Comment on gmd-2023-29

Stephanie Fiedler, Vaishali Naïk, Fiona M. O’Connor, Chris Smith, Robert Pincus, Paul T. Griffiths, Ryan J. Kramer, Toshihiko Takemura, Robert J. Allen, Ulaş İm, Matthew Kasoar, Angshuman Modak, Steven T. Turnock, Apostolos Voulgarakis, Duncan Watson‐Parris, Daniel M. Westervelt, Laura J. Wilcox, Alcide Zhao, W. J. Collins, Michael Schulz, Gunnar Myhre, Piers ForsterOpenalex Unknown 2023 · DOI · OpenAlex

Abstract. The climate science community aims to improve our understanding of climate change due to anthropogenic influences on atmospheric composition and the Earth's surface. Yet not all climate interactions are fully understood and diversity in climate model experiments persists as assessed in the latest Intergovernmental Panel on Climate Change (IPCC) assessment report. This article synthesizes current challenges and emphasizes opportunities for advancing our understanding of climate change and model diversity. The perspective of this article is based on expert views from three multi-model intercomparison projects (MIPs) – the Precipitation Driver Response MIP (PDRMIP), the Aerosol and Chemistry MIP (AerChemMIP), and the Radiative Forcing MIP (RFMIP). While there are many shared interests and specialisms across the MIPs, they have their own scientific foci and specific approaches. The partial overlap between the MIPs proved useful for advancing the understanding of the perturbation-response paradigm through multi-model ensembles of Earth System Models of varying complexity. It specifically facilitated contributions to the research field through sharing knowledge on best practices for the design of model diagnostics and experimental strategies across MIP boundaries, e.g., for estimating effective radiative forcing. We discuss the challenges of gaining insights from highly complex models that have specific biases and provide guidance from our lessons learned. Promising ideas to overcome some long-standing challenges in the near future are kilometer-scale experiments to better simulate circulation-dependent processes where it is possible, and machine learning approaches for faster and better sub-grid scale parameterizations where they are needed. Both would improve our ability to adopt a smart experimental design with an optimal tradeoff between resolution, complexity and simulation length. Future experiments can be evaluated and improved with sophisticated methods that leverage multiple observational datasets, and thereby, help to advance the understanding of climate change and its impacts.

Introduction: The Perturbation-Response Paradigm

Predicting the Earth's future climate is fundamentally hindered by "model spread"—the tendency for different Earth System Models (ESMs) to produce varying results under identical inputs. A primary source of this uncertainty is the perturbation-response paradigm: the complex chain of events where a change in atmospheric composition (perturbation) translates into an energy imbalance (forcing) and eventually a physical change in temperature or precipitation (response).

Historically, these steps were studied in isolation. Aerosol experts, chemists, and radiation physicists often employed disparate definitions and diagnostic tools, complicating cross-model comparisons. This paper synthesizes the collective wisdom from three major climate modeling initiatives—AerChemMIP, PDRMIP, and RFMIP—to establish a unified framework for characterizing these interactions.

Methodological Framework: Bridging the MIPs

The collaboration between these three Multi-model Intercomparison Projects (MIPs) covers the entire causal chain of climate change:

  • AerChemMIP: Focuses on the "upstream" phase—how emissions evolve into atmospheric concentrations of aerosols and reactive gases.
  • RFMIP: Targets the "middle" phase—how those concentrations alter the Earth's radiation budget.
  • PDRMIP: Examines the "downstream" phase—how different forcing agents (e.g., black carbon vs. CO2) uniquely impact global precipitation patterns.

A significant methodological advancement is the standardization of Effective Radiative Forcing (ERF) calculations. Unlike older metrics, ERF accounts for "rapid adjustments"—fast atmospheric changes, such as cloud shifts, that occur before the ocean warms. The community has converged on two primary estimation methods: the Fixed Sea-Surface Temperature (SST) method and the Regression (Gregory) method.

Methods for calculating Effective Radiative Forcing (ERF)

Conceptual Challenges: Navigating the "Abyss"

The authors introduce the "Abyss" framework to categorize current limitations in climate modeling. The Computing Capacity Abyss represents the inherent trade-off between model resolution, complexity, and simulation length. Researchers must often sacrifice one to achieve the others, as illustrated in the "trade-off triangle."

