the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 30 Aug 2024
Flavor identification of the stratospheric sudden warmings based on the downward tropospheric influence
Abstract. The downward impact of sudden stratospheric warming events (SSWs) on the troposphere is still uncertain. Using the ERA5 reanalysis data, 52 SSWs are identified over the period from 1940–2022, and 33 downward-propagating (DW) SSWs with noticeable impacts on the troposphere are selected with the remaining 19 SSW non-downward-propagating (NDW). The DW events are further classified into three types that are followed by cold surges over the Eurasia (EA), over the North America (NA), and over both (BOTH), respectively. Although the stratospheric polar vortex is weakened and deformed for both DWs and NDWs, the former are stronger and lead to more significant negative Northern Annular Mode (NAM) and North Atlantic Oscillation (NAO) response in the troposphere. For DWs, the anomalous high develops in the polar region, which deflects to lower latitudes, consistent with the frequent appearance of the polar high and the midlatitude blockings. The shape of the anomalous polar high varies with the DWs type, and the extension and deflection of the anomalous high lead to different surface responses. The DWs are also accompanied by a southward shift of the precipitation belt especially over the oceanic and coastal regions. The NDW SSWs show relatively weaker impact on the troposphere, which is primarily related to the weaker amplitude of the stratospheric disturbance. The differences among three types of DWs include diverse NAM structures in the stratosphere, various spatiotemporal evolutions of the NAO pattern in the sea level pressure, different forcing by planetary waves, and varying number ratios between displacement and split. This study reveals the diversity of the DW events and distinguish their potential impact on both continents in the Northern Hemisphere.
- Preprint
(9630 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2179', Anonymous Referee #1, 18 Sep 2024
Summary
The authors classify sudden stratospheric warmings (SSWs) into those that propagate downward (DW) and those that don’t (NDW), and then further classify the DW events into those with cold air outbreaks over Eurasia (EA), North America (NA), and BOTH. They look at the dynamical differences before and after each type of event.
Overall
There are potentially interesting findings here, and the presentation/figure quality is very good, but overall it is difficult to argue that the results are very meaningful when most of it is a result of how the authors constructed the analysis. In addition, there simply aren’t enough samples in the reanalysis record (52) to divide further into 4 categories and expect statistically meaningful or interpretable results. Finally, the article is greatly hindered by a lack of proof-reading and grammar issues. Overall, I can’t recommend publication at this time.
Major Comments
- Sample size. I think it is hard to make any substantial statistical arguments with 33 DW events and 19 NDW events. Note that White et al.2019 find that at least 35 events must be present to get robust differences between DW and NDW events (https://journals.ametsoc.org/view/journals/clim/32/1/jcli-d-18-0053.1.xml). The 33 DW events are then even further divided into EA, NA, and BOTH surface impacts; for NA, there are only 5 samples, according to Table 1. This makes the significance levels for NA almost meaningless. Moreover, most of the paper discusses differences between EA, NA, and BOTH, but no statistical tests are applied to actually quantify the differences. My feeling is that almost none of the noted differences in these categories are likely to be statistically different, because the sample size is simply too small. This is why using a model to get many thousands of SSWs would be useful.
- Use of stratospheric reanalysis prior to 1958. This is just model data at that point, not reality- in which case, why not use model data to up your statistics? If model data simply is not available (which it should be- many high-top CMIP pi-Control runs could likely simulate downward coupling very well), at least don’t use reanalysis before 1958, as any SSWs prior to then are almost certainly going to be in model land anyway and not strongly constrained by observations.
- A huge part of the analysis section discusses differences between EA, NA, BOTH DWs and NDWs without seeming to acknowledge/recognize that most of the differences in the post SSW period are there *by construction* of how the authors classified these events. If you classify NA events by cold air temperatures over NA, they are going to have cold air temperatures over NA, and the corresponding large-scale meteorology to give you those cold air temperatures. Additionally, there are several statements about how NDWs are “less organized” and don’t show similar spatial structures; but this isn’t a fair comparison, because the NDWs *aren’t* divided up by regional surface temperature responses and instead average together all the variability. A more apples-to-apples comparison would be if you also divided the NDWs up in the same way (but again, this would cause very small samples). Overall, I don’t think much is learned except in the pre-SSW period since that is somewhat more independent from the classification method itself.
