Eintrag weiter verarbeiten
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization
Gespeichert in:
Zeitschriftentitel: | Journal of Advances in Modeling Earth Systems |
---|---|
Personen und Körperschaften: | , |
In: | Journal of Advances in Modeling Earth Systems, 11, 2019, 1, S. 376-399 |
Format: | E-Article |
Sprache: | Englisch |
veröffentlicht: |
American Geophysical Union (AGU)
|
Schlagwörter: |
author_facet |
Bolton, Thomas Zanna, Laure Bolton, Thomas Zanna, Laure |
---|---|
author |
Bolton, Thomas Zanna, Laure |
spellingShingle |
Bolton, Thomas Zanna, Laure Journal of Advances in Modeling Earth Systems Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization General Earth and Planetary Sciences Environmental Chemistry Global and Planetary Change |
author_sort |
bolton, thomas |
spelling |
Bolton, Thomas Zanna, Laure 1942-2466 1942-2466 American Geophysical Union (AGU) General Earth and Planetary Sciences Environmental Chemistry Global and Planetary Change http://dx.doi.org/10.1029/2018ms001472 <jats:title>Abstract</jats:title><jats:p>Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high‐resolution quasi‐geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data‐driven approaches can be exploited to predict both subgrid and large‐scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in‐depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse‐resolution climate models.</jats:p> Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization Journal of Advances in Modeling Earth Systems |
doi_str_mv |
10.1029/2018ms001472 |
facet_avail |
Online Free |
finc_class_facet |
Technik Chemie und Pharmazie Geologie und Paläontologie Geographie Physik |
format |
ElectronicArticle |
fullrecord |
blob:ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuMTAyOS8yMDE4bXMwMDE0NzI |
id |
ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuMTAyOS8yMDE4bXMwMDE0NzI |
institution |
DE-Ch1 DE-L229 DE-D275 DE-Bn3 DE-Brt1 DE-D161 DE-Zwi2 DE-Gla1 DE-Zi4 DE-15 DE-Pl11 DE-Rs1 DE-105 DE-14 |
imprint |
American Geophysical Union (AGU), 2019 |
imprint_str_mv |
American Geophysical Union (AGU), 2019 |
issn |
1942-2466 |
issn_str_mv |
1942-2466 |
language |
English |
mega_collection |
American Geophysical Union (AGU) (CrossRef) |
match_str |
bolton2019applicationsofdeeplearningtooceandatainferenceandsubgridparameterization |
publishDateSort |
2019 |
publisher |
American Geophysical Union (AGU) |
recordtype |
ai |
record_format |
ai |
series |
Journal of Advances in Modeling Earth Systems |
source_id |
49 |
title |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_unstemmed |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_full |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_fullStr |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_full_unstemmed |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_short |
Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_sort |
applications of deep learning to ocean data inference and subgrid parameterization |
topic |
General Earth and Planetary Sciences Environmental Chemistry Global and Planetary Change |
url |
http://dx.doi.org/10.1029/2018ms001472 |
publishDate |
2019 |
physical |
376-399 |
description |
<jats:title>Abstract</jats:title><jats:p>Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high‐resolution quasi‐geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data‐driven approaches can be exploited to predict both subgrid and large‐scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in‐depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse‐resolution climate models.</jats:p> |
container_issue |
1 |
container_start_page |
376 |
container_title |
Journal of Advances in Modeling Earth Systems |
container_volume |
11 |
format_de105 |
Article, E-Article |
format_de14 |
Article, E-Article |
format_de15 |
Article, E-Article |
format_de520 |
Article, E-Article |
format_de540 |
Article, E-Article |
format_dech1 |
Article, E-Article |
format_ded117 |
Article, E-Article |
format_degla1 |
E-Article |
format_del152 |
Buch |
format_del189 |
Article, E-Article |
format_dezi4 |
Article |
format_dezwi2 |
Article, E-Article |
format_finc |
Article, E-Article |
format_nrw |
Article, E-Article |
_version_ |
1792347987033194503 |
geogr_code |
not assigned |
last_indexed |
2024-03-01T18:03:57.744Z |
geogr_code_person |
not assigned |
openURL |
