A Sensitivity Study of the Moist Singular Vectors to Temporal and Spatial Scales in GRAPES-GEPS Global Ensemble Prediction System
-
摘要: 湿奇异向量（Moist Singular Vectors，简称MSVs）是包含了湿物理切线性过程计算得到的奇异向量。研究MSVs对最优化时间间隔（optimization time interval，简称OTI）及模式水平分辨率的敏感性对提高集合预报效果至关重要。本文基于中国气象局数值预报中心自主研发的全球/区域同化和预报系统（Global/Regional Assimilation and Prediction System，简称GRAPES）——全球集合预报系统（Global ensemble prediction system，简称GEPS）业务版本研究了4组不同时空尺度（不同OTI和水平分辨率）下的MSVs，从能量模、能量谱、空间剖面等方面分析热带外MSVs特征，并从等压面变量评分、降水评分、降水概率预报等方面评估不同初值的集合预报效果。结果表明：提高MSVs水平分辨率可使其扰动具有较大的增长率，缩短OTI后MSVs能量向上传播的趋势更明显，并可以在中尺度范围产生较大SVs扰动。不同OTI下初始MSVs相似性较低，结构差异较大。从集合预报的结果来看，OTI为24 h试验的集合扰动能量增长较大，集合离散度在预报的0～96 h有明显提升，特别是2 m温度，且近地面要素的outlier评分也有明显改进。进一步分析发现，提高水平分辨率和缩短OTI的MSVs能够提高降水概率预报，而降水评分显示，同一水平分辨率下，OTI越短评分越好，但是提高MSVs的水平分辨率并不一定会提升小雨到中雨量级的降水评分。
-
关键词:
- 湿奇异向量 /
- 最优时间间隔 /
- 集合预报 /
- GRAPES-GEPS全球集合预报系统
Abstract: The Singular Vectors (SVs) that include the linearized moist physical process in calculations are called Moist SVs (MSVs). The sensitivity study of MSVs to horizontal resolutions and optimization time intervals (OTI) is important for the ensemble forecasting system. Based on the operational version of Global/Regional Assimilation and Prediction System-Global ensemble prediction system (GRAPES-GEPS), which is independently developed by the China meteorological administration’s numerical forecast center, this paper analyzes the characteristics of the subtropical MSVs and their ensemble forecasts under four groups of experiments with different horizontal and temporal resolutions. The characteristics of MSVs in terms of energy norm, energy spectrum and spatial profile are analyzed, and the evaluation of the ensemble forecast with the four groups of experiments is made in terms of isopressure variable scores, precipitation scores, and precipitation probability predictions. An increase in the horizontal resolution of MSVs leads to an increase in the growth rate of their perturbation. The upward propagation of MSV energy is more obvious than the downward propagation with the reduced OTI, which also produces relatively large SV perturbations in the mesoscale ranges. Under different OTIs, the initial MSVs are less similar to each other and their structures are different from each other. From the perspective of ensemble forecasting, the average ensemble perturbed energy with the 24-h OTI increases greatly, and the ensemble spread is improved for the 0- to 96-h prediction, especially for the 2-m temperature and the outlier scores of the near-surface variables. It is further found that increasing the horizontal and temporal resolutions can improve the precipitation probability prediction. The precipitation scores show that at the same spatial resolution, the shorter the OTI, the better the scores, while increasing the horizontal resolution of the MSVs fails to improve the precipitation scores for the light to moderate rains. -
