Preprints
https://doi.org/10.48550/arXiv.2408.01581
https://doi.org/10.48550/arXiv.2408.01581
02 Oct 2024
 | 02 Oct 2024

Huge Ensembles Part II: Properties of a Huge Ensemble of Hindcasts Generated with Spherical Fourier Neural Operators

Ankur Mahesh, William Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard

Abstract. In Part I, we created an ensemble based on Spherical Fourier Neural Operators. As initial condition perturbations, we used bred vectors, and as model perturbations, we used multiple checkpoints trained independently from scratch. Based on diagnostics that assess the ensemble's physical fidelity, our ensemble has comparable performance to operational weather forecasting systems. However, it requires several orders of magnitude fewer computational resources. Here in Part II, we generate a huge ensemble (HENS), with 7,424 members initialized each day of summer 2023. We enumerate the technical requirements for running huge ensembles at this scale. HENS precisely samples the tails of the forecast distribution and presents a detailed sampling of internal variability. For extreme climate statistics, HENS samples events 4σ away from the ensemble mean. At each grid cell, HENS improves the skill of the most accurate ensemble member and enhances coverage of possible future trajectories. As a weather forecasting model, HENS issues extreme weather forecasts with better uncertainty quantification. It also reduces the probability of outlier events, in which the verification value lies outside the ensemble forecast distribution.

Share

Journal article(s) based on this preprint

04 Sep 2025
Huge ensembles – Part 2: Properties of a huge ensemble of hindcasts generated with spherical Fourier neural operators
Ankur Mahesh, William D. Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard
Geosci. Model Dev., 18, 5605–5633, https://doi.org/10.5194/gmd-18-5605-2025,https://doi.org/10.5194/gmd-18-5605-2025, 2025
Short summary
Ankur Mahesh, William Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard

Interactive discussion

Status: closed

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CEC1: 'Comment on egusphere-2024-2422', Juan Antonio Añel, 29 Oct 2024
  • RC1: 'Comment on egusphere-2024-2422', Peter Düben, 02 Nov 2024
  • RC2: 'Comment on egusphere-2024-2422', Anonymous Referee #2, 18 Jan 2025
  • AC1: 'Response to CEC1 Comment', Ankur Mahesh, 19 Jan 2025
  • AC2: 'Comment on egusphere-2024-2422', Ankur Mahesh, 18 Feb 2025
  • AC3: 'Changes to uploaded manuscript in response to reviewer comments', Ankur Mahesh, 03 Apr 2025

Interactive discussion

Status: closed

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CEC1: 'Comment on egusphere-2024-2422', Juan Antonio Añel, 29 Oct 2024
  • RC1: 'Comment on egusphere-2024-2422', Peter Düben, 02 Nov 2024
  • RC2: 'Comment on egusphere-2024-2422', Anonymous Referee #2, 18 Jan 2025
  • AC1: 'Response to CEC1 Comment', Ankur Mahesh, 19 Jan 2025
  • AC2: 'Comment on egusphere-2024-2422', Ankur Mahesh, 18 Feb 2025
  • AC3: 'Changes to uploaded manuscript in response to reviewer comments', Ankur Mahesh, 03 Apr 2025

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload
AR by Ankur Mahesh on behalf of the Authors (03 Apr 2025)  Author's response   Author's tracked changes   Manuscript 
ED: Referee Nomination & Report Request started (08 Apr 2025) by Po-Lun Ma
RR by Peter Düben (25 Apr 2025)
RR by Anonymous Referee #2 (27 Apr 2025)
ED: Publish as is (06 May 2025) by Po-Lun Ma
AR by Ankur Mahesh on behalf of the Authors (01 Jun 2025)  Manuscript 

Journal article(s) based on this preprint

04 Sep 2025
Huge ensembles – Part 2: Properties of a huge ensemble of hindcasts generated with spherical Fourier neural operators
Ankur Mahesh, William D. Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard
Geosci. Model Dev., 18, 5605–5633, https://doi.org/10.5194/gmd-18-5605-2025,https://doi.org/10.5194/gmd-18-5605-2025, 2025
Short summary
Ankur Mahesh, William Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard

Data sets

Trained Machine Learning Model Weights Ankur Mahesh et al. https://portal.nersc.gov/cfs/m4416/earth2mip_prod_registry/

Model code and software

Ensemble Inference Code Ankur Mahesh et al. https://github.com/ankurmahesh/earth2mip-fork/tree/HENS

Ankur Mahesh, William Collins, Boris Bonev, Noah Brenowitz, Yair Cohen, Peter Harrington, Karthik Kashinath, Thorsten Kurth, Joshua North, Travis A. O'Brien, Michael Pritchard, David Pruitt, Mark Risser, Shashank Subramanian, and Jared Willard

Viewed

Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.

Total article views: 487 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
481 0 6 487 0 0
  • HTML: 481
  • PDF: 0
  • XML: 6
  • Total: 487
  • BibTeX: 0
  • EndNote: 0
Views and downloads (calculated since 02 Oct 2024)
Cumulative views and downloads (calculated since 02 Oct 2024)

Viewed (geographical distribution)

Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.

Total article views: 473 (including HTML, PDF, and XML) Thereof 473 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 04 Sep 2025
Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Short summary
We use machine learning to create a massive database of simulated weather extremes. This database provides a large sample size, which is essential to characterize the statistics of extreme weather events and study their physical mechanisms. Also, such large simulations can be beneficial to accurately forecast the probability of low-likelihood extreme weather.
Share