Multigrid beta filter for faster computation of ensemble covariance localization

Yokota, Sho; Rancic, Miodrag; Lei, Ting; Purser, R. James; De Pondeca, Manuel S. F. V.

doi:https://doi.org/10.5194/gmd-18-7815-2025

Articles | Volume 18, issue 20

https://doi.org/10.5194/gmd-18-7815-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/gmd-18-7815-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 18, issue 20

Development and technical paper

|

27 Oct 2025

Development and technical paper |

| 27 Oct 2025

Multigrid beta filter for faster computation of ensemble covariance localization

Sho Yokota, Miodrag Rancic, Ting Lei, R. James Purser, and Manuel S. F. V. De Pondeca

Data sets

ICs, LBCs, and observations for EnVar NOAA National Centers for Environmental Prediction https://doi.org/10.5281/zenodo.15744386

ICs for EnKF (mem001-015) NOAA National Centers for Environmental Prediction https://doi.org/10.5281/zenodo.15747450

ICs for EnKF (mem016-030) NOAA National Centers for Environmental Prediction https://doi.org/10.5281/zenodo.15747477

Model code and software

Rapid Refresh Forecast System (RRFS) Sho Yokota https://doi.org/10.5281/zenodo.15193112

Short summary

Covariance localization to mitigate sampling error of ensemble-based forecast error covariances is one of the main parts of the calculation in ensemble-variational data assimilation for the atmosphere. This study clarifies that the multigrid beta filter-based localization makes it several times faster than the conventional recursive filter-based one without significantly changing the analysis if a coarser filter grid is applied and filters except for the coarsest resolution are omitted.