- Text needs to be edited for grammar and just general proof-reading. I started off adding suggestions for Technical Edits, but to be honest, could not keep up with the level of errors, and so at some point I stopped adding to the Technical edit list below. In many cases, the edits go beyond a language barrier and just seem to be due to a lack of reading the paper carefully before submitting (e.g., sentences are repeated twice, results are explained for different regions than are stated, etc).
Minor Comments
Line 5, 33, and throughout: suggest using “sudden stratospheric warming” phrasing instead to be consistent with most recent literature (Baldwin et al. 2021; see also, Butler et al. 2015, BAMS, for historical context in using this phrasing).
Line 20, 22: Not clear what is meant by “deflection of the anomalous high”. Do you mean “shift”?
Line 22, 26: If there are only 33 DW events, are they further divided into type of SSW? This seems like the statistics would be very poor. See also, Major Comment #1.
Line 24: Isn’t it somewhat by construction that the NDWs have a weaker impact on the troposphere?
Line 28: Is it really showing robust diversity across events, or just the large contribution of sampling variability?
Line 38: Butler et al (2015) was about defining SSWs, and did not quantify drivers of SSWs. A more appropriate reference here might be Polvani and Waugh (2004)
Line 44-45: I’m not sure what is trying to be said here, but this is not true. Stratosphere-troposphere coupling is a two-way process, e.g., there are certain tropospheric patterns that precede stratospheric variability, which then couples back to the surface. Maybe instead “Stratospheric variability associated with SSWs affects the troposphere through stratosphere-troposphere coupling processes” ? Alternatively, just remove this sentence and start the paragraph with “Several mechanisms…”
Line 53-55: Yes, but this can really only explain the downward descent of the anomalies to the troposphere. There are aspects of wave-mean flow interaction that cannot fully explain the amplified response at the surface. It’s been proposed that the eddy feedbacks in the troposphere must play some role, but this is not well understood. For a review, see Kidston et al. 2015 (https://www.nature.com/articles/ngeo2424)
Line 60: This needs to specify “The negative phase of the NAM is”, not just “Changes in the NAM are”, since what is discussed thereafter are specific to one phase.
Line 68-69: Is this really true? Or is it just the large internal variability present during any individual SSW event that dominates how much influence the SSW will be able to have on the surface? The dynamics of SSWs are by and large very similar across events- so I’m not sure I agree that the differences between SSWs lead to differences in their impacts. For example, see Maycock and Hitchcock 2015- when you have a lot of simulated SSWs, differences in their surface impacts arise largely from sampling, not because of inherent differences in the SSWs (https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2015GL066754)
Line 85: Does there need to be? Given the huge range of possible internal variability, I think it would be almost impossible to cleanly classify by surface impact.
Line 105: The definition that is described here is that from Charlton and Polvani (2007). The WMO never had a definition that specifically was based on zonal winds alone; see Butler et al. (2015). Likewise, on line 115, this aspect of the definition was developed in Charlton and Polvani (2007), not White et al. (2019).
Line 113-114: Not sure what is meant here by “zonal winds oscillating between westerlies and easterlies.” As the corrigendum in CP07 states, the events must be separated by more than 20 days of consecutive westerlies; not just 20 days apart. This number of days arises from roughly double the radiative timescales in the mid-stratosphere, to avoid double counting what might otherwise be the same dynamical event.
Line 127: The average NAM is negative at what level? Both? Either?
