the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2024
How Does the Latitude of Stratospheric Aerosol Injection Affect the Climate in UKESM1?
Abstract. Stratospheric Aerosol Injection (SAI) refers to a climate intervention method by which aerosols are intentionally added to the lower stratosphere to enhance sunlight reflection and offset some of the adverse effects of global warming. The climate outcomes of SAI depend on the location, amount, and timing of injection, as well as the material used. Here, we isolate the role of the latitude of SO2 injection by comparing different scenarios which have the same global-mean temperature target, altitude of injection, and hemispherically symmetric injection rates. These are: injection at the equator (EQ), and injection at 15° N and S (15N+15S), at 30° N and S (30N+30S), and at 60° N and S (60N+60S). We show that injection at the equator leads to many undesirable side effects, such as a residual Arctic warming, significant reduction in tropical precipitation, reductions in high-latitude ozone, tropical lower stratospheric heating, and strengthening of the stratospheric jets in both hemispheres. Additionally, we find that the most efficient injection locations are the subtropics (15 and 30° N and S), although the 60N+60S strategy only requires around 30 % more SO2 injection for the same amount of cooling; the latter also leads to much less stratospheric warming but only marginally increases high-latitude surface cooling. Finally, while all the SAI strategies come with trade-offs, we demonstrate that the 30N+30S strategy has, on balance, the least negative side effects and is easier to implement than a multi-latitude controller algorithm; thus it is a good candidate strategy for an inter-model comparison.
- Preprint
(1682 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
- RC1: 'Comment on egusphere-2024-1565', Anonymous Referee #1, 16 Jul 2024
-
RC2: 'Comment on egusphere-2024-1565', Anonymous Referee #2, 16 Jul 2024
This study compares the climate response to stratospheric aerosol injection in the earth system model UKESM under varying injection latitude. It is a clear presentation of an important new set of simulations, which will be a useful reference going forwards. It makes a valuable contribution in clarifying the trade-offs between different possible injection latitudes. I would like to see it published given some minor revisions.
General comments:
In the abstract, the statement "we demonstrate that the 30N+30S strategy has, on balance, the least negative side effects" is too strong. There is no comprehensive assessment of the side effects or their impacts presented here, and their results suggest 60° injection has smaller side-effects on some metrics. I suggest the authors make a more limited statement, or at least specify on what metric they are making this judgement.
It would be good to include maps of the global surface climate response, at least for temperature and precipitation, under the different scenarios. This could be in Supplementary if article length is an issue. While not critical to the arguments being made here, these simulations will be a useful reference point for others and I suspect many will want to see the full spatial picture.
The zonal line plots (fig 1, fig 2a,c,d) don't currently indicate the ensemble spread. This information would be useful to include (perhaps via a new figure in Supplementary if it is too cluttered here), since it’s not currently obvious where the differences between scenarios are robust to internal variability.
Figure 2C: there are two aspects here which could be a little misleading, and which adding some more subplots (or a new fig in supplementary) could resolve. First, the equatorial injection strategy looks as though it strongly undercools the arctic relative to the other strategies, but this is partly an artifact of the controller failing to meet the 1.5K target for this strategy. Perhaps also plotting the residuals for each strategy relative to the baseline global mean temperature it actually achieves (e.g. relative to a ~1.75K world for equatorial injection) would show more clearly the relative differences in latitudinal pattern of cooling between the strategies. Second, plots like these obscure the lower latitude structure because of the strong feedback driven arctic amplification of residuals (as already discussed in the text). Perhaps it would be worth also plotting the residual warming scaled by the local warming from pre-industrial at the baseline?
Minor points:
- Line 95: "The larger injection rate is thus due to the lower efficacy of tropical forcing (Kang and Xie, 2014) and to the confinement of aerosols inside the tropical pipe, enhancing the formation of larger aerosols which sediment faster." A third potential contribution is that local forcing is sub-linear in local AOD, so concentrating AOD in one region reduces the global mean forcing.
- Figure 2b: define the error bars here. Are they the ensemble spread?
- Line 99: "The larger injection rates for 60N+60S, on the other hand, arise due to faster 100 removal of aerosols when injected near the descending branch of the Brewer Dobson Circulation (BDC)". There are other potential contributions here; latitudinal even-ness of the AOD distribution (as above), and latitudinal variation in insolation, underlying albedo, and feedbacks
- Label text is too small on the multi-panel figures (4,5,6)
- Line 227-228 perhaps rephrase to make it clear that it is in CESM2 that the high latitude runs were seasonal and lower altitude.