url_ver=Z39.88-2004&ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fvufind.svn.sourceforge.net%3Agenerator&rft.title=Applications+of+Deep+Learning+to+Ocean+Data+Inference+and+Subgrid+Parameterization&rft.date=2019-01-01&genre=article&issn=1942-2466&volume=11&issue=1&spage=376&epage=399&pages=376-399&jtitle=Journal+of+Advances+in+Modeling+Earth+Systems&atitle=Applications+of+Deep+Learning+to+Ocean+Data+Inference+and+Subgrid+Parameterization&aulast=Zanna&aufirst=Laure&rft_id=info%3Adoi%2F10.1029%2F2018ms001472&rft.language%5B0%5D=eng |
SOLR | |
_version_ | 1792347987033194503 |
author | Bolton, Thomas, Zanna, Laure |
author_facet | Bolton, Thomas, Zanna, Laure, Bolton, Thomas, Zanna, Laure |
author_sort | bolton, thomas |
container_issue | 1 |
container_start_page | 376 |
container_title | Journal of Advances in Modeling Earth Systems |
container_volume | 11 |
description | <jats:title>Abstract</jats:title><jats:p>Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high‐resolution quasi‐geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data‐driven approaches can be exploited to predict both subgrid and large‐scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in‐depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse‐resolution climate models.</jats:p> |
doi_str_mv | 10.1029/2018ms001472 |
facet_avail | Online, Free |
finc_class_facet | Technik, Chemie und Pharmazie, Geologie und Paläontologie, Geographie, Physik |
format | ElectronicArticle |
format_de105 | Article, E-Article |
format_de14 | Article, E-Article |
format_de15 | Article, E-Article |
format_de520 | Article, E-Article |
format_de540 | Article, E-Article |
format_dech1 | Article, E-Article |
format_ded117 | Article, E-Article |
format_degla1 | E-Article |
format_del152 | Buch |
format_del189 | Article, E-Article |
format_dezi4 | Article |
format_dezwi2 | Article, E-Article |
format_finc | Article, E-Article |
format_nrw | Article, E-Article |
geogr_code | not assigned |
geogr_code_person | not assigned |
id | ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuMTAyOS8yMDE4bXMwMDE0NzI |
imprint | American Geophysical Union (AGU), 2019 |
imprint_str_mv | American Geophysical Union (AGU), 2019 |
institution | DE-Ch1, DE-L229, DE-D275, DE-Bn3, DE-Brt1, DE-D161, DE-Zwi2, DE-Gla1, DE-Zi4, DE-15, DE-Pl11, DE-Rs1, DE-105, DE-14 |
issn | 1942-2466 |
issn_str_mv | 1942-2466 |
language | English |
last_indexed | 2024-03-01T18:03:57.744Z |
match_str | bolton2019applicationsofdeeplearningtooceandatainferenceandsubgridparameterization |
mega_collection | American Geophysical Union (AGU) (CrossRef) |
physical | 376-399 |
publishDate | 2019 |
publishDateSort | 2019 |
publisher | American Geophysical Union (AGU) |
record_format | ai |
recordtype | ai |
series | Journal of Advances in Modeling Earth Systems |
source_id | 49 |
spelling | Bolton, Thomas Zanna, Laure 1942-2466 1942-2466 American Geophysical Union (AGU) General Earth and Planetary Sciences Environmental Chemistry Global and Planetary Change http://dx.doi.org/10.1029/2018ms001472 <jats:title>Abstract</jats:title><jats:p>Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high‐resolution quasi‐geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data‐driven approaches can be exploited to predict both subgrid and large‐scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in‐depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse‐resolution climate models.</jats:p> Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization Journal of Advances in Modeling Earth Systems |
spellingShingle | Bolton, Thomas, Zanna, Laure, Journal of Advances in Modeling Earth Systems, Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization, General Earth and Planetary Sciences, Environmental Chemistry, Global and Planetary Change |
title | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_full | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_fullStr | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_full_unstemmed | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_short | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
title_sort | applications of deep learning to ocean data inference and subgrid parameterization |
title_unstemmed | Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization |
topic | General Earth and Planetary Sciences, Environmental Chemistry, Global and Planetary Change |
url | http://dx.doi.org/10.1029/2018ms001472 |