Stable Machine-Learning Parameterization of Subgrid Processes in a Comprehensive Atmospheric Model Learned From Embedded Convection-Permitting Simulations

TL;DR

This paper develops a stable machine learning-based parameterization for subgrid processes in climate models, enabling long-term, high-complexity simulations with improved accuracy and reduced computational costs.

Contribution

It introduces a stable hybrid model with near operational complexity, incorporating explicit cloud microphysics and land coupling, achieving multi-year stability and realistic climatology.

Findings

Achieved 5-year stable simulations with low temperature and water vapor biases.

Demonstrated realistic cloud condensate climatology within the MMF framework.

Achieved skillful online performance with an expressive U-Net architecture and physical constraints.

Abstract

Modern climate projections often suffer from inadequate spatial and temporal resolution due to computational limitations, resulting in inaccurate representations of sub-grid processes. A promising technique to address this is the Multiscale Modeling Framework (MMF), which embeds a kilometer-resolution cloud-resolving model within each atmospheric column of a host climate model to replace traditional convection and cloud parameterizations. Machine learning (ML) offers a unique opportunity to make MMF more accessible by emulating the embedded cloud-resolving model and reducing its substantial computational cost. Although many studies have demonstrated proof-of-concept success of achieving stable hybrid simulations, it remains a challenge to achieve near operational-level success with real geography and comprehensive variable emulation that includes, for example, explicit cloud condensate coupling. In this study, we present a stable hybrid model capable of integrating for at least 5 years with near operational-level complexity, including coarse-grid geography, seasonality, explicit cloud condensate and wind predictions, and land coupling. Our model demonstrates skillful online performance, achieving a 5-year zonal mean tropospheric temperature bias within 2K, water vapor bias within 1 g/kg, and a precipitation RMSE of 0.96 mm/day. Key factors contributing to our online performance include an expressive U-Net architecture and physical thermodynamic constraints for microphysics. With microphysical constraints mitigating unrealistic cloud formation, our work is the first to demonstrate realistic multi-year cloud condensate climatology under the MMF framework. Despite these advances, online diagnostics reveal persistent biases in certain regions, highlighting the need for innovative strategies to further optimize online performance.