- Line 238: "The main takeaway is that the 30N+30S strategy is one of the most efficient strategies in terms of amounts of SO2 needed". Perhaps better to give a rough statement of the relative efficiencies. There aren't many strategies compared, so "one of the best" is not all that meaningful. My interpretation is 30° is only slightly less efficient (5%?) than the best efficiency strategy tested of 15°.
- Line 249: perhaps worth stressing this point more strongly - given a strong uplift in efficiency for seasonal strategies (and a stronger uplift at 60° than at 30°) its not implausible that 60 seasonal would be the most efficient strategy in UKESM.
- Figure 3: I wonder if it would be worth adding some more data to the plot to better explore the drivers. Adding at least the model's pre-industrial range might help to inform the discussion around lines 143-144 of the expected changes under warming.
Citation: https://doi.org/10.5194/egusphere-2024-1565-RC2 -
RC3: 'Comment on egusphere-2024-1565', Anonymous Referee #3, 16 Jul 2024
Summary :
This work used UKESM1 to perform a set of SAI SO2 injection strategies, designed to be easily compared with previous work performed with CESM2-WACCM. A single simulation was performed, following the experimental design of the ARISE-SAI experiment, previously applied within CESM2; which uses a feedback algorithm to meet multiple surface objectives by controlling yearly SO2 input at 15 N, 15S, 30N and 30S. The additional four simulations performed sought to isolate the impacts of the latitudinal location of injection on the climate response to SAI and separately simulated injections at the equator, 15N/15S, 30N/30S, and 60N/60S; fixing the altitude and season of injection. Taken together, these five simulations:
- Enabled comparison between the climate response to SAI in UKESM1 and CESM2-WACCM to understand the ability to extrapolate conclusions drawn from single model studies
- Enabled understanding of how injection latitude impacts the climate response to SAI in UKESM1
In general the response in UKESM was found to be similar to that of CESM2, apart from marked differences in the shifting of the ITCZ, motivating future study of the differences in the mechanistic controls on this feature between models. Comparing between the fixed latitude injections in UKESM, injection at 30N ad 30S required the smallest amount of injected SO2 to reach set targets, as compared with other injection locations, and minimized undesirable side effects. Injection at the equator was found to be the least efficient with the largest undesired modifications to features like ozone and dynamics. Together these findings motivate the 30N/30S injection strategy as part of the GeoMIP G6-1.5K-SAI simulations, and suggest that the robustness climate impacts, like shifting of the ITCZ would benefit from an intermodel study.
General Comments:
In general, the narrative of this work could be strengthened by providing more introductory context. In particular, providing greater background pertaining to the feedback algorithm and how that simulation is important in the context of this work (e.g. the arise simulation does not isolate the latitude of injection, so why are you using it?). It is eventually clear that this work uses the ARISE simulation to ground comparison between CESM and UKESM, as well as to motivate the use of the 30N/30S injection for use in GeoMIP, but this narrative could be clearer from the introduction to set the reader up. Line by line suggestions to this point are listed below under “minor comments.”
Additionally, language in the abstract is a slightly misleading; specifically “we demonstrate that the 30N+30S strategy has, on balance, the least negative side effects…” while in the context of the remainder of this paper the reader can understand this line, it overstates the depth of the analysis of the side effects. If the word count allows this should cite the specific effects analyzed and could be reworded to emphasize this conclusion motivates future simulation, versus future deployment, strategy.
Minor Comments:
Line 20: “enables a control …” – should be more descriptive of the algorithm applied, and more clear that this is a PID controller seeking to optimize a set of multiple objectives
Line 31: “The contribution of the aerosol lifetime effects was found to be five to six times larger than that of the water vapor feedback.” Unclear exactly what the aerosol lifetime is contributing to; and exactly how. In general it would be useful for the author to define “efficiency” as used in this paper – e.g. cooling per tg SO2 injected? Total reduction of the TOA imbalance per Tg SO2 injected? This should be directly defined in the introduction
Line 40: How did Bednarz et al. vary cooling? Was it an idealized study or does this equate to a reduction of injected SO2?
