Variable Masking for Climate Emulation in ACE

By Yermakov, Arcomano, McGibbon +2Oct 14, 2026

Score████░░░░░░4.4

VerdictWorth a reader's time today.

Abstract

In the recent AI Model Intercomparison Project (AIMIP) benchmark, the ACE2.1-ERA5 emulator was shown to underestimate the true near-surface warming trend both during the training period and the out-of-sample extrapolation period. In this work, we present M-ACE2S (Masked-ACE2S), a modified version of the stochastic ACE2S model, which uses variable masking during training and reduces the reliance of M-ACE2S on any single input channel. We show that M-ACE2S improves long-term near-surface air temperature prediction. We also present a novel data normalization strategy called "global mean removal", which, combined with masking, further improves the near-surface temperature forecasts. We show that M-ACE2S outperforms the ACE2S baseline on in-sample data for ERA5 while improving training stability and generalizability of near-surface temperature to unseen forcing when paired with global mean removal.

Alexey Yermakov, Troy Arcomano, Jeremy McGibbon, Brian Henn, Oliver Watt-Meyer

The editor's rubric

Heuristic review

Dimension	Level	Weight	What that level means
Leverage	███░░ 3	16%	A method or resource many groups across the field will adopt within a year.
Magnitude	███░░ 3	20%	Large gain: roughly 2x, or a clear new state of the art on a hard, unsaturated problem.
Evidence	███░░ 3	20%	Solid: multiple benchmarks or cohorts, ablations, fair baselines, released code or data.
Novelty	██░░░ 2	10%	A new combination of known ideas.
Trajectory	██░░░ 2	14%	Some room to improve with obvious engineering.
Stakes	███░░ 3	20%	Meaningful benefit to many people within a few years.

Editor’s rationale

Heuristic triage from title and abstract text only, not a reading of the paper. Cues found: method (we propose, we report, named contribution); gains (outperforms); verification (held-out test, code released); stakes (global scale, climate). Red flags: derivative (modified version).

How the score was computed

rank-2026-10-07

Score████░░░░░░4.4

Score = 10 × (75% × adjusted merit / 10 + 15% × attention + 10% × freshness)

Merit: 5.5 / 10; Weighted rubric, evidence-gated.
Adjusted merit: 4.7 / 10; Shrunk toward the desk prior by editor confidence (44%).
Attention: 0%; Citations, upvotes, points, mentions.
Freshness: 91%; Half-life decay since publication.

Citations0 (reference 15, via semantic-scholar, Oct 16, 2026, 07:29 UTC)

The record

Reviewed by heuristic-v6 on Oct 16, 2026, 07:29 UTC. Paper type: method.
Categories: physics.ao-ph
TOP, No.4 in the Climate & Energy edition of October 16, 2026.