Impact of solar geoengineering on wildfires in the 21st century in CESM2/WACCM6

Tang, Wenfu; Tilmes, Simone; Lawrence, David M.; Li, Fang; He, Cenlin; Emmons, Louisa K.; Buchholz, Rebecca R.; Xia, Lili

doi:https://doi.org/10.5194/acp-23-5467-2023

Articles | Volume 23, issue 9

https://doi.org/10.5194/acp-23-5467-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/acp-23-5467-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 23, issue 9

Research article

|

16 May 2023

Research article |

| 16 May 2023

Impact of solar geoengineering on wildfires in the 21st century in CESM2/WACCM6

Wenfu Tang, Simone Tilmes, David M. Lawrence, Fang Li, Cenlin He, Louisa K. Emmons, Rebecca R. Buchholz, and Lili Xia

Download

Final revised paper (published on 16 May 2023)
Supplement to the final revised paper
Preprint (discussion started on 16 Jan 2023)
Supplement to the preprint

Interactive discussion

Status: closed

RC1:
'Peer Review: Impact of Solar Geoengineering on Wildfires in the 21st Century in CESM2/WACCM6', Anonymous Referee #1, 15 Feb 2023

This study investigates whether and how solar geoengineering affects wildfire impacts in CESM2 simulations by examining the major contributing factors to wildfire activity and how they change in response to geoengineering. The authors find that, in most regions, solar geoengineering causes reductions in wildfire activity, largely due to reduced surface temperatures and increased relative humidity and soil moisture, despite reduced precipitation.
Overall, this is a sound study with meaningful results, and reading it was interesting and informative; wildfires are an important aspect of the climate system, and a study about how solar geoengineering does or does not offset possible changes under global warming is an important contribution to the literature. However, the manuscript needs some work before it can be published. I have two major criticisms. Firstly, the authors need to clarify their methods in some areas; it is not always clear how the authors computed certain numbers presented in their results, and the authors should explain more clearly whether they are averaging over a certain period or how they accounted for certain factors when computing statistics. Secondly, many of the figures are unpolished or incomplete, with missing uncertainties, ensemble spread, or statistical significance; inconsistent color schemes; frustrating or confusing color scales; and captions that do not fully explain how data are calculated. Correcting these issues would improve the clarity, ease of reading, and repeatability of the study. Specific comments and technical corrections are provided in the supplement, and I recommend that the study be accepted with minor revisions so that these findings can be published.

Citation: https://doi.org/10.5194/acp-2022-834-RC1
- AC1: 'Reply on RC1', Wenfu Tang, 03 Apr 2023
  
  The comment was uploaded in the form of a supplement: https://acp.copernicus.org/preprints/acp-2022-834/acp-2022-834-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/acp-2022-834-AC1
RC2:
'Comment on acp-2022-834', Anonymous Referee #2, 14 Mar 2023

The authors seek to quantify how geoengineering might be expected to affect wildfire occurrence, extent, and carbon emissions towards the end of the 21st century. They compare four different scenarios: two of the world without any geoengineering (one "cooler" - SSP2-4.5 - and one "warmer" - SSP5-8.5), one in which idealized (solar shade) geoengineering is applied to force global mean radiative forcing (RF) in the "warmer" scenario to follow that in the "cooler" scenario, and one in which stratospheric sulfates are emitted to achieve the same objective. To do so, they use the fully-coupled global chemistry-climate model WACCM, which includes an interactive treatment of wildfires. They evaluate wildfire outcomes principally through the metrics of total burned area and total carbon emissions, finding that - by both metrics - geoengineering "overcompensates" for the increase expected under SSP5-8.5, bringing total wildfires below the amount projected by SSP2-4.5. They also note some smaller differences in outcomes between the two geoengineering scenarios, and that there is some nonuniformity in the spatial distribution of outcomes.

The question asked by the authors is interesting, important, and timely, and the methods applied are appropriate. WACCM is a world-renowned climate model and the scenarios chosen are both well described and reasonably well known, making them more relevant to the community at large. The paper is incremental in nature, as it does not produce any astonishing new findings, but the content is nonetheless novel. The conclusions drawn are also supported by the data produced.

With this in mind, I have no major concerns regarding the content or framing of this paper. I have listed below some minor tehcnical and presentation concerns. Once these are addressed, I believe the manuscript will be appropriate for publication in ACP.

