Training a convolutional neural network to conserve mass in data assimilation
<p>In previous work, it was shown that the preservation of physical properties in the data assimilation framework can significantly reduce forecast errors. Proposed data assimilation methods, such as the quadratic programming ensemble (QPEns) that can impose such constraints on the calculation...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2021-02-01
|
| Series: | Nonlinear Processes in Geophysics |
| Online Access: | https://npg.copernicus.org/articles/28/111/2021/npg-28-111-2021.pdf |
| _version_ | 1818611182550908928 |
|---|---|
| author | Y. Ruckstuhl T. Janjić S. Rasp |
| author_facet | Y. Ruckstuhl T. Janjić S. Rasp |
| author_sort | Y. Ruckstuhl |
| collection | DOAJ |
| description | <p>In previous work, it was shown that the preservation of physical properties in the data assimilation framework can significantly reduce forecast errors. Proposed data assimilation methods, such as the quadratic programming ensemble (QPEns) that can impose such constraints on the calculation of the analysis, are computationally more expensive, severely limiting their application to high-dimensional prediction systems as found in Earth sciences. We, therefore, propose using a convolutional neural network (CNN) trained on the difference between the analysis produced by a standard ensemble Kalman filter (EnKF) and the QPEns to correct any violations of imposed constraints. In this paper, we focus on the conservation of mass and show that, in an idealised set-up, the hybrid of a CNN and the EnKF is capable of reducing analysis and background errors to the same level as the QPEns.</p> |
| first_indexed | 2024-12-16T15:26:16Z |
| format | Article |
| id | doaj.art-34ac20e37de84ad398ca2f4b5418843f |
| institution | Directory Open Access Journal |
| issn | 1023-5809 1607-7946 |
| language | English |
| last_indexed | 2024-12-16T15:26:16Z |
| publishDate | 2021-02-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | Nonlinear Processes in Geophysics |
| spelling | doaj.art-34ac20e37de84ad398ca2f4b5418843f2022-12-21T22:26:30ZengCopernicus PublicationsNonlinear Processes in Geophysics1023-58091607-79462021-02-012811111910.5194/npg-28-111-2021Training a convolutional neural network to conserve mass in data assimilationY. Ruckstuhl0T. Janjić1S. Rasp2Meteorological Institute Munich, Ludwig-Maximilians-Universität München, Munich, GermanyMeteorological Institute Munich, Ludwig-Maximilians-Universität München, Munich, GermanyClimateAi, San Francisco, CA, USA<p>In previous work, it was shown that the preservation of physical properties in the data assimilation framework can significantly reduce forecast errors. Proposed data assimilation methods, such as the quadratic programming ensemble (QPEns) that can impose such constraints on the calculation of the analysis, are computationally more expensive, severely limiting their application to high-dimensional prediction systems as found in Earth sciences. We, therefore, propose using a convolutional neural network (CNN) trained on the difference between the analysis produced by a standard ensemble Kalman filter (EnKF) and the QPEns to correct any violations of imposed constraints. In this paper, we focus on the conservation of mass and show that, in an idealised set-up, the hybrid of a CNN and the EnKF is capable of reducing analysis and background errors to the same level as the QPEns.</p>https://npg.copernicus.org/articles/28/111/2021/npg-28-111-2021.pdf |
| spellingShingle | Y. Ruckstuhl T. Janjić S. Rasp Training a convolutional neural network to conserve mass in data assimilation Nonlinear Processes in Geophysics |
| title | Training a convolutional neural network to conserve mass in data assimilation |
| title_full | Training a convolutional neural network to conserve mass in data assimilation |
| title_fullStr | Training a convolutional neural network to conserve mass in data assimilation |
| title_full_unstemmed | Training a convolutional neural network to conserve mass in data assimilation |
| title_short | Training a convolutional neural network to conserve mass in data assimilation |
| title_sort | training a convolutional neural network to conserve mass in data assimilation |
| url | https://npg.copernicus.org/articles/28/111/2021/npg-28-111-2021.pdf |
| work_keys_str_mv | AT yruckstuhl trainingaconvolutionalneuralnetworktoconservemassindataassimilation AT tjanjic trainingaconvolutionalneuralnetworktoconservemassindataassimilation AT srasp trainingaconvolutionalneuralnetworktoconservemassindataassimilation |