the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Mar 2026
Dependence of Hadley circulation on the representation of tropical deep convection
Abstract. Parameterizations of subgrid-scale processes are an important source of bias in climate simulations and uncertainty in climate projections. Both the bias and the uncertainty can be reduced by resolving the most energetic small-scale processes based on the first principles via increasing model resolution, as oppose to relying on empirical or ad hoc (though physically motivated) parametrizations. We made a first step in this direction using a suite of time-slice and fixed-season experiments, specifically designed to deal with the heavy computational costs encountered when using a global 5 km model for climate and climate change simulations. We concentrate on the effect of the representation of deep convection for the Hadley Circulation (HC) and its response to warming.
We find that the time-mean state of HC is sensitive to the representation of deep convection: The HC is significantly weaker and the deep convection is significantly reduced in the case when deep convection is resolved instead of parametrized. Resolving convection leads to less heavy rain and less rain in the ascending branch of the HC – consistent with a weaker HC – but more total precipitation when taking also light rain into account. As a response to a 4 K global warming, the HC becomes broader, deeper, and weaker, no matter whether convection is resolved or parameterized. There is an indication that the weakening of HC is further intensified when deep convection is resolved. As a response to the warming, heavy rain and the rain associated with the HC ascending branch become stronger regardless of convection representation, while the total precipitation is more strongly enhanced when convection is resolved.
- Preprint
(1746 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1168', Anonymous Referee #1, 11 May 2026
-
RC2: 'Comment on egusphere-2026-1168', Anonymous Referee #2, 16 May 2026
Review of ‘Dependence of Hadley circulation on the representation of tropical deep convection’ by von Storch et. al. (WCD)
This manuscript asks an interesting question: how does resolving convection affect the strength of the Hadley cell. I was intrigued by the paper title and by one of the main results that the strength of the Hadley cell decreases when convection is resolved. However, sadly, the paper fell well short of my expectations.
The paper uses a say-what-you-see approach with no attempt made to understand mechanisms as to how or why the Hadley cell weakens if convection is resolved. Rather, a bunch of metrics are defined and used to quantify various aspects of the Hadley strength, width and height. These metrics are discussed laboriously over many figures that are presented mostly in box-and-whisker form (indeed, I suggest reducing the number of figures that are of the exact same format). From a scientific paper, I expect much more than just stating some observations from the experiments with no explanation, or at least attempt at an explanation.
Further, I am not convinced that an apples-to-apples comparison between the two runs is possible in its current setup. As stated on lines 67-68, only the radiation parameterization is the same between runs. All other aspects of the atmospheric part of the model are different. Indeed, it is stated on lines 90-91 that ‘we assume that the treatment of convection plays a key role’. This is a huge assumption to make given all of the differences, not least the difference in resolution itself that can have a big influence on the width and strength of the Hadley cell (e.g., see Lockwood et al. 2025 for a recent paper).
Following on from not being able to definitively say that the difference in convection schemes is key for the change in strength of the Hadley cell, the change in midlatitude eddies is also not accounted for, or even discussed. Such eddies play a huge role in determining the strength and width of the Hadley cell (e.g., see Davis and Birner 2019 and references therein for a recent paper, but also a plethora of papers from the 80s, 90s and 00s by Held, Hou, Lindzen, Schneider to name just a few authors), and I imagine that increasing the model resolution will significantly influence their scale, strength (perhaps by differences in moist effects), and breaking patterns. Currently your experimental setup does not constrain such an effect as the midlatitude eddies can change and thus changes in the Hadley cell cannot be attributed solely to tropical convection differences.
Overall, my suggestion is to reject the paper in its current format. Even though the subject matter is interesting (indeed, I was excited to read the paper given the title!), it does not provide enough new science to justify publication. I do hope the authors reimagine and rewrite the paper in such a way that they can answer the interesting question they have posed.
Specific Comments:
Line 10: ‘leads to less heavy rain and less rain in the ascending branch of the HCHC—consistent with a weaker HC—but more total precipitation when taking also light rain into account ‘. This is quite confusing wording (the ‘less rain’ part) so I suggest rewording.
Line 21: ‘2005’s’ is oddly specific given you said ‘around’ too!
Lines 55-60: I would make clear that the low-resolution configuration is the unresolved convection run, and the Sapphire configuration is the resolved-convection run.
Line 66: Why are snow and graupel now included in the list of hydrometeors? I understand that graupel can only form in high resolution models as its formation requires highly localized strong updrafts that enable supercooled liquid to form, but including it in the high resolution model may take away from the precipitation (which is presumably only rain and snow) used in the precipitation metric described on lines 116-123.
