Implementation of Hybrid En-3DVAR Assimilation Scheme in Rapid Cycling Assimilation System
-
摘要: 基于北京市气象局快速循环同化系统RMAPS-ST以及对流尺度集合预报系统RMAPS-EN,构建了En-3DVAR集合变分混合同化系统,将该系统应用到业务快速循环同化系统中并进行试验,分别在冷启动与循环启动环境下对比了混合同化系统(Hybrid)与三维变分(3DVAR)的同化预报效果。获得的结论如下:单点试验结果表明,混合同化系统分析增量的分布与集合预报离散度分布具有较好的对应关系;在冷启动和循环启动中,三维变分的分析增量都表现出各向同性的特点,混合同化分析增量均表现出一定的流依赖特征;降水个例分析表明,在冷启动环境中,Hybrid与3DVAR效果相当,而在循环启动中,Hybrid的降水预报相对于3DVAR有较明显的改进效果;批量试验检验结果表明,冷启动中,Hybrid与3DVAR的评分大致相当,而在循环启动中,Hybrid相对于3DVAR的评分有明显改进;集合离散度和背景场误差的相关性分析表明二者在循环启动环境下具有更好的相关性。Abstract: A hybrid En-3DVAR (ensemble-three-dimensional variational data assimilation) was constructed based on the rapid refresh multi-scale analysis and prediction system–short term (RMAPS-ST) and the RMAPS-ensemble (RMAPS-EN) at Beijing Meteorological Bureau. The hybrid and 3DVAR schemes were conducted in the cold run and cycle run settings based on RMAPS-ST. The authors obtained the following conclusions: The single observation test showed that there was a good correspondence distribution between the hybrid data analysis (DA) increment and the ensemble spread. The hybrid DA-increment exhibited flow-dependent characteristics in both cold run and cycle run, but the 3DVAR DA-increment was isotropic. A rainfall case study revealed that the hybrid and 3DVAR had almost the same influence on the precipitation in the cold run setting, but the hybrid scores were significantly higher than those of the 3DVAR in the cycle run setting. The analysis of the ensemble spread and background error showed that they were better correlated in the cycle run than in the cold run.
-
图 3 2018年7月15日12时(协调世界时,下同)区域集合预报RMAPS-EN起报12 h预报时效的500 hPa纬向风集合离散度(阴影,单位:m s−1)和业务RMAPS-ST的d02区域(图1)分析场的位势高度场(等值线,单位:gpm)分布
Figure 3. Horizontal pattern of ensemble spread of 500-hPa zonal wind (shadings, units: m s−1) from 12-h forecast with ensemble forecast system of RMAPS-EN initiated at 1200 UTC 15 July 2018, and the corresponding geopotential height (solid lines, units: gpm) from the operational RMAPS-ST analysis of domain d02 in Fig. 1
图 5 2018年7月16日00时四种同化方案在模式层第16层纬向风的分析增量(单位:m s−1):(a)3DVAR-cold方案;(b)Hybrid-cold方案;(c)3DVAR-cycle方案;(d)Hybrid-cycle方案
Figure 5. Analysis increments (units: m s−1) of zonal wind at the 16th model level for the four schemes at 0000 UTC 16 July 2018: (a) 3DVAR-cold scheme (3DVAR in the cold run); (b) hybrid-cold scheme (hybrid in the cold run); (c) 3DVAR-cycle scheme (3DVAR in the partial cycle run); (d) hybrid-cycle scheme (hybrid in the partial cycle run)
图 6 2018年7月16日00时起报的18 h预报时效的(a)3DVAR-cold方案、(b)Hybrid-cold方案700 hPa高度风场(箭头,单位:m s−1)及相对湿度(阴影),以及(c)在冷启动、循环设置下Hybrid相对于3DVAR方案的700 hPa高度风场(箭头,单位:m s−1)、相对湿度的差异分布。(d–f)同(a–c),但为循环启动设置下物理量的分布
Figure 6. Distributions of wind vectors (arrows, units: m s−1) and relative humidity (shadings) at 700 hPa from the 18th-hour forecast obtained from (a) 3DVAR-cold scheme, (b) hybrid-cold scheme, and (c) the differences between 3DVAR-cold scheme and hybrid-cold scheme at 0000 UTC 16 July 2018. (d–f) As in (a–c), but for distributions for 3DVAR-cycle scheme and hybrid-cycle scheme