Line 134-5: Provide citations for this statement, or refer to your own Figure
Line 147-8, 152: This method was actually from Mitchell et al. 2011, and then was adapted by Seviour et al. 2013: Seviour, W. J. M., D. M. Mitchell, and L. J. Gray (2013), A practical method to identify displaced and split stratospheric polar vortex events, Geophys. Res. Lett., 40, 5268–5273
Line 164: There are a lot of other undefined variables here
Line 190-215: are any of these differences across DWs significant though? My guess is no, but that should be assessed here (the significance of the individual composite anomalies is assessed, but this is different than stating that the composites themselves are different and then testing that- you need to do a difference in means test, using the sample size/standard deviation of each composite group). Given that NA only has about 5 samples, it’s going to be hard to judge whether you’re not just seeing sampling variability.
Line 211: By requirement, it has to have negative NAM850 at least 50% of the time, no matter what type of DW, right? So not sure what is meant here about “short in the persistent time” for NA. Also, I bet if you randomly resample EA and BOTH for 5 samples to equal the sample size of NA, you can get composites that look like NA, just by chance.
Line 214-215: This is by construction though, no?
Line 215: specify “over the North Atlantic”
Figure 2, line 224-25: I’m worried about how smooth these look. For the blue curve (NA) there are five samples being put in to make this very smooth looking distribution. (this assumes these are 60-day averages, not all 60 days being put into the PDF… not totally clear from the caption).
Like 226-228: I’m not convinced these mean values (or the PDFs in general) are statistically different, particularly for the DWs. You could do a Kolmogorov-Smirnov test to see if the PDFs are separable. Moreover, since the NAO is highly correlated to the NAM, this result (at least the difference between NDW and DW) is also somewhat by construction. Also the values cited here say “NAM” but the NAO is being shown in Figure 2 so I assume that’s what is meant?
Figure 3: would be beneficial to add some measure of tropospheric circulation to these figures, such as Z500. That would also help tie into Figure 1.
Line 239-255: It should be re-emphasized throughout that the cold anomalies over the region of definition (e.g., NA, EA) are by construction, so the fact that these regions show different anomalies is not unexpected nor does it tell you much about why they are different in other regions.
Line 247, 287-88: This isn’t a fair comparison. Here, you’ve divided the DW into categories that are defined based on persistent cold events in a given region, so they’re going to look very distinct and organized—by construction-- compared to the NDW events which are thrown into a pool of all types of surface impacts. A more fair comparison would be to divide the NDW events using the same classifications used in the DW events based on regional cold air outbreaks, and compare those groups…
Figure 4: would be easier to read if instead of I-IV the regions (Europe, Asia, etc) were provided in the label on the plot
Line 256: ? There is no temperature anomaly reversal over Europe for NA in Figure 4
Line 270-290: In many cases, the highest IPV is associated with the Greenland block (e.g. BOTH and NA) yet the authors are focused on much smaller positive anomalies over the N. Pacific as a source for the NA cold. Yet it’s been shown in the literature that the Greenland blocking pattern (also associated with the negative NAM/NAO) would advect cold air over at least eastern North America. To me this seems to suggest this is the dominant forcing of cold over NA.
Line 304-305: kind of? The pattern is shifted much further poleward for NDW, and looks less spatially coherent.
Line 325: NA looks more baroclinic to me than EA?
Line 330: Again, the majority of post-SSW impacts at 100 hPa are going to be reflecting the event definition…
Line 335-345: Throughout this section it’s often not clear which level is being referred to, 100 hPa or SLP- so make sure it’s evident in each instance.
Line 352-353: This doesn’t seem apparent from Figure 8. At least around 100 hPa, NA seems to have the strongest convergence (blue shading)
Line 370-71: I’m not clear what the first part of this sentence has to do with the second part.
Figure 8-10: I’m confused why the EP flux convergence anomalies are so substantial for days 0-10 in Figure 8, yet when decomposed into wave 1 and 2 in Figures 9-10, they basically disappear/are much weaker. Since wave 1 and wave 2 are the primary components to the total wave flux, I would check and make sure there’s not an error in Figure 8…
Line 395-397: It is very difficult to parse this type of parenthetical sentence. Please separate into two (or possibly 3-4) sentences.
