Lamb et al. (2025) Perspectives on Systematic Cloud Microphysics Scheme Development With Machine Learning

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Journal of Advances in Modeling Earth Systems
• Year: 2025
• Date: 2025-12-31
• Authors: Kara D. Lamb, Clare E. Singer, Kaitlyn Loftus, Hugh Morrison, Margaret Powell, Joseph Ko, Jatan Buch, Arthur Z. Hu, Marcus van Lier Walqui, Pierre Gentine
• DOI: 10.1029/2025ms005341

Research Groups

Not specified in the provided abstract.

Short Summary

This perspectives paper synthesizes recent progress and outlines challenges and opportunities for applying machine learning to improve cloud microphysics parameterizations, aiming to reduce significant parametric and structural uncertainties in weather and and climate models.

Objective

• To outline the challenges that must be addressed to apply machine learning (ML) toward cloud microphysics scheme development.
• To synthesize recent progress in using data-driven methods, including ML, to improve cloud microphysics parameterizations.
• To highlight opportunities to address key uncertainties (parametric, structural, aleatoric, and epistemic errors) in cloud microphysics.

Study Configuration

• Spatial Scale: From small-scale cloud microphysical processes (unresolved in large-eddy, weather, or climate models) to their significant impact at the climate scale (global).
• Temporal Scale: From the short-term evolution of liquid droplets and ice crystals to long-term climate scale impacts, aiming for consistency across temporal scales.

Methodology and Data

• Models used: The paper discusses the improvement of existing cloud microphysical schemes within weather and climate models; it does not present new model simulations.
• Data sources: High-fidelity simulations and observations.

Main Results

• Cloud microphysics remains a major source of parametric and structural uncertainty in weather and climate models.
• Machine learning (ML) offers significant potential to minimize these uncertainties by leveraging high-fidelity simulations and observations.
• ML can improve microphysical schemes through both bottom-up and top-down constraints.
• Promising methods include differentiable programming, ML-enhanced sampling strategies, and the creation of large-scale benchmark datasets to bridge the gap between observations and models.

Contributions

• Provides a comprehensive synthesis of recent advancements in applying data-driven methods, particularly ML, to enhance cloud microphysics parameterizations.
• Identifies and categorizes key uncertainties (parametric, structural, aleatoric, epistemic) in cloud microphysics and discusses how ML can address them.
• Outlines a roadmap for future research by highlighting specific opportunities and promising methodologies (e.g., differentiable programming, ML-enhanced sampling) and data needs (benchmark datasets) to improve consistency across scales.

Funding

Not specified in the provided abstract.

Citation

@article{Lamb2025Perspectives,
  author = {Lamb, Kara D. and Singer, Clare E. and Loftus, Kaitlyn and Morrison, Hugh and Powell, Margaret and Ko, Joseph and Buch, Jatan and Hu, Arthur Z. and Walqui, Marcus van Lier and Gentine, Pierre},
  title = {Perspectives on Systematic Cloud Microphysics Scheme Development With Machine Learning},
  journal = {Journal of Advances in Modeling Earth Systems},
  year = {2025},
  doi = {10.1029/2025ms005341},
  url = {https://doi.org/10.1029/2025ms005341}
}