图 7 2018年7月16日18时至2018年7月17日00时的6 h累计降水量分布实况以及四种方案对该时段降水量的预报(单位:mm):(a)实况;(b)3DVAR-cold方案预报;(c)Hybrid-cold方案预报;(d)3DVAR-cycle方案预报;(e)Hybrid-cycle方案预报
Figure 7. Six-hour accumulated precipitation (units: mm) obtained from observation and forecast from the four schemes from 1800 UTC 16 July to 0000 UTC 17 July 2018: (a) Observation; (b) forecast from 3DVAR-cold scheme; (c) forecast from hybrid-cold scheme; (d) forecast from 3DVAR-cycle scheme; (e) forecast from hybrid-cycle scheme
图 8 2018年7月16日00时至7月17日00时的逐6 h累计降水量的四种方案(3DVAR-cold、hybrid-cold、3DVAR-cycle、hybrid-cycle)预报的不同降水量级的TS评分和Bias评分:(a)0.1 mm TS评分;(b)1 mm TS评分;(c)10 mm TS评分;(d)0.1 mm Bias评分;(e)1 mm Bias评分;(f)10 mm Bias评分
Figure 8. The TS (threat score) and Bias score of 6-h accumulative forecast precipitation from 0000 UTC 16 July to 0000 UTC 17 July 2018 in four schemes (3DVAR-cold scheme, hybrid-cold scheme, 3DVAR-cycle scheme, hybrid-cycle scheme): (a) 0.1 mm TS; (b) 1 mm TS; (c) 10 mm TS; (d) 0.1 mm Bias score; (e) 1 mm Bias score; (f) 10 mm Bias score
图 9 四种方案(3DVAR-cold、hybrid-cold、3DVAR-cycle、hybrid-cycle)(a)2 m高度处温度、(b)10 m高度处纬向风的12 h、24 h预报均方根误差
Figure 9. RMSE of (a) temperature at 2-m height, (b) zonal wind at 10-m height for the 12th and 24th forecast hours in the four schemes (3DVAR-cold scheme, hybrid-cold scheme, 3DVAR-cycle scheme, hybrid-cycle scheme)
表 1 试验设置
Table 1. Test settings
试验方案 背景场 背景误差协方差 同化方法 3DVAR-cold ECMWF分析场 NMC静态 3DVAR Hybrid-cold ECMWF分析场 NMC静态+集合 Hybrid 3DVAR-cycle RMAPS-ST 3 h循环预报场 NMC静态 3DVAR Hybrid-cycle RMAPS-ST 3 h循环预报场 NMC静态+集合 Hybrid 表 2 不同背景场设置(cold和cycle)下Hybrid同化方案相较于3DVAR同化方案的各物理量不同预报时效的均方根误差改进百分比
Table 2. The improvement RMSE (Root Mean Square Error) percentage of different variables at different forecast lead time for the hybrid relative to 3DVAR in different backgrounds (cold and cycle runs)
物理量 不同预报时效的均方根误差改进百分比(cold) 不同预报时效的均方根误差改进百分比(cycle) 6 h 12 h 18 h 24 h 6 h 12 h 18 h 24 h 200 hPa纬向风 2.59% −2.13% −0.58% 1.97% 10.45% −0.45% 1.62% 1.18% 500 hPa纬向风 3.56% 0.41% 1.44% −0.11% 3.34% 1.02% 0.58% 0.79% 850 hPa纬向风 1.19% 0.33% 1.29% −2.61% 1.24% 2.81% 2.43% −1.05% 200 hPa经向风 3.91% −2.43% −2.87% −0.58% 8.61% 3.13% −0.27% 2.87% 500 hPa经向风 1.99% 0.82% 1.90% 1.17% 3.22% 4.09% 6.91% 1.04% 850 hPa经向风 0.07% 0.76% 1.00% 1.08% 0.89% 1.20% 2.16% −2.32% 500 hPa温度 0.96% −1.77% -0.17% −2.69% 3.16% 5.04% 1.10% −0.78% 850 hPa温度 −0.06% 0.19% 0.77% −0.81% 1.45% 4.03% 1.44% 0.24% 500 hPa位势高度 −0.35% 1.55% 2.10% 2.38% −0.65% 2.73% 1.43% 1.66% 700 hPa相对湿度 −0.02% 2.02% 2.69% 0.34% 1.02% 3.15% 1.18% 0.58% 850 hPa相对湿度 −0.02% −0.40% 0.18% −0.47% 1.04% 1.26% 1.08% 1.21% -