Line 418-421: This makes it sound like the authors are implying the air is actually coming from the polar stratosphere, which is simply not true!
Line 429: It’s different because you defined it to be so! If you classified by precipitation pattern instead, the patterns would look different, *by construction*.
Technical Edits
Line 19: “response” -> “responses”
Line 29: “distinguish”-> “distinguishes”
Line 37: “the stratospheric precondition might play a…” -> “found that stratospheric preconditioning might play a decisive role in inducing the SSW event, by determining the intensity…”. Also, can you add a reference for this statement?
Line 39, and throughout: “that SSW is caused” -> use “a” before SSW when used in the singular, or else say “that SSWs are caused” (as mentioned in Major comments, text should be edited throughout for English style)
Line 46: “mechanism” -> “mechanisms”
Line 51: “affect stratospheric mean flows, and changes in stratospheric background flows in turn affect” -> “affect stratospheric mean flows, which in turn affect”
Line 58: “is usually projected onto the negative phases” -> “usually projects onto the negative phase of”
Line 64: remove “populated”, not sure what is meant; also double use of “extreme”
Line 69: either “extent” or “degree”, not both
Line 82: “the DW SSWs show a dipping NAM signals” -> I think you mean to say “the DW SSWs show dripping NAM signals” (?)
Line 88: change to “although by definition the NAM…”
Line 89: need question mark at end instead of period
Line 101: repetitive phrasing of “isobaric levels vertical levels”. This is also not a complete sentence (change “ranging” to “range” and then put the surface data part in a separate sentence)
Lines 105-115: Suggest carefully re-reading and editing this section; many grammar issues or sentence structure problems.
Line 118: change “extracted” to “calculated”
Line 147: change to “The classification of SSWS into split and displacement events is…”
Line 165: EP-flux acronym is never defined
Line 174: “characterize” -> “characterizes”
Line 179: just say “The NAM index can be used to describe”… Also, can remove “for DWs”. It describes the extent of downward propagation whether they are DW or NDW.
Line 182: remove “radical”
Line 194-195: for NA? or what is this sentence referring to?
Line 245: “EA” should be “NA”. Also I think that western Europe should be “eastern Europe”? (warm anomalies in western Europe seem to weekend compared to before SSW, in 3g)
Figure 6, 8-10: labels for negative values on colorbar are not shown
Line 309: change to “the circulation structure is organized differently for..”
Line 320: “the lakes” -> the “Great Lakes”?
Line 337-338: I think I get what you’re trying to say but the phrasing here needs improvement
Line 344: “even insignificantly detectable” is awkward- also there appears to be stippling over most of the Arctic so not sure what this refers to?
Line 412: just need “weaker” or “smaller” not both
Line 415: sentence repeats itself
Line 466: “pace” -> “place”
Citation: https://doi.org/10.5194/egusphere-2024-2179-RC1 -
CC1: 'Comment on egusphere-2024-2179', Tao Wang, 17 Oct 2024
Comments on “Flavor identification of the stratospheric sudden warmings based on the downward tropospheric influence” by Lu and Rao
Summary
Using the reanalysis, this study analyzes the possible impact of the downward-propagating SSWs on the continental climate. To by best knowledge, this is the first study to classify the downward-propagating SSWs into three types: North America, Eurasia, and BOTH. This classification is established on the existing evidence that the composite SSW shows wide coldness anomalies over the Eurasia and/or North America. However, the coldness over both continents is not synchronous for all SSWs. It will improve the understanding of the diversity of the SSW in influencing the near surface. In general, this study is very interesting and worth publishing after a revision.
Specific comments
- The classification of downward-propagating SSWs is mainly based on the cold anomalies over both continents. Not all of cold temperature anomaly variations are caused by the stratospheric variability, and the composite might filter out the contribution of other variability if the sample size is large enough. Is there any possibility to increase the sample size if the model data are used? For example, CMIP6 provides a large model ensemble dataset, which contains much more samples than ERA5. If those data are used, the stability of the composite results can be well confirmed.