Line 42: The paragraph beginning with “looking into the future..” Felt out of place as is. This seemed disjointed from the logic flow and felt it could be better situated in the conclusions. It could be more useful to frame the introduction to motivate the use of the ARISE setup in UKESM as a way to compare the complex feedback simulation to the other four simulations – if one were to frame the paper this way, it would be important to emphasize the similarity of the conclusions between the ARISE and fixed latitude injection scenarios (e.g. does arise essentially recommend a 30N/30S injection? Are the climate outcomes about the same?).
Line 72-75: “unlike ARISE-SAI …” : When defining the simulations performed it would be useful to clarify what season these injections are performed during; or rather are they continuous? Might considering adding injection timing as a column in Table 1.
Line 83: “…anomalous AOD...” the meaning of this is relatively unclear. Is this the AOD anomaly taken as the deseasonalized and detrended AOD ?
Line 90: Would be useful to formally define efficiency.
Line 99: Would be interesting to include the values for the differences in volume-mean or mass-mean particle size between injection locations!
Line 136: There is an expected relationship between ITCZ and the hemispheric temperature difference, this relationship is shown by the dashed line in figure 3, with a calculated slope of X. It was challenging to follow this paragraph, because to the reader it says the ITCZ is controlled by the hemispheric temperature difference, but actually it isn’t … I would try to be more direct in this paragraph and lead with the many controls on the ITCZ, which explain why the linear correlation is a relatively bad fit. It would also be useful to compare the SAI-ARISE CESM2 results to the UKESM results here – especially given this a main point in the conclusion. Could include in Figure 3 to give context to the reader.
Figure 4/5/6: When using a two sided ttest on spatially resolved data, the p value must be corrected for the false discovery rate. Refer to: http://dx.doi.org/10.1175/BAMS-D-15-00267.1
Line 169: “… causes a weakening of the tropospheric jets, …” I was curious what the seasonal (DJF, MAM, SON etc) figures look like for the zonal wind changes. It’s not immediately clear what season the injections were performed during, however, if one expects to strengthen the stratospheric polar vortex, based on previous studies, this should delay the springtime breakdown of the vortex, resulting in a poleward shift of the eddy driven tropospheric jet. I wonder if the weakening seen here, might appear to be a poleward shift if parsed by season vs an annual average change for the given period.
Conclusions:
+ Might consider a discussion of how season of injection (e.g. performing the same simulations in a different season) might change the outcomes.
+ could be strengthened by a discussion of the known differences in UKESM and CESM2 alongside a more detailed discussion of the differences in response between the models. This was a major motivation of the investigation and does not feel fully fleshed out.
Citation: https://doi.org/10.5194/egusphere-2024-1565-RC3 - AC1: 'Comment on egusphere-2024-1565', Matthew Henry, 08 Oct 2024
Status: closed
- RC1: 'Comment on egusphere-2024-1565', Anonymous Referee #1, 16 Jul 2024
-
RC2: 'Comment on egusphere-2024-1565', Anonymous Referee #2, 16 Jul 2024
This study compares the climate response to stratospheric aerosol injection in the earth system model UKESM under varying injection latitude. It is a clear presentation of an important new set of simulations, which will be a useful reference going forwards. It makes a valuable contribution in clarifying the trade-offs between different possible injection latitudes. I would like to see it published given some minor revisions.
General comments:
In the abstract, the statement "we demonstrate that the 30N+30S strategy has, on balance, the least negative side effects" is too strong. There is no comprehensive assessment of the side effects or their impacts presented here, and their results suggest 60° injection has smaller side-effects on some metrics. I suggest the authors make a more limited statement, or at least specify on what metric they are making this judgement.
It would be good to include maps of the global surface climate response, at least for temperature and precipitation, under the different scenarios. This could be in Supplementary if article length is an issue. While not critical to the arguments being made here, these simulations will be a useful reference point for others and I suspect many will want to see the full spatial picture.
The zonal line plots (fig 1, fig 2a,c,d) don't currently indicate the ensemble spread. This information would be useful to include (perhaps via a new figure in Supplementary if it is too cluttered here), since it’s not currently obvious where the differences between scenarios are robust to internal variability.
