BACKGROUND ERROR COVARIANCE CHARACTERISTICS BASED ON GSI ASSIMILATION SYSTEM AND ITS EFFECT ON PREDICTION RESULTS
-
摘要: 在同化系统中使用更合理的背景误差协方差对于得到更良好的同化效果至关重要。首先采用NMC方法针对中国区域构建更适合WRF-ARW区域预报系统的B矩阵,并对比分析了其与GSI同化系统预设的NCEP预报系统的B矩阵在分析变量间的平衡关系、分析控制变量的标准差、水平和垂直特征尺度等方面的特征差异。参照这些特征差异设计单点观测试验、背景误差协方差调优参数敏感性试验,确定针对中国区域构建B矩阵的最佳调优参数。并讨论其对一次季风低压强降水天气过程的循环同化和预报效果的影响。结果表明,采用最佳调优参数使用针对中国区域构建B矩阵的试验(Sen6)对V风分量场和相对湿度场的预报性能改进显著,同时也引出了GSI同化系统背景误差协方差参数调优(尤其是水平特征尺度参数调整)的两难问题。在此基础上,采用Hybrid同化方法使用针对中国区域构建B矩阵的循环同化试验(Hyb3)可以进一步改善预报效果,并在一定程度上修正个例模拟雨带的位置。Abstract: To improve assimilation results, a more reasonable background error covariance is crucial for the assimilation system. In this paper, first, the NMC method is adopted to construct a B matrix that is more suitable for the WRF-ARW regional prediction system for the Chinese region. Then the characteristics of the B matrix and the NCEP prediction system preset by the GSI assimilation system are compared and analyzed. The single-point observation test and background error covariance tuning parameter sensitivity test are designed with reference to the differences in their characteristics to determine the optimal tuning parameters for the development of B matrix for the Chinese region. The results show that the Sen6 experiment with the optimal tuning parameters and the B matrix constructed for the Chinese region has the best prediction effect. In particular, the prediction performance of V wind component field and relative humidity field has been improved significantly. At the same time, the dilemma of background error covariance parameter adjustment (especially horizontal length scale parameter adjustment) in the GSI assimilation system is also introduced. On this basis, the Hyb3 cyclic assimilation experiment with hybrid assimilation method and B matrix constructed for the Chinese region can further improve prediction results, and to some extent, the location of the simulated rain belt is modified.
-
Key words:
- data assimilation /
- GSI assimilation system /
- background error covariance
-
表 1 GSI同化系统背景误差协方差调整选项
文件 参数 功能 anavinfo as/tsfc_sdv 调整每个分析控制变量ψ、χu、Tu、Psu和RH的方差 namelist vs 全局调整所有分析控制变量的垂直特征尺度 hzscl(3) 全局调整所有分析控制变量的水平特征尺度 hswgt(3) 全局调整所有分析控制变量的方差 表 2 单点观测试验最佳拟合调优参数
调优因子 U风分量单点试验 温度单点试验 比湿单点试验 垂直特征尺度 0.5 0.7 0.7 水平特征尺度 3.5×hzscl2 6.0×hzscl2 6.0×hzscl2 方差权重 1.6×hswgt 1.2×hswgt 1.4×hswgt 表中hzscl2=0.373, 0.746, 1.500;hswgt=0.45, 0.30, 0.25是GSI同化系统针对区域应用推荐的调优参数。水平特征尺度和方差权重调优参数中的三个数值是递归滤波的三个系数。 表 3 背景误差协方差调优参数敏感性试验设计
