Demonstration only. Corollary is a non-commercial research demo — no warranty, no support, not for commercial use or clinical decisions. Read the disclaimer

All models

Multi-Modal Foundation · allenai

SamudrACE-E3SMv3

A stochastic coupled atmosphere–ocean emulator for simulating long-term preindustrial climate variability.

What SamudrACE-E3SMv3 is

SamudrACE-E3SMv3 is a machine learning emulator developed by the Allen Institute for AI (Ai2). Belonging to the Ai2 Climate Emulator (ACE) family, it extends atmospheric emulation to coupled atmosphere–ocean dynamics, reproducing the behaviour of the Energy Exascale Earth System Model version 3 (E3SMv3).

The model addresses the computational expense of running high-resolution, multi-century Earth system simulations. It allows climate researchers to study coupled variability under fixed preindustrial control forcing conditions over timescales spanning from individual days to hundreds of years.

Trained on simulation outputs provided by the United States Department of Energy (DOE), the model was developed and evaluated using the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC).

How it works

SamudrACE-E3SMv3 operates as a stochastic emulator of the fully coupled E3SMv3 Earth system model. It is designed to capture coupled atmosphere–ocean interactions and internal climate variability rather than deterministic atmospheric trajectories alone.

The model was trained on simulation data generated by E3SM under preindustrial control forcing conditions using the fme machine learning framework.

Key facts

Model family
Ai2 Climate Emulator (ACE)
Base system
Energy Exascale Earth System Model version 3 (E3SMv3)
Framework
fme
Forcing scenario
Preindustrial control

What you can use it for

Long-Term Climate Variability Simulation
Simulating coupled atmosphere–ocean dynamics and internal variability across multi-day to centennial timescales.
Preindustrial Control Emulation
Generating extended baseline climate realizations under fixed preindustrial forcing without running full numerical simulations.
Stochastic Coupled Earth System Exploration
Studying stochastic behaviour and feedback mechanisms in fully coupled ocean-atmosphere systems.

Run it here

We list this model but cannot run it yet: no pipeline_tag — needs a human.

Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.

Strengths and limits

Good at

  • Emulates fully coupled atmosphere–ocean interactions across timescales from days to centuries.
  • Incorporates stochastic processes to represent unforced internal climate variability.
  • Trained directly on comprehensive Earth system model outputs from DOE E3SMv3.

Watch out for

  • Intended strictly as a research tool and cannot be applied to operational climate forecasting.
  • Limited to preindustrial control forcing conditions, meaning it does not simulate historical or future greenhouse gas warming trajectories.

Inputs and outputs

You provide

  • Sequence or text

    A protein or nucleotide sequence, or free text — whichever this model was trained on.

You get back

  • Mean-pooled representation, one vector per input.

Licence

Check the licence before commercial use · apache-2.0

The apache-2.0 licence carries terms worth reading before you depend on it.

Questions

Can I use SamudrACE-E3SMv3 commercially?

Check the licence before commercial use. The apache-2.0 licence carries terms worth reading before you depend on it.

Can this model be used for operational climate prediction?

No. The model is built as an exploratory research tool and should not be used for operational forecasting.

Which climate forcing scenario is represented?

The model simulates conditions strictly under preindustrial control forcing.

What is the relationship between ACE and SamudrACE-E3SMv3?

SamudrACE-E3SMv3 is a member of the Ai2 Climate Emulator (ACE) family, specifically extending the framework to coupled atmosphere–ocean emulation using E3SMv3 data.

What computing resources were used to develop the model?

The model was trained and evaluated on the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC).

How to cite it

Stochastic Emulation of a Fully Coupled Preindustrial E3SMv3 Simulation, https://doi.org/10.1029/2025GL119340

Written from SamudrACE-E3SMv3's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.