Guidance on Localization for Strongly Coupled Atmosphere-Ocean Data Assimilation
Abstract
We address the question of how to implement vertical localization in a strongly coupled atmosphere-ocean system, where observations of the atmosphere can directly impact the ocean and vice-versa. Localization serves to mitigate the impact of sampling errors on ensemble-derived estimates of background error covariance matrices and thus a properly specified localization scheme is essential. Vertical localization methods have been developed for multiple earth system components separately. However, guidance on how to localize the cross-domain error covariances used in strongly coupled systems is needed. In this work we use forecasts from a global 1-degree coupled atmosphere-ocean model to inform a straightforward vertical localization strategy for strongly coupled atmosphere-ocean data assimilation systems. Our proposed state-dependent localization strategy is developed for R-matrix localization with several high-impact observing systems. We compare errors associated with this practical localization strategy to errors with several "optimal" localization schemes. We also present the pathway to including this localization strategy in a strongly coupled version of NOAA's Unified Forecast System.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMNG32A..04S