Recent Modelling Studies Systematically Underestimate the Warming from IMO2020 Shipping Regulations

Smith, Josh; Henry, Matthew; Yoshioka, Masaru; Johnson, Ben; Haywood, Jim

doi:10.5194/egusphere-2026-2654

Preprints

https://doi.org/10.5194/egusphere-2026-2654

Preprints

20 May 2026

| 20 May 2026

Status: this preprint is open for discussion and under review for Atmospheric Chemistry and Physics (ACP).

Recent Modelling Studies Systematically Underestimate the Warming from IMO2020 Shipping Regulations

Josh Smith, Matthew Henry, Masaru Yoshioka, Ben Johnson, and Jim Haywood

Abstract. The 2020 International Maritime Organisation regulations (IMO2020) reduced shipping SO₂ emissions by roughly 80%, decreasing the cooling effect of sulphate aerosols on marine clouds, leading to a positive radiative forcing. Recent Global Climate Model (GCM) studies agree on a positive Effective Radiative Forcing (ERF) of ~0.10 W m^-2 from IMO2020. However, these studies rely on parameterisations for sub-grid scale emission processes with assumptions on primary sulphate fraction, particle size, and injection altitude, which contradict observational evidence for shipping exhaust plumes. Using the UKESM1.1 climate model, we conduct sensitivity experiments to quantify the impact of these uncertainties. We find that reallocating primary sulphate from the accumulation and coarse modes to the Aitken mode increases the IMO2020 ERF from 0.10 W m^-2 to between 0.19 and 0.31 W m^-2, and additionally increasing primary sulphate fraction increases this further up to 0.41 W m^-2. This sensitivity is driven primarily by the cloud radiative effect (ΔCRE) responding to an order-of-magnitude increase in modelled aerosol number emissions for the same sulphur mass, and is consistent with earlier shipping studies using other GCMs. Because recent GCM estimates rely on the same biased sub-grid emission assumptions, we argue this underestimate is structural across recent studies, and we find that the default-parameter experiment with a 0.10 W m^-2 forcing significantly underestimates regional ΔCRE values relative to published satellite observations. An IMO2020 ERF 2 to 4 times the current consensus would explain a larger portion of Earth's energy imbalance since 2020 and of the recent global temperature surge.

Received: 08 May 2026 – Discussion started: 20 May 2026

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.

Download & links

Preprint (PDF, 7502 KB)

Supplement (3332 KB)

Download & links

Josh Smith, Matthew Henry, Masaru Yoshioka, Ben Johnson, and Jim Haywood

Status: open (until 02 Jul 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-2654', Anonymous Referee #1, 30 May 2026 reply
The authors present a detailed study using the UKESM1.1 climate model to evaluate the sensitivity of the effective radiative forcing (ERF) from the IMO2020 shipping regulations to near-source emission plume parameterizations. They show that shifting primary sulphate mass from the default accumulation and coarse modes to the smaller Aitken mode significantly increases the modelled ERF from 0.10 W m⁻² up to 0.31 W m⁻², and that increasing the mass fraction further amplifies this effect. They conclude that recent global climate model studies converge on a tighter, lower range because they share structural biases in sub-grid plume configurations.
Overall, the paper addresses an important and debated topic regarding the radiative "unmasking" effect of marine fuel regulations. It provides a good discussion around how this fits with the related literature and I enjoyed reading it. However, several major conclusions are currently overstated, and critical observational and physical constraints should be addressed before the manuscript is suitable for publication in ACP.

Major Comments:
Single-Model Limitations and Overstated Scope: The manuscript frames its findings as a definitive exposure of a systematic, structural underestimate across all recent global climate modeling literature. However, these parameter sensitivity experiments themselves are conducted entirely within a single model architecture (UKESM1.1) running in an atmosphere-only configuration. Because individual climate models feature highly distinct background aerosol fields, boundary layer representations, and vertical velocity distributions, the claim that the entire field underestimates forcing by a factor of 2 to 4 is an overinterpretation. This framing needs to be toned down to reflect a specific model sensitivity or a structural risk within shared parameter lines, rather than an absolute systemic truth.