[1] Anderson J L. 2001. An ensemble adjustment Kalman filter for data assimilation [J]. Mon. Wea. Rev., 129(12): 2884−2903. doi:10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2 [2] Barker D M, Huang W, Guo Y R, et al. 2004. A three-dimensional variational data assimilation system for MM5: Implementation and initial results [J]. Mon. Wea. Rev., 132(4): 897−914. doi:10.1175/1520-0493(2004)132<0897:ATVDAS>2.0.CO;2 [3] Benjamin S G, Weygandt S S, Brown J M, et al. 2016. A North American hourly assimilation and model forecast cycle: The rapid refresh [J]. Mon. Wea. Rev., 144(4): 1669−1694. doi: 10.1175/MWR-D-15-0242.1 [4] Bishop C H, Etherton B J, Majumdar S J. 2001. Adaptive sampling with the ensemble transform Kalman filter. Part I: Theoretical aspects [J]. Mon. Wea. Rev., 129(3): 420−436. doi:10.1175/1520-0493(2001)129<0420:ASWTET>2.0.CO;2 [5] Bishop C H, Holt T R, Nachamkin J, et al. 2009. Regional ensemble forecasts using the ensemble transform technique [J]. Mon. Wea. Rev., 137(1): 288−298. doi: 10.1175/2008MWR2559.1 [6] Bratseth A. 1986. Statistical interpolation by means of successive corrections [J]. Tellus A, 38(5): 439−447. doi: 10.1111/j.1600-0870.1986.tb00476.x [7] Buehner M. 2005. Ensemble-derived stationary and flow-dependent background-error covariances: Evaluation in a quasi-operational NWP setting [J]. Quart. J. Roy. Meteor. Soc., 131(607): 1013−1043. doi: 10.1256/qj.04.15 [8] Buehner M, Houtekamer P L, Charette C, et al. 2010a. Intercomparison of variational data assimilation and the ensemble Kalman filter for global deterministic NWP. Part I: Description and single-observation experiments [J]. Mon. Wea. Rev., 138(5): 1550−1566. doi: 10.1175/2009MWR3157.1 [9] Buehner M, Houtekamer P L, Charette C, et al. 2010b. Intercomparison of variational data assimilation and the ensemble Kalman filter for global deterministic NWP. Part II: One-month experiments with real observations [J]. Mon. Wea. Rev., 138(5): 1567−1586. doi: 10.1175/2009MWR3158.1 [10] 陈敏, 范水勇, 仲跻芹, 等. 2010. 全球定位系统的可降水量资料在北京地区快速更新循环系统中的同化试验 [J]. 气象学报, 68(4): 450−463. doi: 10.11676/qxxb2010.044Chen Min, Fan Shuiyong, Zhong Jiqin, et al. 2010. An experimental study of assimilating the global position system–precipitable water vapor observations into the rapid updated cycle system for the Beijing area [J]. Acta Meteorologica Sinica (in Chinese), 68(4): 450−463. doi: 10.11676/qxxb2010.044 [11] Chen L L, Chen J, Xue J S, et al. 2015. Development and testing of the GRAPES regional ensemble-3DVAR hybrid data assimilation system [J]. J. Meteor. Res., 29(6): 981−996. doi: 10.1007/s13351-015-5021-y [12] Clayton A M, Lorenc A C, Barker D M. 2013. Operational implementation of a hybrid ensemble/4D-Var global data assimilation system at the Met Office [J]. Quart. J. Roy. Meteor. Soc., 139(675): 1445−1461. doi: 10.1002/qj.2054 [13] Courtier P, Andersson E, Heckley W, et al. 1998. The ECMWF implementation of three-dimensional variational assimilation (3D-Var). I: Formulation [J]. Quart. J. Roy. Meteor. Soc., 124(550): 1783−1807. doi: 10.1002/qj.49712455002 [14] 范水勇, 陈敏, 仲跻芹, 等. 2009. 