图 1 （a）初始时刻及（b）最后演化时刻的能量模垂直分布（虚线表示动能KIN，实线表示总能量）
Figure 1. Vertical distributions of energy norm at (a) the initial time and (b) the evolved final time. The dashed lines indicate the kinetic energy (KIN), and the solid lines indicate the total energy
图 2 （a）初始时刻及（b）最后演化时刻奇异向量SVs在第26层（约500 hPa）能量谱（单位：10−9 J kg−1）
Figure 2. Energy spectrum (units: 10−9 J kg−1) of SVs (Singular Vectors) at level 26 (about 500 hPa) at (a) the initial time and (b) the evolved final time
图 3 （a–d）四组试验初始时刻（5月1日12:00，协调世界时，下同）北半球前15个MSVs在第26层（约500 hPa）扰动位温（填色，单位：10−3 K）水平分布，黑色线为控制预报500 hPa位势高度
Figure 3. Perturbed potential temperature (shaded, units: 10−3 K) of the first 15 MSVs in the Northern Hemisphere at level 26 (about 500 hPa) at the initial time (1200 UTC 1 May) in (a–d) the four numerical experiments. The black solid line indicates the 500-hPa geopotential height with controlled forecast
图 4 不同预报时长集合预报5天平均扰动能量（单位：J kg−1）随高度的分布：（a）0 h；（b）24 h；（c）48 h；（d）120 h
Figure 4. Distributions of ensemble forecast on 5-day average perturbation energy (units: J kg−1) with height at different forecast time: (a) 0 h; (b) 24 h; (c) 48 h; (d) 120 h
图 5 离散度（左列）、均方根误差（中间列）以及集合一致性（右列）随时间的变化：（a1−a3）500 hPa纬向风（单位：m s−1）；（b1–b3）850 hPa纬向风（单位：m s−1）；（c1–c3) 近地面10米纬向风（单位：m s−1）；（d1–d3）近地面2米温度（单位：°C）
Figure 5. The spread (left column), root mean square error (RMSE; middle column), and consistency (right column) over time: (a1–a3) Zonal wind of 500 hPa (units: m s−1); (b1–b3) zonal wind of 850 hPa (units: m s−1); (c1–c3) 10-m zonal wind near the ground (units: m s−1); (d1–d3) 2-m temperature near the ground (units: °C)
图 6 四组试验的集合预报outlier评分：（a）500 hPa纬向风；（b）850 hPa纬向风；（c）近地面10米纬向风；（d）近地面2米温度
Figure 6. Outlier scores of the ensemble forecast: (a) Zonal wind of 500 hPa; (b) zonal wind of 850 hPa; (c) 10-m zonal wind near the ground; (d) 2-m temperature near the ground
图 7 （a–d）四组试验北半球2 m温度的集合离散度水平分布（填色，单位：°C）
Figure 7. Horizontal distribution of the spread (shaded, units: °C) of 2-m temperature in the Northern Hemisphere in (a–d) the four numerical experiments
图 8 2019年5月4～5日（a）24 h累计实况降水量（单位：mm）分布及（b–e）四组试验大于25 mm量级降水的概率预报分布
Figure 8. (a) The distribution of 24-h cumulative observed precipitation and (b–e) the distribution of the probability forecast of precipitation greater than 25 mm in the four numerical experiments from 1200 UTC May 4 to 1200 UTC May 5, 2019
图 9 四组降水集合预报的（a、b）AROC（相对作用曲线面积）评分及（c、d）Brier降水评分：（a、c）10 mm降水量级；（b、d）25 mm降水量级
Figure 9. (a, b) AROC (Area under the Relative Operating characteristic Curve) precipitation scores and (c, d) Brier Scores of the four numerical experiments: (a,c) 10-mm precipitation; (b, d) 25-mm precipitation
表 1 时空分辨率对照试验的设置
Table 1. Settings of horizontal resolution and OTI
试验名称 计算SVs的水平分辨率 OTI R25t48 2.5°×2.5° 48 h R25t24 2.5°×2.5° 24 h R15t48 1.5°×1.5° 48 h R15t24 1.5°×1.5° 24 h 
下载: 导出CSV
表 2 R25t24与R25t48前5个MSVs第20层的纬向风扰动相似性