The Computing Capacity Abyss: Trade-offs between resolution, complexity, and simulation length

Complementing this is the Process Understanding Abyss, which refers to physical or chemical interactions—such as certain biogeochemical feedbacks—that are either missing from models or so poorly understood that they cannot be accurately simulated. For example, while CO2 is the primary driver of warming, methane emissions from warming wetlands represent a major positive feedback that remains highly uncertain across current models.

Key Insights and Results

  • Forcing Quantification: By 2014 (relative to 1850), aerosols remain the largest source of uncertainty in radiative forcing, exhibiting a wide model spread in their cooling effect.
  • Standardization as a Force Multiplier: The move toward common diagnostics, such as "double radiation calls" to isolate instantaneous forcing, has allowed researchers to pinpoint why models disagree, moving beyond simply noting the existence of a spread.
  • Precipitation Drivers: Results from PDRMIP highlight that precipitation response depends heavily on the vertical distribution of energy absorption in the atmosphere, rather than just the total energy change at the top of the atmosphere.

Future Outlook: High-Resolution and Machine Learning

To bridge the identified "abysses," the paper identifies two essential technological paths:

  1. Kilometer-Scale Modeling: Moving to "storm-resolving" scales (1–5 km) allows models to explicitly simulate clouds and circulation, bypassing the need for error-prone sub-grid parameterizations.
  2. Machine Learning (ML): ML can be used to create "emulators"—fast, data-driven versions of complex model components like atmospheric chemistry—to significantly reduce computational costs.

By combining these approaches with improved diagnostics and better integration with observational datasets, the climate science community can move toward more robust predictions of climate change and its societal impacts.

Summary updated

Abstract

<strong class="journal-contentHeaderColor">Abstract.</strong> The climate science community aims to improve our understanding of climate change due to anthropogenic influences on atmospheric composition and the Earth's surface. Yet not all climate interactions are fully understood and diversity in climate model experiments persists as assessed in the latest Intergovernmental Panel on Climate Change (IPCC) assessment report. This article synthesizes current challenges and emphasizes opportunities for advancing our understanding of climate change and model diversity. The perspective of this article is based on expert views from three multi-model intercomparison projects (MIPs) &ndash; the Precipitation Driver Response MIP (PDRMIP), the Aerosol and Chemistry MIP (AerChemMIP), and the Radiative Forcing MIP (RFMIP). While there are many shared interests and specialisms across the MIPs, they have their own scientific foci and specific approaches. The partial overlap between the MIPs proved useful for advancing the understanding of the perturbation-response paradigm through multi-model ensembles of Earth System Models of varying complexity. It specifically facilitated contributions to the research field through sharing knowledge on best practices for the design of model diagnostics and experimental strategies across MIP boundaries, e.g., for estimating effective radiative forcing. We discuss the challenges of gaining insights from highly complex models that have specific biases and provide guidance from our lessons learned. Promising ideas to overcome some long-standing challenges in the near future are kilometer-scale experiments to better simulate circulation-dependent processes where it is possible, and machine learning approaches for faster and better sub-grid scale parameterizations where they are needed. Both would improve our ability to adopt a smart experimental design with an optimal tradeoff between resolution, complexity and simulation length. Future experiments can be evaluated and improved with sophisticated methods that leverage multiple observational datasets, and thereby, help to advance the understanding of climate change and its impacts.
BibTeX
@inproceedings{fiedler2023comment,
  author = {Stephanie Fiedler and Vaishali Naïk and Fiona M. O’Connor and Chris Smith and Robert Pincus and Paul T. Griffiths and Ryan J. Kramer and Toshihiko Takemura and Robert J. Allen and Ulaş İm and Matthew Kasoar and Angshuman Modak and Steven T. Turnock and Apostolos Voulgarakis and Duncan Watson‐Parris and Daniel M. Westervelt and Laura J. Wilcox and Alcide Zhao and W. J. Collins and Michael Schulz and Gunnar Myhre and Piers Forster},
  title = {{Comment on gmd-2023-29}},
  booktitle = {Openalex Unknown},
  year = {2023},
  doi = {10.5194/gmd-2023-29-rc1},
}
Figure