Missing Bottom-Up Observational Validation: The authors rely heavily on aircraft campaign observations from ACRUISE to physically motivate their parameter adjustments (such as the shift to the Aitken mode and higher primary sulphate fractions). However, a direct bottom-up comparison with this data is completely missing. Showing that tweaking parameters improves a model's spatial correlation with top-down CERES satellite anomalies is not proof that the underlying plume microphysics are handled correctly. The paper needs to include a quantitative evaluation showing that the updated configurations (e.g., Ait or Ait_10) actually reproduce realistic size distributions and near-source properties (and ideally cloud droplet responses) when verified directly against the in-situ aircraft measurements.

Confounding SST Feedbacks: The experiments use fixed 2005–2014 sea surface temperatures (SSTs) and sea ice configurations to isolate ERF. While this is standard practice for diagnosing uncoupled radiative forcing, it leaves the study structurally blind to alternative physical explanations. Compelling recent work has demonstrated that a substantial portion of the post-2020 global temperature surge in general and shallow cloud anomalies in the North Pacific in particular may be a response to changing SST patterns and coupled ocean-atmosphere dynamics (Zelinka et al. 2026; Tam et al. 2026; Ceppi et al. 2026). By clamping the SSTs, the model interprets these discrepancies entirely through aerosol sensitivity to match top-of-atmosphere fluxes. This is by construction, and makes sense in order to focus on the aerosol forcing, but the authors should explicitly discuss the other contribution to the observed CERES flux changes that this study explicitly doesn’t address and argue why they assume it to be small.

Discrepancies in Regional Constraints: The comparison with regional satellite observations in Section 4.4 requires a more critical, balanced interpretation. While the default Base model underestimates Diamond’s (2023) observed annual mean corridor forcing, the updated Ait_10 experiment yields an annual mean of 1.3 W m⁻², which overestimates the observation (~0.5 W m⁻²) by more than a factor of two. Furthermore, the large gap between Zhang et al.’s (2025) regional analysis sub-total (0.074 W m⁻²) and the model’s global 0.41 W m⁻² is dismissed too easily, glossing over the risk of model over-sensitivity due to grid-scale emission dilution.

Specific Comments:
P8 L191-193: The text notes that Ait_88nm has a larger total ERF than Ait but a smaller ΔCRE term, attributing the difference to noise. This, and the results in Figure 3 (particularly the RA in High 4.5 and Ait) seem to be at odds with the statement that ‘IRF and clear-clean adjustments remain relatively stable across the experiments’. Please expand this discussion.

P9 L195: There is a double word typo in going from from 2.5%.

P13 L300: Typo in text; change gobal to global.

References
Zelinka, M. D., Myers, T. A., Qin, Y., Chao, L.-W., Klein, S. A., Po-Chedley, S., Ma, P.-L., Wall, C. J., Ceppi, P., and Gettelman, A.: Recent cloud trends and extremes reaffirm established bounds on cloud feedback and aerosol-cloud interactions, Commun Earth Environ, https://doi.org/10.1038/s43247-026-03461-8, 2026.

Tam, R. Y. S., Myers, T. A., Zelinka, M. D., Proistosescu, C., Lin, Y.-J., and Marvel, K.: Meteorological drivers of the low-cloud radiative feedback pattern effect and its uncertainty, Atmos. Chem. Phys., 26, 4289–4311, https://doi.org/10.5194/acp-26-4289-2026, 2026

Ceppi, P., Wilson Kemsley, S., Andersen, H., Andrews, T., Kramer, R. J., Nowack, P., Wall, C. J., and Zelinka, M. D.: Emerging low-cloud feedback and adjustment in global satellite observations, Atmos. Chem. Phys., 26, 4153–4171, https://doi.org/10.5194/acp-26-4153-2026, 2026

Reply
Citation: https://doi.org/10.5194/egusphere-2026-2654-RC1
RC2: 'Comment on egusphere-2026-2654', Anonymous Referee #2, 17 Jun 2026 reply

The manuscript “Recent Modelling Studies Systematically Underestimate the Warming from IMO2020 Shipping Regulations” use a single climate model (UKESM1.1) to conduct sensitivity experiments to quantify the impact of sub-grid scale emission processes on the estimated Effective Radiative Forcing following the IMO regulations for SO2 shipping emissions.