- This study classifies the SSW using the t2m anomalies, which is based on the fact that major SSWs show larger and more significant t2m composite than rainfall composite. Are the zonal band of rainfall anomalies over North Atlantic sensitive to the threshold of downward-propagating SSW type?
Other comments
- L38: determine => determines; Butler et al., => remove “,”
- L44: lead => leads
- L46: mechanism => mechanisms
- L56: and have => remove
- L64: populated => growing ; extreme extreme => remove one
- L101-102: Reviesed as “The isobaric levels extend from 1000 hPa to 1 hPa, and thehorizontal resolution of the data is 0.25° latitude by 0.25° longitude.”
- L105: SSW => SSWs; All => all
- L107: The 1 November => 1 November
- L111: using => used; do => does
- L113: twice => more than once
- L117: ERA => remove
- L120: the cosine => cosine
- L125: Natarajan et al., => Natarajan et al.
- L130: White et al., (2019) => White et al. (2019)
- L147, 151: Esler et al., (2009) => Esler et al. (2009)
- L174: Characterize => characterizes
- L186: return => returns
- L188: are present => is present
- L198: persist => persists
- L231: interval => intervals
- L232: region => regions
- L243: move => moves
- L259:stable moderately warm state => moderately warm stable state
- L267: Lu and Ding, 2015; => Lu and Ding, 2015
- L309: the circulation structured is different => the circulation structure is differently organized
- L330: amplitude .. are => amplitude is
- L337: cut => cuts
- L342: midlatitude => midlatitudes
- L364: all for => for all
- L385: reverse => reverses
- L392: elongate => elongates
- L409: finding => findings
- L412: weaker smaller => weaker
- L421: ocean regions =>oceanic regions
- L424: 0-10 day => 0-10 days
- L431: show => shows
- L440: show => shows
Citation: https://doi.org/10.5194/egusphere-2024-2179-CC1 -
RC2: 'Comment on egusphere-2024-2179', Anonymous Referee #2, 12 Nov 2024
Review of “Flavor identification of the stratospheric sudden warmings based on the downward tropospheric influence” by Lu and Rao.
I agree with Reviewer 1 that the statistics are not sufficient to divide the SSWs into four categories. From Table 1, there are 13 BOTH, 14 EA, 6 NA and 19 NDW. Those are very small numbers. In addition, I noticed that 4 of the 6 NA occur in the 2000s, and 6 of the 13 BOTH also occur in the same decade (1970s). There are also many more EA towards the end of the considered time period than early on. Thus, besides the comments from Reviewer 1 that these numbers are not sufficient for statistical analysis and that data before 1958 is questionable, I wonder whether the composites include things like a climate change trend (EA) or potentially decadal variability (BOTH, NA), which both might have nothing to do with SSWs.
Furthermore, the cited paper by Jucker (2016) has shown that composite evolution and surface impact can be very different depending on the exact definition of “downward propagation”. For instance, they show that simply checking for expected anomalies after SSW onset captures periods of internal variability where these anomalies might already exist before the SSW can influence the surface. Compare this, for instance, to the statement on lines 314-315, “The negative NAM pattern has well developed before the SSW onset for the type BOTH.”
I am also a bit sceptical about the use of the sign of the NAM as a measure of downward propagation (lines 125-129): The NAM is a zonal mean quantity, but EA and NA are explicitly defined as strongly zonally asymmetric surface anomalies (impact in one region but not another). So why to the authors think using the NAM is the best way to define downward influence of SSWs? Is it not possible that they are missing several occurrences where EA or NA are anomalously cool but the zonal mean is still neutral (and thus this would be classified as NDW)?
I think these are major points concerning the design of the study which need to be re-considered, and I therefore do not recommend publication in the current form.
Citation: https://doi.org/10.5194/egusphere-2024-2179-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 149 | 24 | 15 | 188 | 6 | 7 |
- HTML: 149
- PDF: 24
- XML: 15
- Total: 188
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1