摘要
优化和改进变分资料同化系统中的背景误差协方差模型,使之能够正确反映随流型演变的不确定性信息,是提高中期数值预报精度的重要技术手段。集合四维变分资料同化的一大关键是如何根据自身的背景误差协方差模型提取和应用流依赖背景信息。为了提高变分资料同化解算的效率,通过将某些变量划分为平衡部分和非平衡部分,平衡部分依据平衡约束关系与一个特定的变量相联系,剩余的非平衡部分在各变量间互不相关。随着数值预报模式的不断优化,非平衡方差在总方差中的影响越来越重要。本文介绍了YH4DVAR的背景误差协方差模型和集合四维变分资料同化系统架构,重点分析了平衡算子。通过集合方法估计得到了散度、温度、地面气压的非平衡项流依赖方差。最后,为了减少有限样本噪声对方差估计的影响,对非平衡项方差进行了校正和滤波。
Optimizing and improving the background error covariance model, which promote the better representation on flow-dependent uncertain information, are important technology to improve the precision of medium-range numerical forecast system. A key issue of ensemble four-dimensional variational data assimilation is the way to estimate and apply flow-dependent background information according to its own background error covariance model. In general, some variables can be divided into balanced and unbalanced parts. The balanced parts are associated with specific variables under the constraint of balanced relationship. The residual unbalanced parts are independent. With the model of numerical weather prediction improving continuously, unbalanced variances play more and more important roles in the total variance. In this paper, we briefly introduced the background error covariance model of YH4DVAR and Ensemble Four-dimensional Variational Data Assimilation system, and the balanced operators are analyzed specifically. The flow-dependent unbalanced variances are estimated successfully for divergence, temperature and surface pressure by an ensemble method. Finally, some calibration and filtering technologies are imposed on unbalanced variances to reduce the sampling noises.
出处
《地球科学前沿（汉斯）》
2020年第7期637-647,共11页
Advances in Geosciences
关键词
资料同化
背景误差协方差
非平衡项方差
集合四维变分资料同化
流依赖
校正
滤波
Data Assimilation
Background Error Covariance
Unbalanced Variances
Ensemble Four-Dimensional Variational Data Assimilation
Flow-Dependent
Calibration
Filtering
