Numerical Simulations with Data Assimilation Using an Adaptive POD Procedure
Numerical Simulations with Data Assimilation Using an Adaptive POD Procedure
复制标题
DOI：
10.1007/978-3-642-12535-5_18
复制
发表时间：
2009-06
期刊：
影响因子：
--
通讯作者：
G. Dimitriu;N. Apreutesei;R. Stefanescu
中科院分区：
文献类型：
--
作者：
G. Dimitriu;N. Apreutesei;R. Stefanescu
在这项研究中，应用模型降阶的本征正交分解(POD)方法来构造简单的对流扩散方程的降阶控制空间。在简化的控制空间中进行了与这些模式相关的几个4D-Var资料同化试验。重点对自适应POD程序的性能进行了评估，以经典的4D-Var(全模式)和POD 4D-Var资料同化得到的解为例。尽管模型动力学中存在一些扰动因素，但与其他方法相比，自适应POD格式表现出了更好的数值稳健性，并提供了准确的结果。
In this study the proper orthogonal decomposition (POD) methodology to model reduction is applied to construct a reduced-order control space for simple advection-diffusion equations. Several 4D-Var data assimilation experiments associated with these models are carried out in the reduced control space. Emphasis is laid on the performance evaluation of an adaptive POD procedure, with respect to the solution obtained with the classical 4D-Var (full model), and POD 4D-Var data assimilation. Despite some perturbation factors characterizing the model dynamics, the adaptive POD scheme presents better numerical robustness compared to the other methods, and provides accurate results.