Table 2. Similarity of R25t24and R25t48（Zonal wind disturbance of the first 5 MSVs at level 20）
R25t24与R25t48前5个MSVs的相关系数 SV01
（R25t48）SV02
（R25t48）SV03
（R25t48）SV04
（R25t48）SV05
（R25t48）SV01
（R25t24）0.9018 −0.0198 0.0077 −0.1804 0.0265 SV02
（R25t24）0.0005 0.0048 −0.3905 0.0054 0.0201 SV03
（R25t24）−0.0031 −0.0054 0.5053 −0.0093 −0.0241 SV04
（R25t24）0.0008 −0.0087 0.0052 −0.019 0.0117 SV05
（R25t24）0.0595 −0.7518 0.0032 −0.1697 0.2747 注：加粗数值为相关系数较大值，下同。
下载: 导出CSV
表 3 R15t24与R15t48前5个MSVs第20层的纬向风扰动相似性
Table 3. Similarity of R15t48 and R15t24（Zonal wind disturbance of the first 5 MSVs at level 20）
R15t24与R15t48前5个MSVs的相关系数 SV01
（R15t48）SV02
（R15t48）SV03
（R15t48）SV04
（R15t48）SV05
（R15t48）SV01
（R15t24）−0.842 0.0083 0.0091 −0.1153 −0.0125 SV02
（R15t24）−0.0206 −0.0377 −0.6231 −0.0044 −0.3241 SV03
（R15t24）0.0054 −0.0212 −0.1722 −0.0255 0.2409 SV04
（R15t24）0.0284 −0.0175 0.0028 0.0172 0.005 SV05
（R15t24）−0.0143 0.1806 −0.0038 −0.0808 0.0333
下载: 导出CSV
表 4 时间步长分别为600秒和1200秒的两组试验相似性
Table 4. Similarity of the two experiments with 600s and 1200s integration timesteps
时间步长为1200 s试验和600 s试验的相关系数 SV01
（600 s）SV02
（600 s）SV03
（600 s）SV04
（600 s）SV05
（600 s）SV06
（600 s）SV07
（600 s）SV01（1200 s） 0.0925 0.0116 0.0675 0.9726 0.0336 0.0188 −0.0837 SV02（1200 s） 0.0078 0.1557 0.9011 −0.0015 −0.3285 −0.0124 0.0086 SV03（1200 s） −0.0002 0.0163 0.2209 −0.0044 0.7273 −0.0764 0.0912 SV04（1200 s） 0.0167 0.0122 0.0262 −0.0178 0.0912 0.1329 −0.9794 SV05（1200 s） −0.0084 0.1245 −0.138 0.0345 −0.1544 −0.0824 −0.2349 SV06（1200 s） 0.1423 0.0118 0.1264 −0.2598 0.4571 0.9593 0.0192 SV07（1200 s） 0.0079 0.6792 −0.1395 −0.0238 0.0147 0.0533 −0.0342
下载: 导出CSV
表 5 OTI分别为24 h和48 h的两组试验相似性对比
Table 5. Similarities of the two experiments at different optimization times of 24 h and 48 h
OTI为48 h试验和24 h试验的相关系数 SV01
（24 h）SV02
（24 h）SV03
（24 h）SV04
（24 h）SV05
（24 h）SV06
（24 h）SV07
（24 h）SV01（48 h） −0.0789 −0.0101 −0.0569 −0.8556 0.0108 0.0322 0.0472 SV02（48 h） 0.0115 0.9061 −0.2294 0.0119 −0.0012 −0.0059 0.0074 SV03（48 h） 0.0087 −0.1402 −0.5182 0.0174 0.1627 0.0128 −0.0229 SV04（48 h） −0.1218 0.0639 0.0599 −0.1523 0.0781 0.0931 −0.0113 SV05（48 h） 0.014 −0.0943 −0.4332 −0.0025 0.1845 −0.0369 0.0181 SV06（48 h） 0.0022 0.0364 −0.0117 −0.0011 0.0131 −0.0031 −0.0108 SV07（48 h） 0.0642 −0.0434 −0.2256 −0.2374 −0.017 −0.1517 0.057
下载: 导出CSV
-
[1] Buizza R. 1994. Sensitivity of optimal unstable structures [J]. Quart. J. Roy. Meteor. Soc., 120(516): 429−451. doi: 10.1002/qj.49712051609 [2] Buizza R, Palmer T N. 1995. The singular-vector structure of the atmospheric global circulation [J]. J. Atmos. Sci., 52(9): 1434−1456. doi:10.1175/1520-0469(1995)052<1434:TSVSOT>2.0.CO;2 [3] Buizza R, Palmer T N. 1998. Impact of ensemble size on ensemble prediction [J]. Mon. Wea. Rev., 126(9): 2503−2518. doi:10.1175/1520-0493(1998)126<2503:IOESOE>2.0.CO;2 [4] Buizza R, Milleer M, Palmer T N. 1999. Stochastic representation of model uncertainties in the ECMWF ensemble prediction system [J]. Quart. J. Roy. Meteor. Soc., 125(560): 2887−2908. doi: 10.1002/qj.49712556006 [5] 陈德辉, 沈学顺. 2006. 