Figure 2C: there are two aspects here which could be a little misleading, and which adding some more subplots (or a new fig in supplementary) could resolve. First, the equatorial injection strategy looks as though it strongly undercools the arctic relative to the other strategies, but this is partly an artifact of the controller failing to meet the 1.5K target for this strategy. Perhaps also plotting the residuals for each strategy relative to the baseline global mean temperature it actually achieves (e.g. relative to a ~1.75K world for equatorial injection) would show more clearly the relative differences in latitudinal pattern of cooling between the strategies. Second, plots like these obscure the lower latitude structure because of the strong feedback driven arctic amplification of residuals (as already discussed in the text). Perhaps it would be worth also plotting the residual warming scaled by the local warming from pre-industrial at the baseline?
Minor points:
- Line 95: "The larger injection rate is thus due to the lower efficacy of tropical forcing (Kang and Xie, 2014) and to the confinement of aerosols inside the tropical pipe, enhancing the formation of larger aerosols which sediment faster." A third potential contribution is that local forcing is sub-linear in local AOD, so concentrating AOD in one region reduces the global mean forcing.
- Figure 2b: define the error bars here. Are they the ensemble spread?
- Line 99: "The larger injection rates for 60N+60S, on the other hand, arise due to faster 100 removal of aerosols when injected near the descending branch of the Brewer Dobson Circulation (BDC)". There are other potential contributions here; latitudinal even-ness of the AOD distribution (as above), and latitudinal variation in insolation, underlying albedo, and feedbacks
- Label text is too small on the multi-panel figures (4,5,6)
- Line 227-228 perhaps rephrase to make it clear that it is in CESM2 that the high latitude runs were seasonal and lower altitude.
- Line 238: "The main takeaway is that the 30N+30S strategy is one of the most efficient strategies in terms of amounts of SO2 needed". Perhaps better to give a rough statement of the relative efficiencies. There aren't many strategies compared, so "one of the best" is not all that meaningful. My interpretation is 30° is only slightly less efficient (5%?) than the best efficiency strategy tested of 15°.
- Line 249: perhaps worth stressing this point more strongly - given a strong uplift in efficiency for seasonal strategies (and a stronger uplift at 60° than at 30°) its not implausible that 60 seasonal would be the most efficient strategy in UKESM.
- Figure 3: I wonder if it would be worth adding some more data to the plot to better explore the drivers. Adding at least the model's pre-industrial range might help to inform the discussion around lines 143-144 of the expected changes under warming.
Citation: https://doi.org/10.5194/egusphere-2024-1565-RC2 -
RC3: 'Comment on egusphere-2024-1565', Anonymous Referee #3, 16 Jul 2024
Summary :
This work used UKESM1 to perform a set of SAI SO2 injection strategies, designed to be easily compared with previous work performed with CESM2-WACCM. A single simulation was performed, following the experimental design of the ARISE-SAI experiment, previously applied within CESM2; which uses a feedback algorithm to meet multiple surface objectives by controlling yearly SO2 input at 15 N, 15S, 30N and 30S. The additional four simulations performed sought to isolate the impacts of the latitudinal location of injection on the climate response to SAI and separately simulated injections at the equator, 15N/15S, 30N/30S, and 60N/60S; fixing the altitude and season of injection. Taken together, these five simulations:
- Enabled comparison between the climate response to SAI in UKESM1 and CESM2-WACCM to understand the ability to extrapolate conclusions drawn from single model studies
- Enabled understanding of how injection latitude impacts the climate response to SAI in UKESM1
In general the response in UKESM was found to be similar to that of CESM2, apart from marked differences in the shifting of the ITCZ, motivating future study of the differences in the mechanistic controls on this feature between models. Comparing between the fixed latitude injections in UKESM, injection at 30N ad 30S required the smallest amount of injected SO2 to reach set targets, as compared with other injection locations, and minimized undesirable side effects. Injection at the equator was found to be the least efficient with the largest undesired modifications to features like ozone and dynamics. Together these findings motivate the 30N/30S injection strategy as part of the GeoMIP G6-1.5K-SAI simulations, and suggest that the robustness climate impacts, like shifting of the ITCZ would benefit from an intermodel study.
