the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Uncertainty in aerosol effective radiative forcing from anthropogenic and natural aerosol parameters in ECHAM6.3-HAM2.3
Abstract. Interactions between aerosols, clouds, and radiation remain one of the largest sources of uncertainty in effective radiative forcing (ERF), limiting the accuracy of climate projections. Despite progress, key sources of parametric and structural uncertainty in aerosol–cloud and aerosol–radiation interactions remain poorly quantified. This study addresses this gap using a perturbed parameter ensemble (PPE) of 221 simulations with the ECHAM6.3-HAM2.3 climate model, varying 23 aerosol-related parameters that control emissions, removal, chemistry, and microphysics. The resulting global mean aerosol ERF is -1.24 Wm-2 (5–95 percentile: -1.56 to -0.89 Wm-2). We find that uncertainty in aerosol ERF is dominated by sulfate-related processes, biomass burning, size, and natural emissions. Here, for Aerosol-Cloud Interactions, DMS and biomass burning emissions are important, whereas for Aerosol-Radiation Interactions, sulfate chemistry and dry deposition are important. Despite structural differences across models, the leading causes of ERF uncertainty identified here align with findings from other PPEs.
Comparison with satellite retrievals from POLDER-3/PARASOL reveals persistent model biases in aerosol optical depth (AOD), Ångström exponent (AE), and single-scattering albedo (SSA), many of which fall within the parametric uncertainty bounds of the PPE. Sulfate-related processes account for over 40 % of AOD uncertainty, while AE and SSA uncertainties are strongly influenced by DMS, sea salt, and black carbon properties. PPEs can reduce some structural model biases through parameter adjustments, but others persist. These results highlight the need for combined efforts in parameter perturbation and structural model development to improve confidence in aerosol-forcing estimates and future climate projections.
Competing interests: One of the authors is a member of the editorial board of Atmospheric Chemistry and Physics.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1728 KB) - Metadata XML
-
Supplement
(5575 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-2848', Anonymous Referee #1, 27 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-2848/egusphere-2025-2848-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2025-2848-RC1 -
RC2: 'Comment on egusphere-2025-2848', Anonymous Referee #2, 01 Oct 2025
Review for Bhatti et al., “Uncertainty in aerosol effective radiative forcing from anthropogenic and natural aerosol parameters in ECHAM6.3-HAM2.3”
This work presents a novel PPE design using the ECHAM6.3-HAM2.3 designed to characterize uncertainty in aerosol ERF. The parameters chosen in this work focus heavily on aerosol processes, including aerosol emissions, refractive indices, hygroscopicity, deposition, and nucleation/secondary aerosol processes. Across their PPE, they generate emulators for ERF, AOD, SSA, and AE and conduct regional analysis to identify the largest regional sources of parametric uncertainty for each diagnostic. They further validate their PPE against recent aerosol satellite observations to identify areas of bias in their simulations and point out areas of structural and parametric uncertainty.
Overall, I found the paper to be well written and easy to follow. The results were interesting and linked well to broader context of PPE work done in other models such as UKESM1, HadGEM, and CESM. I think they did a good job of setting the stage for future work looking at model improvements and development in ECHAM6.3-HAM2.3, but at times I found the results to be somewhat vague in the context of model improvement. Below, I offer some minor revisions and some clarification, but I think the paper should be published promptly with some minor changes.
Major comments:
Section 2.2.1: I would like to see an additional paragraph in this section that explains significance of the parameters in the context of the model. It doesn’t have to be parameter by parameter, but it would be nice if the different categories of parameter were acknowledged. While the role of some is straightforward (emission, deposition, refractive index), I find it surprising that kappaSO2 and kappaSS don't contribute to cloud activation and would like to know briefly why that is and what role these values play for aerosol. The significance of PH_PERT also eludes me in this context (controlling aqueous reaction rate?) so it would be nice to hear why these was chosen and what they represent.
There is a lack of emission-size separation in this work for some of the more uncertain parameters such as sea salt and dust. The authors comment that this may be a direction for future work, but I think it would be interesting for the authors to comment on what the impacts of isolating size modifications for dust and sea salt could be given that mass emissions are changing both your aerosol size and your burden. In some cases, it seems the solution to fixing size biases may be to emit more mass, but the authors don’t comment on whether the issue may lie in the size parameterization of these aerosol. Please see specific references in the minor comments below.
Section 3.2.4: Most of the paper gives overall parameter uncertainty/sensitivity in different regions but little has been said about the sign of the parametric changes and how this relates to diagnostics in the model. From a model development side, I'd imagine it would be good to say what direction a parameter should be changed in order to address a given bias. Having this in the following section would be nice for the key uncertain parameters (e.g., 'future work would address increasing emissions of x and decreasing emissions of y to address a given bias'). Could you generate a simple linear correlation that targets the key parameters identified in your AOD, AE, and SSA uncertainty quantification (sections 3.2.1-3.2.3) to understand what direction they need to be tuned?
