Ensemble Kalman filter updates based on regularized sparse inverse Cholesky factors

Ensemble Kalman filter updates based on regularized sparse inverse Cholesky factors
复制标题

基于正则化稀疏逆 Cholesky 因子的集成卡尔曼滤波器更新

DOI:
10.1175/mwr-d-20-0299.1
复制
发表时间:
2021
影响因子:
3.2
通讯作者:
Katzfuss, Matthias
Katzfuss, Matthias
中科院分区:
地球科学2区
文献类型:
--
作者:
Boyles, Will;Katzfuss, Matthias

文献摘要

参考文献

被引文献

相似文献

集成卡尔曼滤波器 (EnKF) 是高维非线性状态空间模型中数据同化的流行技术。 EnKF 表示集合的兴趣分布,这是一种降维形式,即使对于复杂且昂贵的演化算子也能进行直接预测。然而,EnKF 更新步骤涉及基于(通常很小)集成的预测协方差矩阵的估计,这需要正则化。许多现有的正则化技术依赖于空间定位,这可能会忽略长程依赖性。相反,我们提出的方法假设逆协方差矩阵的稀疏 Cholesky 因子,并且非零 Cholesky 条目进一步正则化。由此产生的方法具有高度灵活性和计算可扩展性。在我们的数值实验中,我们的方法比基于锥形的定位更准确,并且对调整参数的错误指定更不敏感。
The ensemble Kalman filter (EnKF) is a popular technique for data assimilation in high-dimensional nonlinear state-space models. The EnKF represents distributions of interest by an ensemble, which is a form of dimension reduction that enables straightforward forecasting even for complicated and expensive evolution operators. However, the EnKF update step involves estimation of the forecast covariance matrix based on the (often small) ensemble, which requires regularization. Many existing regularization techniques rely on spatial localization, which may ignore long-range dependence. Instead, our proposed approach assumes a sparse Cholesky factor of the inverse covariance matrix, and the nonzero Cholesky entries are further regularized. The resulting method is highly flexible and computationally scalable. In our numerical experiments, our approach was more accurate and less sensitive to misspecification of tuning parameters than tapering-based localization.
大空间数据的贝叶斯非平稳和非参数协方差估计
DOI: 10.1214/21-ba1273
发表时间: 2021
期刊: Bayesian Analysis
影响因子: 4.4
作者:
Kidd, Brian;Katzfuss, Matthias
通讯作者: Katzfuss, Matthias
DOI: 10.1007/s11222-021-10077-9
发表时间: 2022
影响因子: 2.2
作者:
Jurek, Marcin;Katzfuss, Matthias
通讯作者: Katzfuss, Matthias
DOI: 10.1080/01621459.2019.1592753
发表时间: 2017-04
影响因子: 3.7
作者:
M. Katzfuss;Jonathan R. Stroud;C. Wikle
通讯作者: M. Katzfuss;Jonathan R. Stroud;C. Wikle
DOI: 10.1214/19-sts755
发表时间: 2021-02-01
影响因子: 5.7
作者:
Katzfuss, Matthias;Guinness, Joseph
通讯作者: Guinness, Joseph