试验 B矩阵 方差比例因子 垂直特征尺度 水平特征尺度 方差权重 Ctl1 NCEP GFS GSI同化系统推荐设置 hzscl1 hswgt Ctl2 NCEP NAM hzscl2 hswgt Ctl3 ARW China hzscl2 hswgt Sen1 ARW China 同4.2节 0.5 3.5×hzscl2 1.2×hswgt Sen2 ARW China 同4.2节 0.5 3.5×hzscl2 1.4×hswgt Sen3 ARW China 同4.2节 0.5 3.5×hzscl2 1.6×hswgt Sen4 ARW China 同4.2节 0.6 4.75×hzscl2 1.2×hswgt Sen5 ARW China 同4.2节 0.6 4.75×hzscl2 1.4×hswgt Sen6 ARW China 同4.2节 0.6 4.75×hzscl2 1.6×hswgt Sen7 ARW China 同4.2节 0.7 6.0×hzscl2 1.2×hswgt Sen8 ARW China 同4.2节 0.7 6.0×hzscl2 1.2×hswgt Sen9 ARW China 同4.2节 0.7 6.0×hzscl2 1.6×hswgt 表中hzscl1=1.7, 0.8, 0.5;hswgt=0.45, 0.30。0.25是GSI同化系统针对全球应用推荐的调优参数。Sen3、Sen7和Sen8试验采用的分别是通过大量单点观测试验调试出的分别最适合U风分量、温度和比湿的调优参数。 表 4 循环同化试验设计
试验 B矩阵 调优参数设置 NoDA - - Ctl1 NCEP GFS GSI同化系统推荐设置 Ctl2 NECP NAM GSI同化系统推荐设置 Sen6 ARW China 同表 3 Hyb1 25% NCEP GFS + 75% Ens GSI同化系统推荐设置 Hyb2 25% NECP NAM + 75% Ens GSI同化系统推荐设置 Hyb3 25% ARW China + 75% Ens 同Sen6 -
[1] PARRISH D F, DERBER J C. The National Meteorological Center's spectral statistical interpolation analysis system[J]. Mon Wea Rev, 1992, 120(8): 1 747-1 763 [2] FISHER M. Background error covariance modelling. Proceedings of ECMWF seminar on recent developments in data assimilation for atmosphere and ocean[C]. United Kingdom: ECMWF, 2003. [3] BELO-PEREIRA M, BERRE L. The use of an ensemble approach to study the background error covariances in a global NWP model[J]. Mon Wea Rev, 2006, 134(9): 2 466-2 489. [4] KLEIST D, PARRISH D F, DERBER J C, et al. Introduction of the GSI into the NCEP global data assimilation system[J]. Wea Forecasting, 2009, 24(6): 1 691-1 705. [5] WU W S. Background error for NCEP's GSI analysis in regional mode[C]. Proceedings of 4th WMO international symposium on assimilation of observation in meteorology and oceanography, Prague, Czech Republic, 2005. [6] 龚建东.资料同化中二维特征长度随模式分辨率变化的分析研究[J].大气科学, 2007, 31(3): 459-467. [7] 刘延安.高光谱红外辐射资料在区域模式中的直接同化及应用研究[D].上海: 华东师范大学, 2014. [8] 陈耀登, 夏雪, 闵锦忠, 等.背景误差协方差变量平衡特征及其对台风同化和预报的影响[J].热带气象学报, 2017, 33(3): 289-298. [9] KALANY E.大气模式、资料同化和可预报性[M].蒲朝霞, 杨福全, 邓北胜, 等, 译.北京: 气象出版社, 2005. [10] 陈耀登, 曾腊梅, HUANG X Y, 等.纬度依赖背景场误差协方差及其对台风同化和预报的影响研究[J].热带气象学报, 2014, 30(4): 654-662. [11] BANNISTER R N. A review of forecast error covariance statistics in atmospheric variational data assimilation. I: Characteristics and measurements of forecast error covariances[J]. Q J R Meteor Soc, 2008, 134(637): 1 951-1 970. [12] 龚俊强, 刘朝顺, 刘延安, 等.探空和飞机观测资料联合同化对台风"苏迪罗"(2015)数值模拟的影响研究[J].热带气象学报, 2019, 35 (1): 99-112. [13] RIZVI S RH, LIU Z Q, HUANG X Y. Generation of WRF-ARW Background Errors for GSI[Z]. 2009. [14] 赵幸.青藏高原和华东地区背景误差协方差特征及同化效果对比研究[D].南京: 南京信息工程大学, 2015 [15] MURPHY A H. Skill scores based on the mean square error and their relationships to the correlation coefficient[J]. Mon Wea Rev, 1988, 116 (12): 2 417-2 424. [16] 薛童, 管兆勇, 徐建军, 等. ATMS和CrIS卫星资料同化对青藏高原天气预报的影响[J].高原气象, 2017, 36(4): 912-929 [17] 张诚忠, 薛纪善, 冯业荣, 等.基于贝叶斯方案的雷达反射率反演水汽及其同化试验[J].热带气象学报, 2019, 35(2): 145-153. [18] 张诚忠, 薛纪善, 黄燕燕, 等.资料同化对2017年登陆广东沿海台风的短期降水与路径预报影响[J].热带气象学报, 2019, 35(5): 577-586.