the Creative Commons Attribution-NonCommercial 4.0 Deed License.
the Creative Commons Attribution-NonCommercial 4.0 Deed License.
A computationally efficient method to model Stratospheric Aerosol Injection experiments
Abstract. Climate model simulations incorporating stratospheric aerosol injection (SAI) generally require more computational resources compared to out-of-the-box applications, due to the importance of stratospheric chemistry. This presents a challenge for SAI research, especially because there are numerous ways and scenarios through which SAI can be implemented. Here, we propose a novel method that allows SAI simulations to be performed without interactive stratospheric chemistry, saving a significant portion of the computational budget. The method requires a pre-existing dataset of an SAI experiment and its corresponding control experiment, with active stratospheric chemistry. The data is converted into a set of relations to determine the forcing fields given any required optical depth of the aerosol field. This makes the method suitable for applications that use dynamical feedback controllers. The results of climate simulations with aerosols prescribed by our method are in close agreement with those from full-complexity model, even for different model versions, resolutions and forcing scenarios.
Status: open (until 17 Jun 2025)
-
RC1: 'Comment on egusphere-2025-1476', Anonymous Referee #1, 10 May 2025
reply
This is a fantastic paper. The work that the authors have done has the power to make complex geoengineering simulations more accessible. The development and exploration of the methodology are done quite well. I do have some comments:
I don’t have a sense of pros and cons, i.e., when this simpler method would work versus when you need a more complex model like WACCM. I don’t expect anything thorough, but if the authors could provide some opinions on this, it would be helpful.
There are situations where the authors claim to have explored a variety of scenarios and conclude that the method is robust to different scenarios. This is only partially true. You may get different answers if you use a different background scenario, for example one with strong mitigation or changes in tropospheric aerosols, as that will change the spatial patterns of forcing. Some appropriate caveats would be useful.
Lines 152-153: I think providing more details about the simulations here would be useful. I got a little lost. I suggest moving Table C1 into the main body of the text and expanding it so that it has more information about the specifications of each simulation.
Bullets on page 3: This is essentially pattern scaling. There’s a lot of literature you can lean on to show that this is a sensible thing to do.
Line 101: What does “similar” mean?
Line 186: I think you mean monotonically?
Section 2.2: It would be useful if you said somewhere that the approximations you make are good enough for this purpose, as the point of a feedback algorithm is to correct for such uncertainties. MacMartin et al. and Kravitz et al. both say this if you need citations.
Line 241: Your errors seem kind of high. Looking at Figure 2, an error of 0.4°C is a lot. This likely means your controller isn’t tuned as well as it could be, which isn’t a big deal, but it would be worth saying so.
Figure 2: I found the panels confusing, in that you’re mixing and matching units.
Line 257: Per the above comment, maybe change “well” to “adequately”.
Line 292: This is correct but also a strawman argument. You didn’t try to restore the climate completely.
Figure 5: I’m having trouble making sense of how important these results are. I wonder if you could compute z-scores (or something like that) so I would know whether the CAM minus WACCM differences are large compared to the natural variability of WACCM.
I did not find the paragraph on lines 376, nor Appendix B, terribly convincing. If you heat the stratosphere by 24°C, you are going to have substantial influences on ozone, and we know that ozone has an influence on surface climate. I would be more comfortable if you simply said that this is what you did, its effects of ozone changes on climate are likely smaller than the effects of the stratospheric heating on climate, and this should be explored further. That puts you on much safer ground.
Lines 400-402: Kravitz et al. (2014) demonstrated that the controller is likely robust to these sorts of differences. That gives some confidence that your controller indeed can handle this.
Lines 427ff: See recent work from the Cornell group, specifically led by Farley or Brody. They’re doing the initial steps of what you propose.
Code availability: Journals tend to want a fixed repository (e.g., Zenodo) rather than a changeable repository (Github).
Citation: https://doi.org/10.5194/egusphere-2025-1476-RC1
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 67 | 0 | 0 | 67 | 0 | 0 |
- HTML: 67
- PDF: 0
- XML: 0
- Total: 67
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1