Hybrid physics-ML & parameterization★ Highlight
Crowdsourcing the Frontier: Advancing Hybrid Physics-ML Climate Simulation via a $50,000 Kaggle Competition
Why it matters. Shows multiple Kaggle-inspired ClimSim emulators run stably online with full microphysics in a coupled model; key milestone for hybrid ML climate simulation.
In one sentence. Kaggle-inspired ML subgrid emulators stay online-stable when coupled to a climate model with cloud microphysics, and some match state-of-the-art zonal-mean bias and global RMSE.
Abstract
Subgrid machine-learning (ML) parameterizations have the potential to introduce a new generation of climate models that incorporate the effects of higher-resolution physics without incurring the prohibitive computational cost associated with more explicit physics-based simulations. However, important issues, ranging from online instability to inconsistent online performance, have limited their operational use for long-term climate projections. To more rapidly drive progress in solving these issues, domain scientists and machine learning researchers opened up the offline aspect of this problem to the broader machine learning and data science community with the release of ClimSim, a NeurIPS Datasets and Benchmarks publication, and an associated Kaggle competition. This paper reports on the downstream results of the Kaggle competition by coupling emulators inspired by the winning teams' architectures to an interactive climate model (including full cloud microphysics, a regime historically prone to online instability) and systematically evaluating their online performance. Our results demonstrate that online stability in the low-resolution, real-geography setting is reproducible across multiple diverse architectures, which we consider a key milestone. All tested architectures exhibit strikingly similar offline and online biases, though their responses to architecture-agnostic design choices (e.g., expanding the list of input variables) can differ significantly. Multiple Kaggle-inspired architectures achieve state-of-the-art (SOTA) results on certain metrics such as zonal mean bias patterns and global RMSE, indicating that crowdsourcing the essence of the offline problem is one path to improving online performance in hybrid physics-AI climate simulation.
Labels
- Topic
- Hybrid physics-ML & parameterization, Benchmarks, datasets & evaluation
- Type
- Evaluation or intercomparison of existing models
- Method
- Other
- Time scale
- Climate projections / centennial
- Domain
- Global
- Data
- Other
- Status
- Research
- Identifiers
- arXiv:2511.20963 DOI:10.1029/2025ms005643 OpenAlex:W4417011631
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.