Wang et al. (2025) Challenges in global climate models to represent cloud response to aerosols: insights from volcanic eruptions

Identification

• Journal: Nature Communications
• Year: 2025
• Date: 2025-12-18
• Authors: Yu Wang, David Neubauer, Ying Chen, George Jordan, Florent Malavelle, Tianle Yuan, Daniel G. Partridge, Paul R. Field, Hao Wang, Minghuai Wang, Martine Michou, Pierre Nabat, Anton Laakso, Gunnar Myhre, Ulrike Lohmann
• DOI: 10.1038/s41467-025-67359-3

Research Groups

• School of GeoSciences, University of Edinburgh, Edinburgh, UK
• Institute for Atmospheric and Climate Science, ETH Zürich, Zürich, Switzerland
• School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham, UK
• Met Office Hadley Centre, Exeter, UK
• Met Office, Exeter, UK
• Goddard Earth Sciences Technology and Research (GESTAR) II, University of Maryland, Baltimore County, Baltimore, MD, USA
• Sciences and Exploration Directorate, Goddard Space Flight Center, Greenbelt, MD, USA
• College of Engineering, Mathematics, and Physical Sciences, University of Exeter, Exeter, UK
• School of Earth and Environment, University of Leeds, Leeds, UK
• School of Atmospheric Sciences, Nanjing University, Nanjing, China
• Météo-France, CNRS, Univ. Toulouse, CNRM, Toulouse, France
• Finnish Meteorological Institute, Kuopio, Finland
• CICERO Center for International Climate Research, Oslo, Norway

Short Summary

This study confronts a large-scale observational constraint of cloud response to aerosols, derived from the Holuhraun-2014 volcanic eruption, against six global climate models. It reveals that models significantly underestimate aerosol-induced cloud cover responses, a major source of uncertainty in climate projections.

Objective

• To confront a large-scale observational constraint of cloud susceptibility, derived from the Holuhraun-2014 volcanic eruption and machine learning with satellite observations, against six diverse global climate models to advance understanding of aerosol-cloud interaction (ACI) simulation uncertainty.

Study Configuration

• Spatial Scale: Global (represented by the Holuhraun-2014 eruption over the North Atlantic, spanning millions of square kilometres, and global CMIP6 simulations).
• Temporal Scale: October 2014 (for Holuhraun eruption simulations); 2000-2014 (for CMIP6 global cloud susceptibility comparison).

Methodology and Data

• Models used: ECHAM6.3-HAM2.3, CESM2.1.0, UKESM1, CNRM-ESM2-1, ECHAM6.3-SALSA2.0, CAM5.3-Oslo (from the VolcACI project). MPI-ESM-1-2-HAM and UKESM1-0-LL (from CMIP6). Various cloud schemes and process sensitivities were tested within ECHAM6.3-HAM2.3 (e.g., REF, REF-XR, P3, P3-XR, PROG_CC).
• Data sources: Machine-learning-based observational constraint of cloud susceptibility (derived from satellite observations, specifically MODIS cloud products from Aqua and Terra), Holuhraun-2014 volcanic eruption emissions, reanalysis data (ERA-Interim or MERRA2) for nudging model winds and surface pressure.

Main Results

• Marine liquid cloud optical depth (COD) responses to aerosols are reasonably well simulated by global climate models (GCMs), but often through compensating errors where an underestimated Twomey effect is balanced by an overestimated liquid water path (LWP) adjustment.
• All six GCMs largely underestimate cloud cover responses to aerosols, with five of them falling outside the 90% confidence level of the observational constraint, indicating a persistent and significant bias.
• This systematic bias in cloud cover response remains despite extensive tuning of five distinct cloud schemes and testing various key cloud microphysical processes within the ECHAM6.3-HAM2.3 model.
• The Holuhraun-2014 eruption provides a globally representative analogue for understanding cloud responses to anthropogenic aerosol perturbations, and model internal variability introduces only a small degree of uncertainty compared to the observed biases.

Contributions

• Provides a robust, large-scale observational constraint for aerosol-cloud interactions (ACI) using a natural experiment (volcanic eruption) combined with machine learning and satellite data.
• Systematically evaluates ACI simulation uncertainty across a diverse group of state-of-the-art GCMs, pinpointing the underestimation of cloud cover response as a major structural bias.
• Highlights the urgent need to improve cloud cover parameterizations and related microphysical processes in GCMs to reduce significant uncertainties in climate projections and estimations of climate sensitivity.
• Proposes the Holuhraun-2014 eruption as a benchmark case for the wider climate community to validate cloud susceptibilities in addition to cloud properties.

Funding

• Swiss National Supercomputing Centre (CSCS) under project ID s1144
• Forth cluster at the University of Edinburgh
• Startup fund for lectureship from the University of Edinburgh
• Mr. Philippe Sarasin and the ETH Zürich Foundation
• UKRI-NERC projects QUESTION (NE/B001024/1)
• UKRI-NERC projects QR-CODE (NE/Z503800/1)
• NERC projects ADVANCE (NE/S015671/1)
• NERC projects CLOSURE (NE/W001713/1)
• European Union’s Horizon 2020 CONSTRAIN grant (820829)

Citation

@article{Wang2025Challenges,
  author = {Wang, Yu and Neubauer, David and Chen, Ying and Jordan, George and Malavelle, Florent and Yuan, Tianle and Partridge, Daniel G. and Field, Paul R. and Wang, Hao and Wang, Minghuai and Michou, Martine and Nabat, Pierre and Laakso, Anton and Myhre, Gunnar and Lohmann, Ulrike},
  title = {Challenges in global climate models to represent cloud response to aerosols: insights from volcanic eruptions},
  journal = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-67359-3},
  url = {https://doi.org/10.1038/s41467-025-67359-3}
}