北京地区高分辨率快速循环同化预报系统性能检验和评估 [J]. 暴雨灾害, 28(2): 119−125. doi: 10.3969/j.issn.1004-9045.2009.02.004Fan Shuiyong, Chen Min, Zhong Jiqin, et al. 2009. Performance tests and evaluations of Beijing local high-resolution rapid update cycle system [J]. Torrential Rain and Disasters (in Chinese), 28(2): 119−125. doi: 10.3969/j.issn.1004-9045.2009.02.004 [15] Hamill T M, Snyder C. 2000. A hybrid ensemble Kalman filter–3D variational analysis scheme [J]. Mon. Wea. Rev., 128(8): 2905−2919. doi:10.1175/1520-0493(2000)128<2905:AHEKFV>2.0.CO;2 [16] Hamill T M, Whitaker J S, Fiorino M, et al. 2011. Global ensemble predictions of 2009’s tropical cyclones initialized with an ensemble Kalman filter [J]. Mon. Wea. Rev., 139(2): 668−688. doi: 10.1175/2010MWR3456.1 [17] 何静, 陈敏, 仲跻芹, 等. 2019. 雷达反射率三维拼图观测资料在北方区域数值模式预报系统中的同化应用研究 [J]. 气象学报, 77(2): 210−232. doi: 10.11676/qxxb2019.005He Jing, Chen Min, Zhong Jiqin, et al. 2019. A study of three-dimensional radar reflectivity mosaic assimilation in the regional forecasting model for North China [J]. Acta Meteorologica Sinica (in Chinese), 77(2): 210−232. doi: 10.11676/qxxb2019.005 [18] Liu C S, Xiao Q N. 2013. An ensemble-based four-dimensional variational data assimilation scheme. Part III: Antarctic applications with advanced research WRF using real data [J]. Mon. Wea. Rev., 141(8): 2721−2739. doi: 10.1175/MWR-D-12-00130.1 [19] Liu C S, Xiao Q N, Wang B. 2008. An ensemble-based four-dimensional variational data assimilation scheme. Part I: Technical formulation and preliminary test [J]. Mon. Wea. Rev., 136(9): 3363−3373. doi: 10.1175/2008MWR2312.1 [20] Liu C S, Xiao Q N, Wang B. 2009. An ensemble-based four-dimensional variational data assimilation scheme. Part II: Observing system simulation experiments with advanced research WRF (ARW) [J]. Mon. Wea. Rev., 137(5): 1687−1704. doi: 10.1175/2008MWR2699.1 [21] Lorenc A C. 1986. Analysis methods for numerical weather prediction [J]. Quart. J. Roy. Meteor. Soc., 112(474): 1177−1194. doi: 10.1002/qj.49711247414 [22] Lorenc A C. 2003. The potential of the ensemble Kalman filter for NWP—A comparison with 4D-Var [J]. Quart. J. Roy. Meteor. Soc., 129(595): 3183−3203. doi: 10.1256/qj.02.132 [23] Lorenc A C. 2017. Improving ensemble covariances in hybrid variational data assimilation without increasing ensemble size [J]. Quart. J. Roy. Meteor. Soc., 143(703): 1062−1072. doi: 10.1002/qj.2990 [24] 马旭林, 陆续, 于月明, 等. 2014. 数值天气预报中集合—变分混合资料同化及其研究进展 [J]. 热带气象学报, 30(6): 1188−1195. doi: 10.3969/j.issn.1004-4965.2014.06.020Ma Xulin, Lu Xu, Yu Yueming, et al. 2014. Progress on hybrid ensemble–variational data assimilation in numerical weather prediction [J]. Journal of Tropical Meteorology (in Chinese), 30(6): 1188−1195. doi: 10.3969/j.issn.1004-4965.2014.06.020 [25] 马旭林, 李琳琳, 周勃旸, 等. 2005. 台风预报误差的流依赖特征及混合资料同化中最优耦合系数 [J]. 