新一代数值预报系统GRAPES研究进展 [J]. 应用气象学报, 17(6): 773−777. doi: 10.3969/j.issn.1001-7313.2006.06.014Chen Dehui, Shen Xueshun. 2006. Recent progress on GRAPES research and application [J]. J. Appl. Meteor. Sci. (in Chinese), 17(6): 773−777. doi: 10.3969/j.issn.1001-7313.2006.06.014 [6] Coutinho M M, Hoskins B J, Buizza R. 2004. The influence of physical processes on extratropical singular vectors [J]. J. Atmos. Sci., 61(2): 195−209. doi:10.1175/1520-0469(2004)061<0195:TIOPPO>2.0.CO;2 [7] Diaconescu E P, Laprise R. 2012. Singular vectors in atmospheric sciences: A review [J]. Earth-Sci. Rev., 113(3–4): 161−175. doi: 10.1016/j.earscirev.2012.05.005 [8] Hoskins B J, Coutinho M M. 2005. Moist singular vectors and the predictability of some high impact European cyclones [J]. Quart. J. Roy. Meteor. Soc., 131(606): 581−601. doi: 10.1256/qj.04.48 [9] 霍振华, 刘永柱, 陈静, 等. 2020. 热带气旋奇异向量在GRAPES全球集合预报中的初步应用 [J]. 气象学报, 78(1): 48−59. doi: 10.11676/qxxb2020.006Huo Zhenhua, Liu Yongzhu, Chen Jing, et al. 2020. The preliminary application of tropical cyclone targeted singular vectors in the GRAPES global ensemble forecasts [J]. Acta Meteor. Sinica (in Chinese), 78(1): 48−59. doi: 10.11676/qxxb2020.006 [10] Komori T, Kadowaki T. 2010. Resolution dependence of singular vectors computed for typhoon SINLAKU [J]. SOLA, 6: 45−48. doi: 10.2151/sola.2010-012 [11] 李晓莉, 刘永柱. 2019. GRAPES全球奇异向量方法改进及试验分析 [J]. 气象学报, 77(3): 552−562. doi: 10.11676/qxxb2019.020Li Xiaoli, Liu Yongzhu. 2019. The improvement of GRAPES global extratropical singular vectors and experimental study [J]. Acta Meteor. Sinica (in Chinese), 77(3): 552−562. doi: 10.11676/qxxb2019.020 [12] 李晓莉, 陈静, 刘永柱, 等. 2019. GRAPES全球集合预报初始条件及模式物理过程不确定性方法研究 [J]. 大气科学学报, 42(3): 348−359. doi: 10.13878/j.cnki.dqkxxb.20190318001Li Xiaoli, Chen Jing, Liu Yongzhu, et al. 2019. Representations of initial uncertainty and model uncertainty of GRAPES global ensemble forecasting [J]. Trans. Atmos. Sci. (in Chinese), 42(3): 348−359. doi: 10.13878/j.cnki.dqkxxb.20190318001 [13] 刘永柱, 沈学顺, 李晓莉. 2013. 基于总能量模的GRAPES全球模式奇异向量扰动研究 [J]. 气象学报, 71(3): 517−526. doi: 10.11676/qxxb2013.043Liu Yongzhu, Shen Xueshun, Li Xiaoli. 2013. Research on the singular vector perturbation of the GRAPES global model based on the total energy norm [J]. Acta Meteor. Sinica (in Chinese), 71(3): 517−526. doi: 10.11676/qxxb2013.043 [14] 刘永柱, 张林, 金之雁. 2017. GRAPES全球切线性和伴随模式的调优 [J]. 应用气象学报, 28(1): 62−71. doi: 10.11898/1001-7313.20170106Liu Yongzhu, Zhang Lin, Jin Zhiyan. 2017. The optimization of GRAPES Global tangent linear model and adjoint model [J]. J. Appl. Meteor. Sci. (in Chinese), 28(1): 62−71. doi: 10.11898/1001-7313.20170106 [15] 刘永柱, 龚建东, 张林, 等. 2019. 线性化物理过程对GRAPES 4DVAR同化的影响 [J]. 