Localization for ensemble DA: objective diagnostic and effcient application
Presentation given at the Workshop on Climate Prediction in the Atlantic-Arctic sector, 7th June 2019. Abstract: Localization is a key aspect of ensemble data assimilation. Its purpose is to reduce the sampling noise affecting covariances estimated with ensembles of limited size. It is necessary for...
| Main Author: | |
|---|---|
| Format: | Conference Object |
| Language: | English |
| Published: |
Zenodo
2019
|
| Subjects: | |
| Online Access: | https://dx.doi.org/10.5281/zenodo.3241192 https://zenodo.org/record/3241192 |
| Summary: | Presentation given at the Workshop on Climate Prediction in the Atlantic-Arctic sector, 7th June 2019. Abstract: Localization is a key aspect of ensemble data assimilation. Its purpose is to reduce the sampling noise affecting covariances estimated with ensembles of limited size. It is necessary for both variational methods (EnVar) and Ensemble Kalman Filter-related methods, but applied in different spaces for this two kinds of methods. The localization function can be diagnosed objectively by combining elements of the sample covariance estimation theory and the linear filtering theory, as shown in Menetrier et al. (2015 a,b). Recent developments of explicit convolution methods to apply localization functions have shown a promising potential for coupled DA systems. |
|---|