大气科学学报, 38(6): 766−775. doi: 10.13878/j.cnki.dqkxxb.20141224001Ma Xulin, Li Linlin, Zhou Boyang, et al. 2005. Flow-dependent characteristics of typhoon forecasting errors and optimal coupling coefficient in hybrid data assimilation [J]. Transactions of Atmospheric Sciences (in Chinese), 38(6): 766−775. doi: 10.13878/j.cnki.dqkxxb.20141224001 [26] Meng Z Y, Zhang F Q. 2007. Tests of an ensemble Kalman filter for mesoscale and regional-scale data assimilation. Part II: Imperfect model experiments [J]. Mon. Wea. Rev., 135(4): 1403−1423. doi: 10.1175/MWR3352.1 [27] Parrish D F, Derber J C. 1992. The national meteorological center’s spectral statistical-interpolation analysis system [J]. Mon. Wea. Rev., 120(8): 1747−1763. doi:10.1175/1520-0493(1992)120<1747:TNMCSS>2.0.CO;2 [28] Rabier F, Järvinen H, Klinker E, et al. 2000. The ECMWF operational implementation of four-dimensional variational assimilation. I: Experimental results with simplified physics [J]. Quart. J. Roy. Meteor. Soc., 126(564): 1143−1170. doi: 10.1002/qj.49712656415 [29] 沈菲菲, 闵锦忠, 许冬梅, 等. 2016. Hybrid ETKF-3DVAR方法同化多普勒雷达速度观测资料. Ⅰ: 模拟资料试验 [J]. 大气科学学报, 39(1): 81−89. doi: 10.13878/j.cnki.dqkxxb.20140210001Shen Feifei, Min Jinzhong, Xu Dongmei, et al. 2016. Assimilation of Doppler radar velocity observations with hybrid ETKF-3DVAR method. Part Ⅰ: Experiments with simulated data [J]. Transactions of Atmospheric Sciences (in Chinese), 39(1): 81−89. doi: 10.13878/j.cnki.dqkxxb.20140210001 [30] Tong M J, Xue M. 2005. Ensemble Kalman filter assimilation of Doppler radar data with a compressible nonhydrostatic model: OSS experiments [J]. Mon. Wea. Rev., 133(7): 1789−1807. doi: 10.1175/MWR2898.1 [31] Toth Z, Zhu Y J, Marchok T. 2001. The use of ensembles to identify forecasts with small and large uncertainty [J]. Wea. Forecasting, 16(4): 463−477. doi:10.1175/1520-0434(2001)016<0463:TUOETI>2.0.CO;2 [32] Wang X G. 2011. Application of the WRF hybrid ETKF-3DVAR data assimilation system for hurricane track forecasts [J]. Wea. Forecasting, 26(6): 868−884. doi: 10.1175/WAF-D-10-05058.1 [33] Wang X G, Snyder C, Hamill T M. 2007a. On the theoretical equivalence of differently proposed ensemble-3DVAR hybrid analysis schemes [J]. Mon. Wea. Rev., 135(1): 222−227. doi: 10.1175/MWR3282.1 [34] Wang X G, Hamill T M, Whitaker J S, et al. 2007b. A comparison of hybrid ensemble transform Kalman filter-optimum interpolation and ensemble square root filter analysis schemes [J]. Mon. Wea. Rev., 135(3): 1055−1076. doi: 10.1175/MWR3307.1 [35] Wang X G, Barker D M, Snyder C, et al. 2008a. A hybrid ETKF-3DVAR data assimilation scheme for the WRF model. Part I: Observing system simulation experiment [J]. Mon. Wea. Rev., 136(12): 5116−5131. doi: 10.1175/2008MWR2444.1 [36] Wang X G, Barker D M, Snyder C, et al. 2008b. A hybrid ETKF-3DVAR data assimilation scheme for the WRF model. Part II: Real observation experiments [J]. Mon. Wea. Rev., 136(12): 5132−5147. doi: 10.1175/2008MWR2445.1 [37] Wang X G, Parrish D, Kleist D, et al. 2013. GSI 