the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Sep 2026
Competing roles of anthropogenic aerosols and greenhouse gases in June–August baroclinicity and storm-track trends
Abstract. Northern Hemisphere summer (June–August) storm tracks have been extensively studied over recent decades, with evidence for weakening trends in parts of the Northern Hemisphere and strengthening trends in the Southern Hemisphere. However, the physical drivers of these trends, and their strong regional variability, are still debated. Here, we investigate recent changes in mid-latitude baroclinicity by analysing trends in Eady Growth Rate and baroclinic mean available potential energy (MAPE) in ERA5 reanalysis data and CMIP6 single-forcing large-ensemble experiments over both the industrial and satellite eras. We further decompose baroclinic MAPE and Eady Growth Rate into their physical components and compare the baroclinicity trends with trends in storm-track intensity. We find that changes in the meridional temperature gradient provide the dominant contribution to recent baroclinicity trends. Significant regional asymmetries are identified between the North Pacific and North Atlantic baroclinicity trends, with the strongest recent weakening occurring over the North Pacific storm-track region. Greenhouse-gas forcing produces a robust positive Southern Hemisphere baroclinicity trend and a mixed weakening response in the Northern Hemisphere, while aerosol perturbations generate substantial regional responses in both hemispheres and partially offset the greenhouse-gas-driven baroclinicity signal. In particular, changes in aerosol emissions during the satellite era are associated with enhanced negative baroclinicity trends over the North Pacific and contribute to the emergence of the greenhouse-gas-driven storm-track response. These results highlight the competing roles of greenhouse gases and aerosols in shaping recent mid-latitude circulation changes, and demonstrate that both forcings must be considered when interpreting historical storm-track trends and projecting their future behaviour.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Weather and Climate Dynamics.
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
(97330 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-5066', Anonymous Referee #1, 21 Oct 2026
-
RC2: 'Comment on egusphere-2026-5066', Anonymous Referee #2, 22 Oct 2026
Reviewer comments to: Cohen et al., submitted for publication in Weather and Climate Dynamics
The manuscript investigates the trends in baroclinicity of both hemispheres during the JJA season. The authors find that recent baroclinicity trends are dominated by meridional temperature gradient trends. The authors also use CMIP6 single-forcing simulations to quantify the contribution from different anthropogenic and natural forcings to the baroclinicity trends.
As the authors note, the storm track trends and the underlying mechanisms have received a lot of attention in recent years. Therefore, it is important to provide additional dynamical insights on top of the understanding established from previous work. I find that the current version of the manuscript, unfortunately, does not do this job perfectly. Also, I am concerned about using the MAPE formula locally, which is derived for the global mean energy. Below are my suggestions for improvements.
Major Comments
1. While there is potential value in analyzing the baroclinicity trends, I think the study has to be better motivated, and a more accurate understanding of previous work is needed throughout the paper.
1.1. Chemke and Coumou (2024) investigated the trends in baroclinicity (see their Fig. 3b) and offer a similar decomposition to the results in this paper. This previous result, which is highly relevant, is not mentioned. It also seems that Chemke and Yuval (2023) is incorrectly cited instead of Chemke and Coumou (2024) throughout the paper.
1.2. Kang et al. (2024b) show the relative importance of greenhouse gases, aerosols, and natural forcing for the circulation trends explicitly (see their Figs. 1–3). Their work also explains the mechanism why the North Pacific and North Atlantic storm track responses to aerosol forcings are different. The claims in lines 52–54 and lines 71–72 are not necessarily true.
1.3. Chemke et al. (2022), which is not referenced, investigated the SH storm track trends and showed that baroclinicity trends do not well explain the storm track trends in reanalysis as well as the reanalysis-model difference. This previous result has to be discussed. Examining SH baroclinicity trends and linking them to storm-track trends requires more attention than is currently given.
1.4. Kang et al. (2024a) show that there is a huge spread in storm track trends across reanalysis products. The robustness of the SH baroclinicity trends across different reanalysis products has to be quantified/discussed.
1.5. Putting together 1.1–1.4, the authors should present in detail what has been done in previous work and motivate the current work, and also connect the baroclinicity trend analyses to the results of previous work.
1.6. Lines 62–66. The references and the descriptions here are inaccurate. The works by Mbengue and Schneider are idealized modeling studies that examine the storm track response to climate-change-like temperature perturbations and do not discuss JJA baroclinicity. Coumou et al. (2015) also do not study the impact of greenhouse gases. A more relevant study the authors may want to cite in those lines is Shaw et al. (2018), which studies CO₂ impacts. In Shaw et al. (2018), land warming (i.e., direct CO₂ effect) dominates the storm track weakening in NH JJA. In this regard, the following sentences on the role of SST warming are inaccurate.