Minor technical issues

The authors appropriately caveat that the estimates of geoengineering's impacts are calculated based on small, variably-sized ensembles. In one of the central figures of the paper (Figure 6), correlations are presented between each "fire" variable (burned area and fire carbon emissions) and several "driver" variables (e.g. surface temperature). Values which are not significant at the p <= 0.1 level are not shown. However, it is not clear how significance is evaluated. Similarly, I am concerned by the differing numbers of simulations between all of the various scenarios. Both are important to this manuscript, and I recommend that the details be not only listed but also discussed in the manuscript main text.

With regards to the question of the number of ensemble members used, I would recommend that the authors perform an analysis where a consistent ensemble size is used. Currently the manuscript appears to be biased by this discrepancy; for example, in Figure 2, the full range of burned areas between ensemble members is shown. However, smaller ensemble sizes will generally result in smaller ranges. Similarly, it would be useful to know how these ensemble members were initialized (e.g. in the geoengineering cases where there are 2 ensemble members, were they initialized with the same initial conditions as 2 of the 5 members from SSP5-8.5?).

On lines 209 to 216, the authors analyze how the estimated total burned area differs between SSPs. One specific claim stands out to me. The authors state that the changes under SSPs 8.5 (~20%) and 7.0 (~10%) are the largest, whereas those in the others scenarios are "relatively small". I would first request that qualitative claims such as this one be replaced with quantitative statements where possible, as I am unsure what "relatively small" means. However in this specific case I am also concerned that this conclusion may be driven more by the form of the evaluation than by a meaningful difference. Visually, it is unclear from Figure 2 that there is any significant difference between the trajectories taken by SSP1-2.6 and SSP3-7.0; however, these are also the two cases that had just one ensemble member. Providing simple quantitative statements (e.g. "increases of less than 4%" is better than "relatively small") would help, but better yet - as discussed above - would be a quantitative discussion of how confident the authors can be when comparing results from a single ensemble member to multi-member averages.

Minor presentation issues

In several figures the color scales used are either confusing or misleading. The two most problematic examples are Figures 5 and 6. In Figure 5, the same color scale is used for 5 different quantities. Worse, for 4 of the quantities, "yellow" (i.e. the central value) is zero; however, for temperature, dark red is zero. This discrepancy is visually confusing. I would suggest using, at the very minimum, a different, one-sided color scale for the temperature change.

Similarly, the color scale used in Figure 6 is visually misleading. The extremes of the scale not only change from panel to panel, but the darkest shade changes from being a positive correlation of ~+0.5 (.e.g. 6a) to a negative correlation of -0.6 (6i). The color white also changes - while usually being 0, sometimes it is not (e.g. 6g, 6h). A consistent, two-sided color scale - for example that used in Figure 5 - running from -1 to +1 (or perhaps -0.6 to +0.6) would greatly help comprehension. Furthermore it is very difficult to read some of the values for the darkest backgrounds (see e.g. the EQAS/SSP2-4.5 value in 6i). Strangely the presentation in Figure S5 is much improved, although a standing issue with this manuscript is that a color scale with both red and green present is likely to cause issues for those with colorblindness.

Why is the p-value of significance different between Figure 6 and Figure S5?

The caption for Figure 8 mentions that the shaded region is relevant to a comparison between delta-BA and delta-C emis, but that does not seem correct based on the axes. I suspect this was incorrectly copied from the caption for Figure 9?

Lines 495-504: This phrasing is incorrect. The comparison being made is between strengths of correlation, not magnitudes of effects. Saying that "impacts of the shown variables (...) on burned area are in general stronger than their impacts on fire carbon emissions" implies that a conclusion is being drawn regarding the size of the impacts, not their degree of correlation with other variables. I would suggest rephrasing for clarity.

Citation: https://doi.org/10.5194/acp-2022-834-RC2
- AC2: 'Reply on RC2', Wenfu Tang, 03 Apr 2023
  
  The comment was uploaded in the form of a supplement: https://acp.copernicus.org/preprints/acp-2022-834/acp-2022-834-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/acp-2022-834-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Wenfu Tang on behalf of the Authors (03 Apr 2023) Author's response Author's tracked changes Manuscript

ED: Publish as is (15 Apr 2023) by Dominick Spracklen

AR by Wenfu Tang on behalf of the Authors (17 Apr 2023)

Download

The requested paper has a corresponding corrigendum published. Please read the corrigendum first before downloading the article.

Article (11814 KB)
Full-text XML

Short summary

Globally, total wildfire burned area is projected to increase over the 21st century under scenarios without geoengineering and decrease under the two geoengineering scenarios. Geoengineering reduces fire by decreasing surface temperature and wind speed and increasing relative humidity and soil water. However, geoengineering also yields reductions in precipitation, which offset some of the fire reduction.