Hierarchical Testing of a Hybrid Machine Learning‐Physics Global Atmosphere Model
Open Access
- 4 March 2026
- journal article
- Published by American Geophysical Union (AGU) in AGU Advances
- Vol. 7 (2)
Abstract
Machine learning (ML)‐based models have demonstrated high skill and computational efficiency, often outperforming conventional physics‐based models in weather and subseasonal predictions. While prior studies have assessed their fidelity in capturing synoptic‐scale atmospheric dynamics, their performance across timescales and under out‐of‐distribution forcing, such as +3K or +4K uniform‐warming forcings, and the sources of biases remain elusive, to establish the model's reliability for Earth science. Here, we design three sets of experiments targeting synoptic‐scale phenomena, interannual variability, and out‐of‐distribution uniform‐warming forcings. We evaluate the Neural General Circulation Model (NeuralGCM), a hybrid model integrating a dynamical core with ML‐based component, against observations and physics‐based Earth system models (ESMs). At the synoptic scale, NeuralGCM captures the evolution and propagation of extratropical cyclones with performance comparable to ESMs. At the interannual scale, when forced by El Niño‐Southern Oscillation sea surface temperature (SST) anomalies, NeuralGCM successfully reproduces associated teleconnection patterns but exhibits deficiencies in capturing nonlinear response. Under out‐of‐distribution uniform‐warming forcings, NeuralGCM simulates similar responses in global‐average temperature and precipitation and reproduces large‐scale tropospheric circulation features similar to those in ESMs. Notable weaknesses include overestimating the tracks and spatial extent of extratropical cyclones, biases in the teleconnected wave train triggered by tropical SST anomalies, and differences in upper‐level warming and stratospheric circulation responses to SST warming compared to physics‐based ESMs. The causes of these weaknesses were explored. Despite the noted weaknesses, NeuralGCM reproduces responses across experiments reasonably and performs comparably to ESMs. By integrating a dynamical core with ML, NeuralGCM shows potential for developing ML‐based ESMs.Keywords
All Related Versions
This publication has 41 references indexed in Scilit:
- Concurrent Inter‐Model Spread of Boreal Winter Westerly Jet Meridional Positions Between the Northern and Southern Hemispheres in CMIP6 ModelsInternational Journal of Climatology, 2024
- Influence of a local diabatic heating source on the midlatitude circulationQuarterly Journal of the Royal Meteorological Society, 2024
- Relation of the double‐ITCZ bias to the atmospheric energy budget in climate modelsGeophysical Research Letters, 2016
- Climate model biases in jet streams, blocking and storm tracks resulting from missing orographic dragGeophysical Research Letters, 2016
- Isolating the roles of different forcing agents in global stratospheric temperature changes using model integrations with incrementally added single forcingsJournal of Geophysical Research: Atmospheres, 2016
- Attributing the forced components of observed stratospheric temperature variability to external driversQuarterly Journal of the Royal Meteorological Society, 2015
- The Brewer‐Dobson circulationReviews of Geophysics, 2014
- What is responsible for the strong observed asymmetry in teleconnections between El Niño and La Niña?Geophysical Research Letters, 2014
- Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth centuryJournal of Geophysical Research: Atmospheres, 2003
- Some simple solutions for heat‐induced tropical circulationQuarterly Journal of the Royal Meteorological Society, 1980