3DVAR-based ensemble–variational hybrid data assimilation for NCEP global forecast system: Single-resolution experiments [J]. Mon. Wea. Rev., 141(11): 4098−4117. doi: 10.1175/MWR-D-12-00141.1 [38] Wei M Z, Toth Z. 2003. A new measure of ensemble performance: Perturbation versus error correlation analysis (PECA) [J]. Mon. Wea. Rev., 131(8): 1549−1565. doi:10.1175//1520-0493(2003)131<1549:ANMOEP>2.0.CO;2 [39] Whitaker J S, Hamill T M. 2002. Ensemble data assimilation without perturbed observations [J]. Mon. Wea. Rev., 130(7): 1913−1924. doi:10.1175/1520-0493(2002)130<1913:EDAWPO>2.0.CO;2 [40] 夏宇, 陈静, 刘艳, 等. 2018. GRAPES混合同化方法在青藏高原区域的初步试验 [J]. 大气科学学报, 41(2): 239−247. doi: 10.13878/j.cnki.dqkxxb.20160119001Xia Yu, Chen Jing, LiuYan, et al. 2018. A tentative experiment of GRAPES En-3DVAR hybrid data assimilation method over the Tibet Plateau [J]. Transactions of Atmospheric Sciences (in Chinese), 41(2): 239−247. doi: 10.13878/j.cnki.dqkxxb.20160119001 [41] Xia Y, Chen J, Zhi X F, et al. 2018. Topographic dependent horizontal localization scale scheme in GRAPES-MESO hybrid En-3DVAR assimilation system [J]. Journal of Tropical Meteorology, 25(2): 245−256. [42] Xia Y, Chen J, Zhi X F, et al. 2020. The impact of ensemble model systematic bias on hybrid data assimilation system [J]. Journal of Meteorology, 34(2): 400−412. doi: 10.1007/s13351-020-9088-8 [43] Zhang F Q, Meng Z Y, Aksoy A. 2006. Tests of an ensemble Kalman filter for mesoscale and regional-scale data assimilation. Part I: Perfect model experiments [J]. Mon. Wea. Rev., 134(2): 722−736. doi: 10.1175/MWR3101.1 [44] 张明阳, 张立凤, 张斌, 等. 2015. 集合变分混合同化背景误差协方差流依赖性分析 [J]. 气象科学, 35(6): 728−736. doi: 10.3969/2015jms.0069Zhang Mingyang, Zhang Lifeng, Zhang Bin, et al. 2015. Flow-dependent characteristics of background error covariance in hybrid variational-ensemble data assimilation [J]. Journal of the Meteorological Sciences (in Chinese), 35(6): 728−736. doi: 10.3969/2015jms.0069 [45] 张涵斌, 李玉焕, 范水勇, 等. 2017. 基于动力降尺度的区域集合预报初值扰动构建方法研究 [J]. 气象, 43(12): 1461−1472. doi: 10.7519/j.issn.1000-0526.2017.12.002Zhang Hanbin, Li Yuhuan, Fan Shuiyong, et al. 2017. Study on initial perturbation construction method for regional ensemble forecast based on dynamical downscaling [J]. Meteor. Mon. (in Chinese), 43(12): 1461−1472. doi: 10.7519/j.issn.1000-0526.2017.12.002 [46] 朱浩楠, 闵锦忠, 杜宁珠. 2016. HBFNEnKF混合同化方法设计及检验 [J]. 大气科学, 40(5): 995−1008. doi: 10.3878/j.issn.1006-9895.1510.15214Zhu Haonan, Min Jinzhong, Du Yuzhu. 2016. Implementation and testing of a hybrid back and forth nudging ensemble Kalman filter (HBFNEnKF) data assimilation method [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 40(5): 995−1008. doi: 10.3878/j.issn.1006-9895.1510.15214 [47] Zhu Y J, Toth Z, Wobus R, et al. 2002. The economic value of ensemble-based weather forecasts [J]. Bull. Amer. Meteor. Soc., 83(1): 73−84. doi:10.1175/1520-0477(2002)083<0073:TEVOEB>2.3.CO;2 -