汛期降水预测集成中的应用.气象, 2005, 31(5):52-54. doi: 10.7519/j.issn.1000-0526.2005.05.011 [18] 赵声蓉.多模式温度集成预报.应用气象学报, 2006, 17(1):52-58. doi: 10.11898/1001-7313.20060109 [19] 智协飞, 林春泽, 白永清, 等.北半球中纬度地区地面气温的超级集合预报.气象科学, 2009, 29(5):569-574. http://www.cnki.com.cn/Article/CJFDTOTAL-QXKX200905002.htm [20] 陈超辉, 李崇银, 谭言科, 等.基于交叉验证的多模式超级集合预报方法研究.气象学报, 2010, 68(4):464-476. doi: 10.11898/1001-7313.20100410 [21] 段明铿, 王盘兴.一种新的集合预报权重平均方法.应用气象学报, 2006, 17(4):488-493. doi: 10.11898/1001-7313.20060416 [22] 林春泽, 智协飞, 韩艳, 等.基于TIGGE资料的地面气温多模式超级集合预报.应用气象学报, 2009, 20(6):706-712. doi: 10.11898/1001-7313.20090608 [23] Krishnamurti T N, Coauthors.Prediction of the diurnal change using a multimodel superensemble.Part I:Precipitation.Mon Wea Rev, 2007, 135:3613-3632.