Jong et al. (2025) A computationally efficient method to model similar and alternate stratospheric aerosol injection experiments using prescribed aerosols in a lower-complexity version of the same model: a case study using CESM(CAM) and CESM(WACCM)

Identification

• Journal: Geoscientific model development
• Year: 2025
• Date: 2025-11-18
• Authors: Jasper de Jong, Daniel Pflüger, Simone Lingbeek, Claudia Wieners, Michiel Baatsen, René R. Wijngaard
• DOI: 10.5194/gmd-18-8679-2025

Research Groups

• Institute for Marine and Atmospheric research Utrecht, Utrecht University, Utrecht, the Netherlands

Short Summary

This study develops a computationally efficient method using pattern-scaling to prescribe stratospheric aerosol injection (SAI) forcing in lower-complexity climate models (CESM(CAM)) based on data from full-complexity models (CESM(WACCM)). The method successfully replicates tropospheric climate responses to SAI across various scenarios and model configurations, significantly reducing computational costs.

Objective

• To develop and validate a computationally efficient method for modeling stratospheric aerosol injection (SAI) experiments in lower-complexity climate models (CESM(CAM)) by prescribing aerosol forcing derived from full-complexity models (CESM(WACCM)), thereby reducing computational resource requirements while accurately reproducing tropospheric climate responses.

Study Configuration

• Spatial Scale:
  • CESM2(WACCM6): 1° nominal horizontal resolution, 70 vertical levels, model top at 6 × 10⁻⁴ Pa (140 km).
  • CESM2(CAM6): 1° nominal horizontal resolution, 32 vertical levels, model top at 300 Pa (40 km).
  • CESM1.0.4(CAM5): 1° nominal horizontal resolution. Also, a high-resolution case with 0.5° atmosphere and 0.1° ocean.
• Temporal Scale:
  • Simulations from 2020 onwards, some extending beyond 2100 with stabilized forcings.
  • Reference periods: [2016–2035] for CESM2, [1990–2009] for CESM1 spinup.
  • Aerosol field derivation period: [2070–2100].
  • Late-century analysis period: [2080–2099].

Methodology and Data

• Models used:
  • CESM2(WACCM6) (Whole Atmosphere Community Climate Model Version 6)
  • CESM2.1.3(CAM6.0) (Community Atmosphere Model Version 6.0)
  • CESM1.0.4(CAM5) (Community Atmosphere Model Version 5)
  • Modal Aerosol Module (MAM4) for aerosol simulation in CESM2 models.
• Data sources:
  • Pre-existing CESM2(WACCM6) simulations (WACCM-Control, WACCM-SAI2020 from Tilmes et al., 2020; Danabasoglu, 2019).
  • Pre-existing CESM1(WACCM) simulations (Geoengineering Large ENSemble (GLENS) from Tilmes et al., 2018; Kravitz et al., 2017).
  • SSP5-8.5 and RCP8.5 scenarios for greenhouse gas and anthropogenic forcings.
  • Pattern-scaling techniques to derive aerosol forcing fields, relating annual intensity of forcing variables to global mean stratospheric aerosol optical depth (n_AOD) using power-law or exponential fits.
  • Proportional-integral (PI) feedback controller (MacMartin et al., 2014; Kravitz et al., 2016, 2017; Tilmes et al., 2020) to dynamically regulate aerosol injection based on global mean surface temperature (GMST) targets.

Main Results

• The proposed method successfully achieves global mean surface temperature (GMST) targets in all CESM(CAM) experiments, with multi-year average deviations typically less than 3% of the GMST change in the control simulation, and less than 0.5% for the direct mimicry scenario (SAI2020).
• Large-scale temperature responses, including interhemispheric and equator-to-pole gradients, are well reproduced in CAM-SAI2020, with deviations from target values less than 15% of their change in the control simulation.
• Regional surface temperature and precipitation responses to SAI forcing are largely similar between WACCM and CAM, with differences generally significantly smaller than interannual inter-model variability in the troposphere (performance index < 1).
• The largest inter-model differences in surface temperature response are observed in the Arctic region and Northeast Asia, where CAM shows more warming, partly attributed to existing model biases and differences in high-latitude climate variability.
• Stratospheric temperature and circulation changes in CAM-SAI2020 correspond well to WACCM-SAI2020 in the troposphere, but discrepancies increase with altitude due to CAM's lack of interactive stratospheric chemistry and higher vertical resolution.
• The method demonstrates robustness across different initial climate states, model versions (CESM1 vs. CESM2), and increased spatial resolution (1° vs. 0.5° atmosphere, 0.1° ocean).
• Delayed SAI scenarios (e.g., CAM-SAI2080) can lead to considerable changes in large-scale temperature gradients not explicitly controlled, such as a 150% decrease in the interhemispheric gradient and a 54% decrease in the equator-to-pole gradient relative to the control simulation, linked to persistent climate changes like Atlantic Meridional Overturning Circulation (AMOC) weakening.

Contributions

• Proposes a novel, computationally efficient method for modeling SAI experiments in lower-complexity models (CESM(CAM)) by prescribing aerosol forcing derived from full-complexity models (CESM(WACCM)).
• Integrates pattern-scaling techniques with a feedback-feedforward controller to dynamically adjust SAI intensity, enabling the achievement of climate targets (e.g., GMST) despite different model sensitivities.
• Improves upon previous linear scaling approaches by approximately capturing non-linear relationships between aerosol optical depth (AOD) and other aerosol-related quantities.
• Validates the method across multiple model versions (CESM1, CESM2), spatial resolutions (1°, 0.5° atmosphere), and SAI forcing scenarios, demonstrating its transferability and robustness for tropospheric climate impact research.
• Provides a valuable tool for generating larger ensembles, exploring new SAI scenarios, and running higher-resolution simulations at significantly reduced computational cost, particularly for studies focused on tropospheric and surface climate impacts.

Funding

• Dutch Ministry for Education, Culture and Science (grant no. 16604027)
• Sector Plan Science and Technology

Citation

@article{Jong2025computationally,
  author = {Jong, Jasper de and Pflüger, Daniel and Lingbeek, Simone and Wieners, Claudia and Baatsen, Michiel and Wijngaard, René R.},
  title = {A computationally efficient method to model similar and alternate stratospheric aerosol injection experiments using prescribed aerosols in a lower-complexity version of the same model: a case study using CESM(CAM) and CESM(WACCM)},
  journal = {Geoscientific model development},
  year = {2025},
  doi = {10.5194/gmd-18-8679-2025},
  url = {https://doi.org/10.5194/gmd-18-8679-2025}
}