1.7. Lines 66–71. There are multiple studies that examine how aerosols impact circulations, but none are mentioned here. For example, Ming et al. (2011) investigate the large-scale circulation impacts of aerosols and analyze the Eady growth rate (see their Fig. 13). Dong et al. (2022) examine the trends in temperature structure from aerosols.
Ming, Y., Ramaswamy, V., & Chen, G. (2011). A model investigation of aerosol-induced changes in boreal winter extratropical circulation. Journal of Climate, 24(23), 6077-6091.
1.8. Putting together 1.6 and 1.7, the paper should put more effort into correctly describing and explaining the physical mechanisms related to each anthropogenic forcing. The current manuscript falls short in doing so; hence, the additional physical insight the study is providing in the single-forcing simulation section is not very clear.
2. The authors calculated the local Eady growth rate and MAPE to quantify local baroclinicity. These quantities may not necessarily represent baroclinicity at each grid point.
2.1. The Eady growth rate is theoretically derived from a model with zonal symmetry. It is, however, widely used in the literature locally, as done in this paper. This is practical, but physically inconsistent. The authors should at least include a cautionary note on this physical inconsistency.
2.2. The MAPE used here is also for the global mean or hemispheric mean. The works cited by the authors calculated one MAPE value from the whole hemisphere. The hemispheric integrated MAPE is useful as it explains the hemispheric mean storm track intensity changes during the seasonal cycle, under climate change, and across different models. This connection between hemispheric MAPE and storm track intensity is built on dynamical theories (Schneider and Walker, 2006). I doubt that the same findings apply locally. Yet, the same formulation is used locally. I think MAPE in this formulation can not be used as a measure of local baroclinicity. Moreover, there are studies that attempted to quantify local baroclinic energy (Novak and Tailleux, 2018; Gertler et al., 2023). The authors should justify the use of MAPE in Eq. (5). Please also note for the readers that Eq. (5) is an approximate (linearized) formula.
Schneider, T., & Walker, C. C. (2006). Self-organization of atmospheric macroturbulence into critical states of weak nonlinear eddy–eddy interactions. Journal of the Atmospheric Sciences, 63(6), 1569-1586.
Novak, L., & Tailleux, R. (2018). On the local view of atmospheric available potential energy. Journal of the Atmospheric Sciences, 75(6), 1891-1907.
Gertler, C. G., O’Gorman, P. A., & Pfahl, S. (2023). Moist available potential energy of the mean state of the atmosphere and the thermodynamic potential for warm conveyor belts and convection. Weather and Climate Dynamics, 4(2), 361-379.
Specific comments
- Lines 56–58. It is confusing to cite Lehmann et al. (2014) and Gertler and O’Gorman (2019) at the end of this sentence because these are the studies that paid attention to the baroclinicity and storm tracks.
- Lines 151–152. Do native model grids mean CMORized grids? It would be better to quantify model and reanalysis using the same vertical levels, if possible.
- Line 167. What does stable results mean here?
- Lines 217–219. This is unclear. Please revise. Should an anomaly be defined to calculate linear trends?
- Lines 270–272. None of these studies investigated North Pacific baroclinicity. Please revise.
- Lines 277–279. The consistency with the Eurasian jet trend is unclear.
Citation: https://doi.org/10.5194/egusphere-2026-5066-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 332 | 291 | 62 | 685 | 53 | 62 |
- HTML: 332
- PDF: 291
- XML: 62
- Total: 685
- BibTeX: 53
- EndNote: 62
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study explores the role of different forcings (green house gas (GHG), aerosol (AER), and natural (NAT)) in shaping the trends in June-August baroclinicity and storm tracks in the historical period, including the full historical period as well as the more recent satellite era. Recent studies have suggested that during the satellite era, the Northern Hemisphere (NH) storm tracks have significantly weakened, while the Southern Hemisphere storm track has significantly increased in activity. Nevertheless, the role played by different forcings are still not clear. The authors used single forcing experiments to explore the impact of the different forcings, and found that GHG and AER forcings play the most important role, with the impact of AER forcing sometimes acting to oppose or mask the impact of those of GHG, except in Northern Hemisphere summer during the satellite period in which the AER forcing works to enhance the impact of the GHG forcing. These results are of interest to the climate community. Nevertheless, I do have serious concerns about the current version of the manuscript which I will detail below.