气象学报, 77(2): 196−209. doi: 10.11676/qxxb2019.013Liu Yongzhu, Gong Jiandong, Zhang Lin, et al. 2019. Influence of linearized physical processes on the GRAPES 4DVAR [J]. Acta Meteor. Sinica (in Chinese), 77(2): 196−209. doi: 10.11676/qxxb2019.013 [16] Lorenz E N. 1965. A study of the predictability of a 28-variable atmospheric model [J]. Tellus, 17(3): 321−333. doi: 10.3402/tellusa.v17i3.9076 [17] Molteni F, Palmer T N. 1993. Predictability and finite-time instability of the northern winter circulation [J]. Quart. J. Roy. Meteor. Soc., 119(510): 269−298. doi: 10.1002/qj.49711951004 [18] Molteni F, Buizza R, Palmer T N, et al. 1996. The ECMWF ensemble prediction system: Methodology and validation [J]. Quart. J. Roy. Meteor. Soc., 122(529): 73−119. doi: 10.1002/qj.49712252905 [19] Palmer T N, Molteni F, Mureau R, et al. 1993. Ensemble prediction [C]//Proceedings of ECMWF Seminar on Validation of Models over Europe. Reading, UK: Shinfield Park, 21–66 [20] Palmer T N, Gelaro R, Barkmeijer J, et al. 1998. Singular vectors, metrics, and adaptive observations [J]. J. Atmos. Sci., 55(4): 633−653. doi:10.1175/1520-0469(1998)055<0633:SVMAAO>2.0.CO;2 [21] Palmer T N, Buizza R, Leutbecher M, et al. 2007. The ensemble prediction system – recent and ongoing developments [R]. ECMWF Technical Memoranda. No. 540. doi: 10.21957/bh84yl1rj [22] Puri K, Barkmeijer J, Palmer T N. 2001. Ensemble prediction of tropical cyclones using targeted diabatic singular vectors [J]. Quart. J. Roy. Meteor. Soc., 127(572): 709−731. doi: 10.1002/qj.49712757222 [23] Tompkins A M, Janisková M. 2004. A cloud scheme for data assimilation: Description and initial tests [J]. Quart. J. Roy. Meteor. Soc., 130(602): 2495−2517. doi: 10.1256/qj.03.162 [24] Walser A, Arpagaus M, Appenzeller C, et al. 2006. The impact of moist singular vectors and horizontal resolution on short-range limited-area ensemble forecasts for two European winter storms [J]. Mon. Wea. Rev., 134(10): 2877−2887. doi: 10.1175/MWR3210.1 [25] Wang J, Wang B, Liu J J, et al. 2020. Application and characteristic analysis of the moist singular vector in GRAPES-GEPS [J]. Adv. Atmos. Sci., 37(11): 1164−1178. doi: 10.1007/s00376-020-0092-9 [26] 薛纪善, 陈德辉. 2008. 数值预报系统GRAPES的科学设计与应用 [M]. 北京: 科学出版社, 1–383.Xue Jishan, Chen Dehui. 2008. Scientific Design and Application of Numerical Prediction System GRAPES (in Chinese) [M]. Beijing: Science Press, 1–383. [27] 叶璐, 刘永柱, 陈静, 等. 2020. 集合预报多尺度奇异向量初值扰动方法研究 [J]. 气象学报, 78(4): 648−664. doi: 10.11676/qxxb2020.042Ye Lu, Liu Yongzhu, Chen Jing, et al. 2020. A study on multi–scale singular vector initial perturbation method for ensemble prediction [J]. Acta Meteor. Sinica (in Chinese), 78(4): 648−664. doi: 10.11676/qxxb2020.042 [28] Zadra A, Buehner M, Laroche S, et al. 2004. Impact of the GEM model simplified physics on extratropical singular vectors [J]. Quart. J. Roy. Meteor. Soc., 130(602): 2541−2569. doi: 10.1256/qj.03.208 -
点击查看大图
计量
- 文章访问数: 103
- HTML全文浏览量: 23
- PDF下载量: 34
- 被引次数: 0