Minor comments:
Line 119: How many members have CDNCmin perturbed? Are these also randomly sampled, and are these members the only ones you use to train your GP emulator? I may be misunderstanding how CDNCmin is accounted for here and would appreciate more clarification.
Table 1: If KAPPA_SO4 and KAPPA_SS are not used for cloud droplet activation, do these values only govern aerosol size? As a follow on, does ECHAM6-HAM account for aerosol hygroscopicity when parameterizing cloud droplet activation from aerosol? Upon second reading I'm wondering if ECHAM6-HAM uses two different hygroscopicities for aerosol and for cloud activation? Related to major comment regarding additional parameter explanation.
Line 178: In addition to references to previous PPE studies, please include the model acronyms associated with each study.
Line 209: Fig. 2b for the Europe ERF uncertainty? I would expect it to be higher than 0.41 when averaged over your Europe region based on this figure but I could be wrong. Please double check and be more clear in your figure references here.
Line 250: Please include reference to figure in this paper, as well as the paper you are referring to for HadGEM (Regayre et al., 2018, I'm assuming).
Line 288: Regarding the comment on CDNCmin compensation for structural error: how do you rule out parametric uncertainty in this (i.e., from interactions of the other parameter choices)?
Line 311: When referring to Fig. 5, please also mention that it will be discussed more in sections 3.2.1-3.2.3.
Line 315: If this is true, what observational means are being shown in the Arctic and Antarctic regions in figure 4? Are you just showing available POLDER retrievals within your Arctic and Antarctic regions? Please clarify here.
Figure 5: Check colors in these uncertainty plots and double check that they are showing the correct parametric value. Panel b shows two boxes for EMI_BF (light green). The top one may be intended to be EMI_FF?
Line 323-324: AOD underestimation appears to be related in large part to biomass burning regions. Coupled with the low uncertainty in many of these regions, is it possible that structural issues hinder representation of BB in ECHAM6-HAM and/or could this be due to too limited of range in some of the parameters chosen (BB emissions, BC refractive index)?
Line 327-328: This seems to indicate that there is very high parametric sensitivity in these regions in ECHAM6-HAM. Does this suggest that parameter choice in these regions is most important for constraining the PPE, or that there may be structural issues related to what I would assume are heavily anthropogenically influenced and dusty regions?
Line 337-338: Could this also be addressed by changing emission parameter ranges? Or is this deemed to put parameter ranges outside of what is physically sound?
Line 348: Consider rewording sentence, changing ‘…AOD bias are shown…’ to ‘…AOD bias and are shown…’
Line 368: Figure reference should be Figure 6e, not Figure 6d
Line 379-380: Does this suggest that ari should not be addressed in isolation from aci?
Line 388-389: The trends in AE suggest that dust and sea salt aerosol are too small. There is high sensitivity to emissions, but emissions don’t differentiate between mass and size impacts. Can you comment on how size is treated in ECHAM6-HAM? Is there any evidence to suggest that emissions may be appropriate, but the size treatment of dust and sea salt may be the issue?
Line 410: What could the implications be for modifications to aerosol size in isolation of mass since this seems like a key issue in this area as well?
Line 415: Does this mean that increasing emissions would accentuate the bias by increasing size? Are emissions the issue or is the size parameterization the issue? I'd be interested in some discussion on this topic.
Line 416-418: Does this mean that increasing emissions would increase the bias or the other way around? I'm a little confused in this section on the sign of your parameter influence on the AE, AOD, and SSA diagnostics.
Line 439: Size and mixing state may also be playing a role. Can you comment on these parameters in ECHAM6-HAM and how they may be contributing to this bias? Does ECHAM6-HAM treat aerosol as volume mixed? That could also be overestimating absorption. Brown et al., 2021 (https://doi.org/10.1038/s41467-020-20482-9) showed an underestimation in SSA over biomass burning regions in this same model.
Line 445-447: Similar to above comment that size and mixing state may be playing a role.
Line 455: As a very minor comment, consider rewording ‘smaller size distributions’ to ‘smaller aerosol’. I say this given that a distribution can vary in number, width, and size which makes smaller seem somewhat arbitrary to me.
Line 473: What does ‘(a lack of)’ refer to? Please reword to be more clear.
Line 473: As per previous comment, consider rewording ‘larger aerosol size distributions’ to ‘larger aerosol sizes’ or ‘larger aerosol diameters’.
Citation: https://doi.org/10.5194/egusphere-2025-2848-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
900 | 89 | 13 | 1,002 | 44 | 18 | 25 |
- HTML: 900
- PDF: 89
- XML: 13
- Total: 1,002
- Supplement: 44
- BibTeX: 18
- EndNote: 25
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1