Hybrid physics-ML & parameterization
Robustness of Neural Network Emulations of Radiative Transfer Parameterizations in a State-of-the-Art General Circulation Model
In one sentence. Shows shallow-network emulators of radiative transfer, built for an older GCM, stay stable and realistic when transplanted into a newer state-of-the-art GCM under AMIP-like runs.
Abstract
The ability of Machine-Learning (ML) based model components to generalize to the previously unseen inputs, and the resulting stability of the models that use these components, has been receiving a lot of recent attention, especially when it comes to ML-based parameterizations. At the same time, ML-based emulators of existing parameterizations can be stable, accurate, and fast when used in the model they were specifically designed for. In this work we show that shallow-neural-network-based emulators of radiative transfer parameterizations developed almost a decade ago for a state-of-the-art GCM are robust with respect to the substantial structural and parametric change in the host model: when used in the AMIP-like experiment with the new model, they not only remain stable, but generate realistic output. Aspects of neural network architecture and training set design potentially contributing to stability of ML-based model components are discussed.
Labels
- Topic
- Hybrid physics-ML & parameterization, Explainability, uncertainty & physics
- Type
- Evaluation or intercomparison of existing models
- Method
- Other
- Time scale
- Climate projections / centennial
- Domain
- Global
- Data
- CMIP / climate model output, Other
- Status
- Research
- Identifiers
- arXiv:2103.07024 DOI:10.5194/gmd-14-7425-2021 OpenAlex:W3137215208
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.