CaltechAUTHORS
  A Caltech Library Service

Physics-informed machine learning: case studies for weather and climate modelling

Kashinath, K. and Mustafa, M. and Albert, A. and Wu, J-L. and Jiang, C. and Esmaeilzadeh, S. and Azizzadenesheli, K. and Wang, R. and Chattopadhyay, A. and Singh, A. and Manepalli, A. and Chirila, D. and Yu, R. and Walters, R. and White, B. and Xiao, H. and Tchelepi, H. A. and Marcus, P. and Anandkumar, A. and Hassanzadeh, P. (2021) Physics-informed machine learning: case studies for weather and climate modelling. Philosophical Transactions A: Mathematical, Physical and Engineering Sciences, 379 (2194). Art. No. 20200093. ISSN 1364-503X. https://resolver.caltech.edu/CaltechAUTHORS:20210223-154127043

Full text is not posted in this repository. Consult Related URLs below.

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20210223-154127043

Abstract

Machine learning (ML) provides novel and powerful ways of accurately and efficiently recognizing complex patterns, emulating nonlinear dynamics, and predicting the spatio-temporal evolution of weather and climate processes. Off-the-shelf ML models, however, do not necessarily obey the fundamental governing laws of physical systems, nor do they generalize well to scenarios on which they have not been trained. We survey systematic approaches to incorporating physics and domain knowledge into ML models and distill these approaches into broad categories. Through 10 case studies, we show how these approaches have been used successfully for emulating, downscaling, and forecasting weather and climate processes. The accomplishments of these studies include greater physical consistency, reduced training time, improved data efficiency, and better generalization. Finally, we synthesize the lessons learned and identify scientific, diagnostic, computational, and resource challenges for developing truly robust and reliable physics-informed ML models for weather and climate processes.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1098/rsta.2020.0093DOIArticle
https://github.com/jinlong83/statistical-constrained-GANSRelated ItemData
https://github.com/maxjiang93/space_time_pdeRelated ItemData
https://github.com/Rose-STL-Lab/Turbulent-Flow-NetRelated ItemData
https://github.com/Rui1521/Equivariant-Neural-NetsRelated ItemData
https://github.com/ashesh6810/Deep-Spatial-TransformersRelated ItemData
https://resources.marine.copernicus.eu/?option=com_csw&view=details&product_id=GLOBAL_ANALYSIS_FORECAST_PHY_001_024Related ItemData
https://portal.edirepository.org/nis/mapbrowse?packageid=edi.200.6Related ItemData
https://doi.org/10.6073/pasta/8f19c5d19d816857e55077ba20570265DOIData
https://prism.oregonstate.eduRelated ItemData
https://github.com/arkadaw9/PGA_LSTMRelated ItemData
https://lter.limnology.wisc.edu/dataRelated ItemData
https://gitlab.com/mspritch/spcam3.0-neural-netRelated ItemData
https://doi.org/10.5281/zenodo.2559313DOIData
ORCID:
AuthorORCID
Kashinath, K.0000-0002-9311-5215
Esmaeilzadeh, S.0000-0001-6122-9122
Azizzadenesheli, K.0000-0001-8507-1868
Chirila, D.0000-0002-6394-4688
White, B.0000-0002-3739-9604
Tchelepi, H. A.0000-0002-3084-6635
Hassanzadeh, P.0000-0001-9425-8085
Additional Information:© 2021 The Author(s). Published by the Royal Society. Manuscript accepted 24/11/2020; Published online 15/02/2021; Published in print 05/04/2021. This article is part of the theme issue ‘Machine learning for weather and climate modelling’. Data accessibility: Data, code and supporting materials are publicly available via the following links: https://github.com/jinlong83/statistical-constrained-GANS; https://github.com/maxjiang93/space_time_pde; https://github.com/Rose-STL-Lab/Turbulent-Flow-Net; https://github.com/Rui1521/Equivariant-Neural-Nets; https://github.com/ashesh6810/Deep-Spatial-Transformers; https://resources.marine.copernicus.eu/?option=com_csw&view=details&product_id=GLOBAL_ANALYSIS_FORECAST_PHY_001_024; https://portal.edirepository.org/nis/mapbrowse?packageid=edi.200.6; https://doi.org/10.6073/pasta/8f19c5d19d816857e55077ba20570265; https://prism.oregonstate.edu/; https://github.com/arkadaw9/PGA_LSTM; https://lter.limnology.wisc.edu/data; https://gitlab.com/mspritch/spcam3.0-neural-net; https://doi.org/10.5281/zenodo.2559313. Authors' contributions: K.K. conceived the idea and designed the structure of the manuscript, wrote the manuscript, and responded to reviewer comments. K.K., M.M., and A.A. led the majority of the research reviewed as case studies in this article. The rest of the authors contributed to the research reviewed as case studies or provided feedback on sections of the manuscript. K.K. dedicates this work to A.A., a colleague and dear friend, who unfortunately was killed in a hit-and-run road accident while he was biking, during the course of preparation of this manuscript. We declare we have no competing interests. No funding has been received for this article.
Subject Keywords:neural networks, physical constraints, turbulent flows, physics-informed machine learning, weather and climate modeling
Issue or Number:2194
Record Number:CaltechAUTHORS:20210223-154127043
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20210223-154127043
Official Citation:Kashinath K et al. 2021 Physics-informed machine learning: case studies for weather and climate modelling. Phil. Trans. R. Soc. A 379: 20200093. https://doi.org/10.1098/rsta.2020.0093
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:108161
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:23 Feb 2021 23:56
Last Modified:23 Feb 2021 23:56

Repository Staff Only: item control page