Relationship between southern hemispheric jet variability and forced response: the role of the stratosphere

Breul, Philipp; Ceppi, Paulo; Shepherd, Theodore G.

doi:https://doi.org/10.5194/wcd-3-645-2022

Articles | Volume 3, issue 2

https://doi.org/10.5194/wcd-3-645-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/wcd-3-645-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 3, issue 2

Research article

|

15 Jun 2022

Research article |

| 15 Jun 2022

Relationship between southern hemispheric jet variability and forced response: the role of the stratosphere

Philipp Breul, Paulo Ceppi, and Theodore G. Shepherd

Download

Final revised paper (published on 15 Jun 2022)
Preprint (discussion started on 30 Nov 2021)

Interactive discussion

Status: closed

RC1: 'Comment on wcd-2021-78', Anonymous Referee #1, 26 Jan 2022

This study helps to clarify the complex relationship among the extratropical jet climatological location, its timescale of meridional shift, and the forced responses seen in observations and models. Especially, it helps to answer (but not fully solve) two long-standing problems: 1. most GCMs overestimate the timescale of the Southern annular mode; 2. the fluctuation-dissipation theorem predicts the forced jet shift to be proportional to its timescale, but no such relationship is found among GCM simulations. This study shows that the prolonged timescale in austral summer is largely due to the variations in the breakdown date of the stratospheric polar vortex. Removing such effect leads to a closer agreement in the annular mode time scale between observations and GCMs, and a much better agreement between the forced response in a simplified barotropic model and the estimation from the fluctuation-dissipation theorem. But it does not lead to a better correlation between the timescale and the forced responses among GCMs. Overall, it is a well written manuscript with carefully designed experiments. I have a few minor comments for the authors to consider.

1. The authors very nicely showed the effect of the polar vortex breakdown date on the annular mode timescale. A perhaps slightly different way to interpret these results is that the jet is in two regime depending on whether the troposphere and the stratosphere is strongly coupled (that is, coupled in spring/autumn vs no coupling in summer/winter or torque on vs off). The fluctuation-dissipation theorem considers variations based a particular mean state, and therefore cannot count for the regime shift before and after polar vortex breakdown (VB). The authors used the regression against the VB date to count for such regime shift. It certainly works as shown in the manuscript, but I wonder if there is a cleaner way to do that. I a bit concerned that the effect of such regime shift may not be linear to the timing of the regime shift. For example, VB occurring at Nov 1 and Nov 5 may be different compared to VB occurring at Dec 1 and Dec 5. One thing that might be worth trying is to construct a conditional annual cycle: S(date of the year)=S1 if before VB, S=S2 if after VB, and calculate anomalies by removing this conditional annual cycle.

2. CMIP5 models in general have a too-late VB breakdown than observations (e.g., Wilcox and Charlton-Perez 2013 JGR). I therefore expect that the VB’s effect peaks at the later time of the year in CMIP5 models than in ERA5. But in Fig. 5, the lowest corrected SAM fraction in models is found at a slightly earlier date than in ERA5. This might be beyond the scope of this manuscript, but I would love to see some discussions regard this. One possibility is that the regression does not isolate the VB effect clean enough as mentioned in my last point.

Editorial comments:

Line 178: “dotted” -> thin

Line 197-198: delete “(we only show… identical)” since both VB-VAR and VB-FIX are shown in Fig. 6.

Fig 5a legend: it may be more meaningful to move the “CMIP5 average” to the two solid lines, rather than the title of all four lines.

Fig. 6: The two legend box of the same figure panel is confusing. It may be better to combine them. Also in Fig. 6a, it may be better to plotting the orange dashed line on top of the black line.

Citation: https://doi.org/10.5194/wcd-2021-78-RC1
RC2: 'Comment on wcd-2021-78', Anonymous Referee #2, 01 Mar 2022

Review

Relationship between Southern Hemispheric jet variability and forced response: the role of the stratosphere

P. Breul et al.

The authors revisit the connection between southern annular mode timescales and the response of the southern hemisphere jet to anthropogenic climate forcings. Simple arguments from the fluctuation dissipation theorem suggest that the two should be to some extent correlated in global models such that those models with a longer timescale should have a larger response. Such a relationship could be valuable as an emergent constraint on projected changes in the circulation; this is thus an important topic that should be of broad interest to the readers of WCD.

This connection has been questioned by previous studies. The present study considers whether the effects of variability in the austral stratospheric polar vortex on SAM timescales could be responsible for the apparent lack of correlation identified in CMIP5 models by Simpson and Polvani (2016). They develop a regression methodology for removing the stratospheric contribution to the timescales based on some idealized integrations of a barotropic model, and apply this to the CMIP5 archive. In the end they find that this does not improve the correlations. Hence this is essentially a negative result.

Given the potential importance of identifying such an emergent constraint, this kind of negative result may still be valuable to report. However, the authors have not done a very good job of highlighting what insights this paper adds to the literature; in fact very little previous work has been cited at all. In particular, many other studies have come to similar conclusions that applying the FDT naively in this way is often quantitatively incorrect; see, e.g.

https://doi.org/10.1016/j.physd.2016.05.010 https://doi.org/10.1175/2010JAS3633.1 https://doi.org/10.1175/JAS-D-16-0099.1

though this is by no means exhaustive.

