13 Oct 2023
 | 13 Oct 2023
Status: this preprint is open for discussion.

TorchClim v1.0: A deep-learning framework for climate model physics

David Fuchs, Steven C. Sherwood, Abhnil Prasad, Kirill Trapeznikov, and Jim Gimlett

Abstract. Climate models are hindered by the need to conceptualize and then parameterize complex physical processes that are not explicitly numerically resolved and for which no rigorous theory exists. Machine learning and artificial intelligence methods (ML/AI) offer a promising paradigm that can augment or replace the traditional parametrized approach with models trained on empirical process data. We offer a flexible and efficient framework, TorchClim, for inserting ML/AI physics surrogates that respect the parallelization of the climate model. A reference implementation of this approach is presented for the Community Earth System Model (CESM), where the authors substitute moist physics and radiative parametrization of the Community Atmospheric Model (CAM) with an ML/AI model. We show that a deep neural network surrogate trained on data from CAM itself can produce a stable model that reproduces the climate and variability of the original model, albeit with some biases. This framework is offered to the research community as an open-source project. The new framework seamlessly integrates into CAM's workflow and code-base and runs with negligible added computational cost, allowing rapid testing of various ML physics surrogates. The efficiency and flexibility of this framework open up new possibilities for using physics surrogates trained on offline data to improve climate model performance and better understand model physical processes.

David Fuchs et al.

Status: open (until 08 Dec 2023)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on egusphere-2023-1954', Anonymous Referee #1, 14 Nov 2023 reply

David Fuchs et al.

Model code and software

Github repository David Fuchs

David Fuchs et al.


Total article views: 573 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
394 169 10 573 7 5
  • HTML: 394
  • PDF: 169
  • XML: 10
  • Total: 573
  • BibTeX: 7
  • EndNote: 5
Views and downloads (calculated since 13 Oct 2023)
Cumulative views and downloads (calculated since 13 Oct 2023)

Viewed (geographical distribution)

Total article views: 564 (including HTML, PDF, and XML) Thereof 564 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 04 Dec 2023
Short summary
Machine learning (ML) of unresolved processes offers many new possibilities for improving weather and climate models, but integrating ML into the models has been an engineering challenge, and there are performance issues. We present a new software framework for this integration, TorchClim, that is scalable, fast, and flexible, and thereby allows a new level of experimentation with the ML approach. We also provide guidance on ML training and demonstrate a skillful hybrid-ML atmosphere model.