Lines 90-91: this is a huge assumption as resolution can change the strength of the HC itself. More broadly though, regarding these simulations, there doesn’t appear to be a potential apples-to-apples comparison between the two experiments you have conducted. There are multiple differences between the runs with various parameterisations turned on and off, e.g., gravity waves (orographic and non-orographic). As you state, only the radiation scheme is the same. I think a more robust comparison can be made if the two experiments are more similar in their setups.
Line 106: ‘norther’ --> northern
Lines 101-113: I would consider marking all of these latitudes on Fig 1 to aid the reader. Also the various pressure levels used.
Line 117: is this total precipitation in including snow and rainfall?
Line 118: ‘excesses’ --> ‘exceeds’
Line 121: I would say that it is directly related to the strength of the upwelling of the HC to be more specific. Also, line 122 doesn’t read well.
Figure 2: I don’t think both 400 and 500hPa need to be plotted. Rather just show one and state whether the other is similar, which indeed looks to be the case. Although the fact that the y axes are different in a and b make it difficult to tell. Also, what are the red crosses? Presumably the single value for the CTRL-5, but it is not said.
Figure 3: The different y axes limits make it difficult to get a good sense of the differences between the various HC heights. Also, the h3 and h5 give quite confusing results when comparing the CTRL-40 runs: the median and lower whisker extend to lower pressures for the h5 isoline, i.e, they reach higher into the troposphere. It seems like it should be opposite to me…
Lines 166-170: this is a global-mean precipitation and so includes all of the changes in midlatitudes and thus the storm tracks. My takeaway from this is that the differences (between CTRL-5 and CTRL-40) in rain outside of the tropics must be extremely large to overcome the large, opposite-signed differences you see in Pasc and Phvy.
Figures 6a-7a: A difference between the two would be useful perhaps.
Figures 8-9: what are the red stars here?
Line 200: the meridional velocity is referred to in many places. I would consider showing something related to that as a way to break the box-and-whisker monotony from having the same kind of plot too frequently.
References:
Davis, N. A., & Birner, T. (2019). Eddy influences on the Hadley circulation. Journal of Advances in Modeling Earth Systems, 11, 1563–1581. https://doi.org/10.1029/2018MS001554
Lockwood, J. F., and Coauthors, 2025: The Effect of Increasing Model Resolution on the Northern Hemisphere Winter Midlatitude Storm Track: An Equatorward Shift due to Contraction of the Hadley Cell. J. Climate, 38, 4539–4551, https://doi.org/10.1175/JCLI-D-24-0414.1.
Olivier Pauluis et al. The Global Atmospheric Circulation on Moist Isentropes.Science 321, 1075-1078(2008). DOI:1126/science.1159649
Citation: https://doi.org/10.5194/egusphere-2026-1168-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 488 | 227 | 49 | 764 | 46 | 57 |
- HTML: 488
- PDF: 227
- XML: 49
- Total: 764
- BibTeX: 46
- EndNote: 57
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This paper sets out to assess the impact of resolved convection on Hadley cell strength in atmospheric model simulations. This goal is both interesting and timely, given the increasing turn toward km-scale simulation in the community.
Unfortunately, I find the paper itself disappointing. The authors set up the ICON model in two configurations: low-resolution with parameterized convection, and high-resolution with explicit convection; they perform control and warm climate simulations, measure the strength of the Hadley cell and precipitation according to some metrics, report these metrics, and stop there. In this reviewer's opinion, this is simply not enough to pass muster as a research paper. It is an example of the "run and show" strategy: run the model, show the results, and leave it to the reader to interpret them. That may be (just about) sufficient to pass a master's thesis, but does not count as creation of new knowledge, which is what a research paper is supposed to report.
The paper's most salient result is that high resolution + explicit convection makes the Hadley cell weaker. There is no attempt to identify the mechanism responsible for this weakening. That is despite the ready availability of many potential mechanisms that have been proposed over the years: the Chemke and Polvani (2021) paper lists six of them, to which one could add eddy-stress related mechanisms in e.g. Bordoni and Schneider (Nat Geoscience 2008). Worse, the paper does not even provide a convincing case that the reduction in strength is solely attributable to explicit convection. In the two control simulations, there are two things changing: the convection scheme, and the resolution itself. But changing resolution can by itself affect the Hadley cell, even with the same convection scheme. So the methodology does not even properly constrain the key question it sets out to address.
In conclusion, my advice to the editor is to reject this manuscript. In my view, no amount of revision of the existing material will lift it above the bar. What is needed is a large amount of new analysis and additional simulations that actually address the question posed, amounting to an entirely new manuscript. That future manuscript may well be worth publishing, and can be reconsidered if and when it is submitted.