General Comments:
In general, the narrative of this work could be strengthened by providing more introductory context. In particular, providing greater background pertaining to the feedback algorithm and how that simulation is important in the context of this work (e.g. the arise simulation does not isolate the latitude of injection, so why are you using it?). It is eventually clear that this work uses the ARISE simulation to ground comparison between CESM and UKESM, as well as to motivate the use of the 30N/30S injection for use in GeoMIP, but this narrative could be clearer from the introduction to set the reader up. Line by line suggestions to this point are listed below under “minor comments.”
Additionally, language in the abstract is a slightly misleading; specifically “we demonstrate that the 30N+30S strategy has, on balance, the least negative side effects…” while in the context of the remainder of this paper the reader can understand this line, it overstates the depth of the analysis of the side effects. If the word count allows this should cite the specific effects analyzed and could be reworded to emphasize this conclusion motivates future simulation, versus future deployment, strategy.
Minor Comments:
Line 20: “enables a control …” – should be more descriptive of the algorithm applied, and more clear that this is a PID controller seeking to optimize a set of multiple objectives
Line 31: “The contribution of the aerosol lifetime effects was found to be five to six times larger than that of the water vapor feedback.” Unclear exactly what the aerosol lifetime is contributing to; and exactly how. In general it would be useful for the author to define “efficiency” as used in this paper – e.g. cooling per tg SO2 injected? Total reduction of the TOA imbalance per Tg SO2 injected? This should be directly defined in the introduction
Line 40: How did Bednarz et al. vary cooling? Was it an idealized study or does this equate to a reduction of injected SO2?
Line 42: The paragraph beginning with “looking into the future..” Felt out of place as is. This seemed disjointed from the logic flow and felt it could be better situated in the conclusions. It could be more useful to frame the introduction to motivate the use of the ARISE setup in UKESM as a way to compare the complex feedback simulation to the other four simulations – if one were to frame the paper this way, it would be important to emphasize the similarity of the conclusions between the ARISE and fixed latitude injection scenarios (e.g. does arise essentially recommend a 30N/30S injection? Are the climate outcomes about the same?).
Line 72-75: “unlike ARISE-SAI …” : When defining the simulations performed it would be useful to clarify what season these injections are performed during; or rather are they continuous? Might considering adding injection timing as a column in Table 1.
Line 83: “…anomalous AOD...” the meaning of this is relatively unclear. Is this the AOD anomaly taken as the deseasonalized and detrended AOD ?
Line 90: Would be useful to formally define efficiency.
Line 99: Would be interesting to include the values for the differences in volume-mean or mass-mean particle size between injection locations!
Line 136: There is an expected relationship between ITCZ and the hemispheric temperature difference, this relationship is shown by the dashed line in figure 3, with a calculated slope of X. It was challenging to follow this paragraph, because to the reader it says the ITCZ is controlled by the hemispheric temperature difference, but actually it isn’t … I would try to be more direct in this paragraph and lead with the many controls on the ITCZ, which explain why the linear correlation is a relatively bad fit. It would also be useful to compare the SAI-ARISE CESM2 results to the UKESM results here – especially given this a main point in the conclusion. Could include in Figure 3 to give context to the reader.
Figure 4/5/6: When using a two sided ttest on spatially resolved data, the p value must be corrected for the false discovery rate. Refer to: http://dx.doi.org/10.1175/BAMS-D-15-00267.1
Line 169: “… causes a weakening of the tropospheric jets, …” I was curious what the seasonal (DJF, MAM, SON etc) figures look like for the zonal wind changes. It’s not immediately clear what season the injections were performed during, however, if one expects to strengthen the stratospheric polar vortex, based on previous studies, this should delay the springtime breakdown of the vortex, resulting in a poleward shift of the eddy driven tropospheric jet. I wonder if the weakening seen here, might appear to be a poleward shift if parsed by season vs an annual average change for the given period.
Conclusions:
+ Might consider a discussion of how season of injection (e.g. performing the same simulations in a different season) might change the outcomes.
+ could be strengthened by a discussion of the known differences in UKESM and CESM2 alongside a more detailed discussion of the differences in response between the models. This was a major motivation of the investigation and does not feel fully fleshed out.
Citation: https://doi.org/10.5194/egusphere-2024-1565-RC3 - AC1: 'Comment on egusphere-2024-1565', Matthew Henry, 08 Oct 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 379 | 134 | 40 | 553 | 18 | 20 |
- HTML: 379
- PDF: 134
- XML: 40
- Total: 553
- BibTeX: 18
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1