Major concerns:
1. The main focus of this study is understanding the factors that gave rise to the observed changes in baroclinicity that had shaped the storm track trends during JJA in both the northern and southern hemispheres. In this regard, my feeling is that the authors should have paid more attention to recent studies that discussed the JJA storm track trends in both hemispheres, and discuss their findings in the context of these studies.
- For example, Chemke and Coumou (2024, which the authors cited) already showed that the NH storm track decrease during the satellite period can be simulated by CMIP6 class models, with GHG and AER forcings both contributing to the decrease, consistent with the results the authors presented here.
- On the other hand, Kang et al (2024a, which the authors also cited) showed that CMIP6 coupled models were unable to simulate the strong storm track increasing trend in the Southern Hemisphere, and their results suggested that observed tropical Pacific and southern ocean SST changes which were different from those simulated by CMIP6 models may play important roles in this discrepancy, with AMIP simulations being more consistent with reanalysis trends. Given the models used in this study appear to be CMIP6 class coupled models, I suppose that they also failed to simulate the strong SH storm track trend found in reanalysis? ERA5 storm track trends were not shown for comparison. Given the model-reanalysis discrepancy in the SH, the authors' SH results should be discussed in light of these discrepancies. If the models are not able to simulate the observed trends, what do the results of the factor analysis really inform us about what caused the observed trends?
2. Given that the main focus is to understand what causes the trends in baroclinicity, my feeling is that the authors have not sufficiently discussed how changes in aerosol might have led to changes in baroclinicity.
- They displayed the aerosol trends (Fig. 9-11), and showed the SST trends (Fig. 12), but it is not clear to me how the aerosol trends can give rise to the SST trends
- In addition, the authors suggested that changes in the SST gradients are consistent with the changes in baroclinicity, but this is very difficult to see when only the SST trends and not the SST gradient trends are shown. In fact, the authors should probably start by showing whether the observed trends in SST gradients are consistent with the trends in baroclinicity found in ERA5 data.
- Given the aerosol trends in Asia during the satellite era is largely similar to the trends in the full historical period but at a larger amplitude (Fig. 9, 10), it is not clear to me why the baroclinicity trend induced by aerosol forcing in the North Pacific over the two periods are of opposite signs. The authors need to explain this since this is one of the key results of the study.
3. The authors showed a lot of information (over 150 figure panels shown in the main section, and over 200 panels in the appendices). Many of the panels do not have it's own sub-figure number (e.g. a, b, c), making it very difficult for readers to figure out exactly which panel the authors were referring to in many of the discussions. In addition, in my opinion the discussion is not well organized. Two different periods and the two hemispheres are interwined in the discussions, making it very hard for readers to keep in mind all the contrasting results. To me, perhaps a reorganization of the discussions of the model results should be made. The authors should have subsections in discussing model results. First a subsection for NH in the satellite period, contrasting model results with reanalysis data, and discussing how each factor in the different model experiments contribute to changes in baroclinicity. The second subsection should focus on NH for the entire historical period, contrasting that with the satellite period. The third subsection should focus on SH during the satellite period, while the fourth subsection on SH during the entire historical period. To me, this might make the discussions of the results more structured and focused, and allow readers to focus on one period and one hemisphere in each subsection and make it easier for readers to catch and better understand the main results.
4. To me, much of the discussions of the model results are too qualitative in nature, largely consisted of comparing (noisy) patterns found in different figure panels by eye. The authors did start their discussion of the reanalysis trends in a more quantitative way, displaying correlations among different factors in Table 1. However, the model-reanalysis comparisons are largely qualitative in nature - perhaps spatial correlations between model and reanalysis fields should be quantified to see how well models simulate trends found in reanalysis? Some quantitative results are shown in Appendix C, but these results are not really discussed in the text. Given that spatial patterns are frequently noisy, and often only scattered "significant" regions or points are found, it may be useful to better quantify trends by discussing more spatially aggregated results, such as those shown in Appendix C? Also, for some reason, the reanalysis trends for the SH are not shown in Figs. C5-6. It also didn't help that the reanalysis spatial trends (Fig. 3-4) are shown in a different map projection (and shifted by ~90 degrees longitude) from those of the model trends (Fig. 5 and later), making it very difficult even to qualitatively compare model trends to reanalysis trends. The color scales used for reanalysis trends (Fig. 3-4) are also different from those used for similar model trends (Fig. 5-8), again making it difficult to compare the two.