In order to be worthy of publication, the present study should be clearly placed in the context of the existing literature, and the authors need to make a compelling case for why this negative result adds something of value to the literature. This may mostly an issue of rewriting, but I have nonetheless recommended major revisions on this basis.

Some further general and specific comments are given below.

General comments

Assumption that the stratosphere is exogenous

The authors have assumed without much comment that the contribution of the stratosphere to the southern annular mode timescales should be 'corrected for', in that it should not be seen as implying a stronger response to an imposed force. This is clearly the case in the barotropic model where the effects of the 'stratosphere' is explicitly an exogenous force. But it's not at all obvious that this should be the case in the real atmosphere (or in climate models). The framing (adopted from Byrne et al. 2017) is that the stratospheric influence renders the temporal statistics of the tropospheric jet 'non- stationary.' But how does this really connect to the FDT? Consider, for instance, a climate model with a well-resolved stratosphere with explicitly fixed or periodic boundary conditions. Such a dynamical system can easily include seasonal variability in the stratosphere and troposphere, including variability in the date of the vortex breakdown that leads to the DJF peak in SAM timescales. Assuming that this dynamical system has some kind of well-defined attractor, the statistics are certainly stationary after some initial transient evolution to the attractor. If we apply the FDT to this dynamical system, on what basis should we consider some parts of the system 'exogenous'?

There is some dynamical intuition that this should be the case, but it's not obvious to me how to justify this intuition in the context of the FDT, and certainly the 'non-stationary' argument can't be right. Why shouldn't the full FDT apply?

I don't expect a full answer to this question, but the authors have just tacitly assumed this point which is not at all obvious. More generally, how do we know that the stratosphere is the only 'exogenous' impact on the SAM timescale? Also, without a clear theoretical basis for this approach, how do we know this 'correction' is even the right thing to do? The barotropic model is hardly a justification.

Barotropic model

The authors have modeled the effects of the stratosphere in the barotropic model as a time-dependent imposed force. This will have the desired effect of modifying the jet, but the nature or origin of such a force in the real system remains a topic of some debate. Does the spatial structure of this imposed force have any effect on the ability of the regression model to correct for the timescale?

Use of breakup date

While the timing of the vortex breakup is probably the single biggest source of year-to- year variability in the austral vortex, it doesn't capture everything. The regression

methodology could just as easily be applied to, say, the anomalous zonal mean zonal wind at 50 hPa for instance; this would include more stratospheric variability.

Use of EOF1 only for GCMs

At various points the claim is made that applying the 'full' FDT calculation in EOF state space would be prohibitive for the GCMs. I don't see any reason this is the case if you are already limiting the state space to the tropospheric 'barotropic' zonal mean zonal wind. The same approach to choosing a subset of EOFs could easily be applied, and some discussion about the validity of restricting the FDT to the leading mode of variability could usefully be made. This I think would be somewhat novel.

Specific comments

Two technical questions that should be clarified in the text (not just in the response):

1) How is are the autocorrelation `timescales' calculated? The integral expressions in the appendix are more general, but only reduce to the decorrelation timescale for specific processes. Yet it is common to assume exponential decay, at least over some fixed set of lags. What was done here, and how is the choice justified?

2) What is the state space used in the appendix? The full dynamical state space of the barotropic model, or just the zonal mean zonal wind?

l152: 'position decreases' -> 'latitude moves equatorward'

Fig. 6: Are the dot-dashed lines in Fig. 6c and d the same as the full lines in Fig 6a and b? The figure caption suggests so, but the label suggests otherwise, and they don't look the same for the Gaussian case. It would be particularly helpful to clearly evaluate the importance of the EOF1 assumption.

l232: I'm not sure that constrained is the right word here - there's no validation of the current result. The range of timescales has been reduced however. A similar comment applies to l280.

Fig. 7: Some comments are made highlighting the fact that models with a climatological jet position that is close to ERA5 have timescales that are generally close to the reanalysis as well. Is this agreement better for 'corrected' timescales than it is for the uncorrected timescales? This is not shown.

l286-287: This has already been done by a number of studies (see those references cited above, for instance).

Citation: https://doi.org/10.5194/wcd-2021-78-RC2
AC1: 'Comment on wcd-2021-78', Philipp Breul, 29 Mar 2022

We attach a file containing our response.

Citation: https://doi.org/10.5194/wcd-2021-78-AC1

Peer review completion

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

AR by Philipp Breul on behalf of the Authors (31 Mar 2022) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (01 Apr 2022) by Yang Zhang

RR by Anonymous Referee #1 (11 Apr 2022)

ED: Publish as is (05 May 2022) by Yang Zhang

AR by Philipp Breul on behalf of the Authors (23 May 2022)

Short summary

Understanding how the mid-latitude jet stream will respond to a changing climate is highly important. Unfortunately, climate models predict a wide variety of possible responses. Theoretical frameworks can link an internal jet variability timescale to its response. However, we show that stratospheric influence approximately doubles the internal timescale, inflating predicted responses. We demonstrate an approach to account for the stratospheric influence and recover correct response predictions.