CaltechAUTHORS
  A Caltech Library Service

Deep learning to infer eddy heat fluxes from sea surface height patterns of mesoscale turbulence

George, Tom M. and Manucharyan, Georgy E. and Thompson, Andrew F. (2021) Deep learning to infer eddy heat fluxes from sea surface height patterns of mesoscale turbulence. Nature Communications, 12 . Art. No. 800. ISSN 2041-1723. PMCID PMC7865057. doi:10.1038/s41467-020-20779-9. https://resolver.caltech.edu/CaltechAUTHORS:20210205-160618946

[img]
Preview
PDF - Published Version
Creative Commons Attribution.

2MB
[img] PDF (Peer Review File) - Supplemental Material
Creative Commons Attribution.

6MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20210205-160618946

Abstract

Mesoscale eddies have strong signatures in sea surface height (SSH) anomalies that are measured globally through satellite altimetry. However, monitoring the transport of heat associated with these eddies and its impact on the global ocean circulation remains difficult as it requires simultaneous observations of upper-ocean velocity fields and interior temperature and density properties. Here we demonstrate that for quasigeostrophic baroclinic turbulence the eddy patterns in SSH snapshots alone contain sufficient information to estimate the eddy heat fluxes. We use simulations of baroclinic turbulence for the supervised learning of a deep Convolutional Neural Network (CNN) to predict up to 64% of eddy heat flux variance. CNNs also significantly outperform other conventional data-driven techniques. Our results suggest that deep CNNs could provide an effective pathway towards an operational monitoring of eddy heat fluxes using satellite altimetry and other remote sensing products.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1038/s41467-020-20779-9DOIArticle
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7865057PubMed CentralArticle
https://doi.org/10.6084/m9.figshare.11920905.v1Related ItemData; scripts; etc.
ORCID:
AuthorORCID
George, Tom M.0000-0002-4527-8810
Manucharyan, Georgy E.0000-0001-7959-2675
Thompson, Andrew F.0000-0003-0322-4811
Additional Information:© The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Received 29 September 2019. Accepted 14 December 2020. Published 05 February 2021. The authors gratefully acknowledge support from Charlie Trimble as well as the David and Lucile Packard Foundation. This work was partially completed during Caltech’s Summer Undergraduate Research Fellowship Program (SURF), and we thank the SURF staff for their assistance. Glenn Flierl provided the QG turbulence code used in this study. This study benefited from conversations with Frederick Eberhardt and RJ Antonello. Author Contributions. G.E.M. conceived the study; T.M.G. and G.E.M. performed the research; all authors analyzed the results and contributed to writing of the paper; the research was supervised by A.F.T. and G.E.M. Data availability. The necessary procedures to generate the data and reproduce the machine learning techniques have been outlined in the manuscript. The datasets and python scripts used in our study have been published in a Figshare repository74. We provided O(105) SSH snapshots of mesoscale turbulence and corresponding domain-averaged eddy heat fluxes as simulated by the two-layer QG model and split into training and validation data; the data and Python/TensorFlow scripts including neural network architectures graphs and hyperparameters that reproduce our training results can be downloaded here: https://doi.org/10.6084/m9.figshare.11920905.v1. If additional data is needed, the QG model that was used to generate the samples is available upon request from the authors. he authors declare no competing interests. Peer review information. Nature Communications thanks Alexa Griesel, Igor Kamenkovich, Ryan Abernathey, and other, anonymous, reviewers for their contributions to the peer review of this work. Peer review reports are available.
Funders:
Funding AgencyGrant Number
Charlie TrimbleUNSPECIFIED
David and Lucile Packard FoundationUNSPECIFIED
Caltech Summer Undergraduate Research Fellowship (SURF)UNSPECIFIED
PubMed Central ID:PMC7865057
DOI:10.1038/s41467-020-20779-9
Record Number:CaltechAUTHORS:20210205-160618946
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20210205-160618946
Official Citation:George, T.M., Manucharyan, G.E. & Thompson, A.F. Deep learning to infer eddy heat fluxes from sea surface height patterns of mesoscale turbulence. Nat Commun 12, 800 (2021). https://doi.org/10.1038/s41467-020-20779-9
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:107942
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:08 Feb 2021 15:15
Last Modified:16 Nov 2021 19:07

Repository Staff Only: item control page