5. Lines 58-60 presented two key questions of this study. It seems to me that the major part of the discussions in the paper addressed key question 1 (how and why baroclinicity has changed), and rather limited effort was devoted to the second question (how this (baroclinicity change) has shaped recent storm track trends). As mentioned above, as far as I can see, reanalysis storm track trends were not even shown to be compared with model trends, and the (potential) inconsistency in the SH winter storm track trend was not discussed. In my opinion, MAPE trends and EKE trends seem to correspond better with each other - which is not surprising due to the energetic relationships between EKE and MAPE, and this should be more quantitatively discussed (e.g. comparing percentage change in MAPE with that in EKE for each hemisphere). This would strengthen the discussion of the second key question.
Other comments:
i. Line 13: "partially offset the greenhouse-gas driven baroclinicity signal" - based on Fig. 5-8, this is really only true for the SH. For the NH, Fig. 7-8 suggests that the aerosol effect dominates over the greenhouse-gas signal especially over the Pacific.
ii. Line 63: "enhanced warming at high latitudes": this mainly occurs in the NH. In the SH, baroclinicity increases because of increase in upper level meridional temperature gradient due to enhanced tropical warming and lower stratospheric cooking - I don't think this was ever mentioned?
iii. Lines 112-113: Since aerosol changes are central to this study, it may be useful for the authors to mention how aerosol trends are simulated in these studies, perhaps with a table or in an appendix or supplemental material.
iv. Lines 152: Can the authors please confirm that these were computed using the native model levels data instead of pressure level data?
v. Line 153: Baroclinic MAPE: Equation (5) is an approximation of the dry MAPE. With changes in the moisture content, it might be useful to discuss that the dry MAPE scales about the same as the moist MAPE under the seasonal cycle as well as global warming (O'Gorman 2010).
vi. Line 190: How many and which pressure levels?
vii. Line 196: Why EKE"vertical"? It's the EKE of the horizontal wind perturbations!
viii. Line 216: trends: Many of the "trends" are not linear - e.g. warming accelerated during the second half of the 20th century, and aerosol trends are clearly not linear. Some time series should be shown to see whether and how linear trends are appropriate.
ix. Line 221: t test: Are serial correlations taken into account in assessment of significance? Many climate modes such as the NAO have low frequency variability and has significant year-to-year serial correlation.
x. Line 255: R = 0.56: Perhaps it may be useful to separate into NH and SH? NH correlation is probably lower?
xi. Line 256: "significant negative trend" - this appears to refer to the zonal mean trends in Fig. A1a and b. Perhaps significance should be indicated in the zonal mean trends? Also, for MAPE in the NH, the zonal mean trend looks more like a poleward shift than a weakening?
xii. Lines 260-261: I'm not sure I can see a southward shift in the S. Pacific. The increase is much further south of the maximum that I would not call that a shift.
xiii. Line 318: Figs. 14 and 15 are cited before 9-13. Perhaps move this sentence to the end of this paragraph?
xiv. Lines 336-337: "The satellite era trends are largely in agreement with recent trends derived from satellite observations": Given that CanESM3 trends seem to be so different from the others, what do satellite obs say about this?
xv. Lines 344-345: It would be useful to show the response in the SST gradient to show how well that corresponds to the baroclinicity trend.
xvi. Lines 346-347: Again it is difficult to see - again showing SST gradient trends might make it easier to see.
xvii. Line 374: "close correspondence": the SST gradient changes seem rather subtle for the AER experiments, and there are positive and negative SST gradient changes over the North Pacific over the satellite era (Fig. 13 third row), but the baroclinicity changes are mainly negative, so it is not clear that the correspondence is really close to me.
xviii. Lines 401-419: As discussed above, this is the only discussion linking baroclinicity trend to storm track trend. In my opinion, a more quantitative discussion should be made, showing whether percentage increase/decrease in EKE can be linked to percentage increase/decrease in MAPE as in O'Gorman (2010).
xix. Line 435: Baroclinic MAPE is also influenced by the increase in temperature gradient in the upper troposphere under global warming.
xx. Line 450: Should discuss why the AER impact changes sign between the full historical period and the satellite period.
xxi. Lines 464-465: Again, in the satellite period in the NH, the trend induced by aerosol is in the same sign as the trend due to GHG, so this statement is not accurate.
xxii. Line 477: Fig. A1: Figures similar to A1 might be useful to display model trends.
xxiii. Lines 493-494: "opposite in sign": not true in the NH satellite period though.