| 全球资料同化中误差协方差三维结构的准确估计与应用Ⅱ:背景误差协方差调整与数值试验分析 |
| |
| 引用本文： | 龚建东,赵刚.全球资料同化中误差协方差三维结构的准确估计与应用Ⅱ:背景误差协方差调整与数值试验分析[J].Acta Meteorologica Sinica,2006,64(6):684-698. |
| |
| 作者姓名： | 龚建东 赵刚 |
| |
| 作者单位： | 国家气象中心,北京,100081 |
| |
| 基金项目： | 国家自然科学基金(40305015)，国家气象中心自筹资金课题“三维变分同化阶段性突出问题研究”(04ZK2003A-09) |
| |
| 摘 要： | 利用NMC方法针对背景误差协方差的方差、三维相关与特征长度来揭示T213L31模式的误差主要特点,并与传统更新矢量方法的计算结果进行了对比与调整。结果表明NMC方法结果与更新矢量方法结果在大体特征上基本吻合,但细节上的差异不可忽视,特别是对背景误差方差与特征长度的估计存在显著的差异,其主要原因是NMC方法倾向于高估天气尺度波的背景误差,而低估次天气尺度到中尺度波的背景误差。通过对背景误差方差、特征长度的调整,显著改善了背景误差功率谱的分布特点,使得NMC方法结果与更新矢量方法结果更为吻合。通过三维变分同化与最优插值中观测与背景误差相对重要性的比较,发现两者结果基本一致,但三维变分同化在850 hPa以下的温度场和300 hPa以上的风场统计结果都表现出背景误差相对于观测误差偏小的特点。背景误差相对于观测误差偏小有助于保证分析场中质量场与风场平衡,消除了大气底层和高层质量场与风场不匹配现象。在数值试验中,针对不同的背景误差均方差与特征长度的特点,分析了分析增量和预报效果的差异,结果表明,准确的背景误差估计与优化工作改善了预报效果,使得北半球三维变分同化的120 h预报效果整体好于现有最优插值。
|
| 关 键 词： | 背景误差协方差 NMC方法估计 更新矢量结果调整 数值对比试验。 |
| 收稿时间： | 2005/9/27 0:00:00 |
| 修稿时间： | 2005年9月27日 |
ACCURATE ESTIMATION AND APPLICATION OF 3-D ERROR COVARIANCE STRUCTURES IN GLOBAL DATA ASSIMILATION Part Ⅱ: Background Error Covariance Structure Adjustments and Numerical Experiments |
| |
| Gong Jiandong,Zhao Gang.ACCURATE ESTIMATION AND APPLICATION OF 3-D ERROR COVARIANCE STRUCTURES IN GLOBAL DATA ASSIMILATION Part Ⅱ: Background Error Covariance Structure Adjustments and Numerical Experiments[J].Acta Meteorologica Sinica,2006,64(6):684-698. |
| |
| Authors: | Gong Jiandong Zhao Gang |
| |
| Abstract: | The so-called NMC-method is used to analyze the three major characteristics of the T213L31 model outputs: error variance,three-dimensional correlation and de-correlation length.Results from the NMC-method are compared with those from the traditional innovation vector method,and then its parameters are tuned based on the comparison.It is shown that results from both methods are roughly similar,but some differences still exist in the background error variance and de-correlation length estimations.It seems that the NMC-method intends to over-estimate background errors for synoptic scale,but under-estimate them for sub-synoptic scale and meso-scale.After the tuning of background error variance and de-correlation length,results from the NMC-method are closer to those of the innovation vector method,and the analysis results are improved.Comparing the ratios of the background to observation errors for the currently operational optimal interpolation system(OI) with the three-dimensional variational data assimilation(3DVAR),it is shown that the two system results are roughly similar, but the background errors for temperatures below 850 hPa and for winds above 300 hPa in the 3D-VAR assimilation are relatively smaller than their observation errors.This is helpful to make sure the mass and wind fields in balance within the analysis,to reduce the unbalance part of the two fields near surface and upper level in the OI system.This paper shows that the accurate estimations and tunings of background error covariance improved the forecast skill,and in the north hemisphere,the forecast for 120 hours made from the analysis from the three dimensional variational data assimilation system is better than that using the currently operational OI system. |
| |
| Keywords: | Background error covariance NMC-method estimation Tuning with innovation vector Numerical contrast experiments |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《Acta Meteorologica Sinica》浏览原始摘要信息 |
| 点击此处可从《Acta Meteorologica Sinica》下载免费的PDF全文 |
|