As for aerosol ERF in general, there are large uncertainties related to magnitude of the ERF due to IMO2020, and a sensitivity study like this is very useful to quantify the impact of these assumptions on the modelling results. However, the conclusion the authors draw (also displayed in the title), based on a single model study and a highly uncertain process include many other processes than the sub-grid scale emission parameterizations, may be overstated. The last sentences of the main text highlight this limitation and recommend that this should be investigated in other models.
Other comments:

L14: Can ∆CRE be observed from satellites?

L21: reflecting additional solar radiation. Delete additional.
L29: Define SOx. 8 Tg per year is not clear when you use the term SOx.
L54: Are there any conclusions from Ahsan et al., 2023, the multi-model study for different emission choices, that can be highlighted in this paper, as this paper is only a single model study?
Figure 2: Again you use SOx which is not clear. The unit in Fig 2a is Tg yr-1, which does not make so much sense for a spatial plot. Use per square meter.
L170: Under the header observational data only CERES data are given. The cloud radiative effect derived from the CERES data is not only dependent on the aerosols, but also feedback processes. Would be useful with more discussions related to this, as this is the only observational-based product directly used.
Figure 5. CERES CRE anomaly is presented as observed. These are not directly observed and assumptions are made. How robust are the CRE anomalies from CERES? You also mention that the estimate in Hansen et al. has been scrutinized for potentially conflating aerosol forcing with coupled SST–cloud feedbacks. In your simulations, you use fixed SSTs, but compare the results with CERES data where SST has evolved. How valuable is this comparison?
L187: Do you need “updated” here?
L204: SEM not defined in the main text.
Figure 4: This figure was a bit confusing. Took me a while to understand that the region was the aggregate of the three listed. As Fig 4d is first introduced, move this to 4a? On line 205-210, to help the reader, refer to the different panels in Fig. 4 when explaining the results of the figures.
L280: What kind of satellite measurements do Zhang et al use?
L308: SEA not defined in the main text.
L325: the aerosols are not emitted (still a large fraction of SO2).

Reply

Citation: https://doi.org/10.5194/egusphere-2026-2654-RC2

Josh Smith, Matthew Henry, Masaru Yoshioka, Ben Johnson, and Jim Haywood

Supplement

https://doi.org/10.5194/egusphere-2026-2654-supplement

Data sets

Data and Code in support of: Recent Modelling Studies Systematically Underestimate the Warming from IMO2020 Shipping Regulations Josh Smith https://doi.org/10.5281/zenodo.20076407

Josh Smith, Matthew Henry, Masaru Yoshioka, Ben Johnson, and Jim Haywood

Viewed

Total article views: 325 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
224	87	14	325	28	9	10

HTML: 224
PDF: 87
XML: 14
Total: 325
Supplement: 28
BibTeX: 9
EndNote: 10

Views and downloads (calculated since 20 May 2026)

Month	HTML	PDF	XML	Total
May 2026	178	64	11	253
Jun 2026	46	23	3	72

Cumulative views and downloads (calculated since 20 May 2026)

Month	HTML	PDF	XML	Total
May 2026	178	64	11	253
Jun 2026	46	23	3	72

Viewed (geographical distribution)

Total article views: 323 (including HTML, PDF, and XML) Thereof 323 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 17 Jun 2026

Short summary

In 2020, new international rules cut sulphur in shipping fuels by roughly eighty percent, removing particles that had been brightening clouds over the oceans and reflecting sunlight. Using a climate model, we tested how this loss of cooling has been calculated and found that recent studies have
likely underestimated the resulting warming by a factor of two to four. This means cleaner shipping fuels may be contributing more to recent global warming than previously recognised.


Total:	0
HTML:	0
PDF:	0
XML:	0