I would like to start by thanking the authors for bringing substantial modifications, that resulted in a more detailed and easier-to-assess manuscript. Given the large differences with respect to the initial version, I treated that as a novel manuscript and provide below a full review.
From a scientific point of view, I believe that the paper still suffers from the ambiguity in the definition of the waves responsible for the analyzed cold spells. Maybe the problem I see behind this study is that it focuses on cold spells, i.e., extreme events that occur on synoptic time scales, while at the same time using a wavenumber-based approach to separate Rossby waves that has been so far applied to climatological, hemispheric-scale studies (about, e.g., energy transport), and not to individual events. Defining planetary waves at instantaneous time steps is quite tricky and probably inexact because the concept of planetary wave (especially when defined simply through a wavenumber range) does not bear a clear physical fingerprint and assumes the presence of a "low-frequency" dynamics that is distinct from the "high-frequency" one. Isolating slow dynamics in the background using filtering in time or space is extremely tricky as pointed out by, e.g., Wirth and Polster (2021, https://doi.org/10.1175/JAS-D-20-0292.1). I have the impression that the whole paper suffers of this ambiguity, which is still well present despite the efforts of the authors: the first set of major comments try to target this point.
Another important aspect that could be further improved is to highlight the value of the scientific insights that can be gained from the analysis. A way to do so would be to specify more explicitly the research questions that this piece of work tries to answer: these questions should emerge clearly from the Introduction as unanswered by previous research. Some additional suggestions are listed in the detailed comments.
From a methodological point of view, the authors have clarified in which aspects their method is different from others (although the novelty could be emphasized a bit more, see detailed comments). As a side effect of moving many informations from the supplement to the main text, several questions about the choice of the methodology have arisen. They are related partly to the previous point, as methodological choices reflect the theoretical framework of climatological waves: this is visible, for instance, in the decision of tracking transient ridges only in regions where the climatological, stationary ridges are found. I have several detailed comments about it and a major comment with a suggestion about how the methodology can be revised to make it less arbitrary and solve many issues.
In summary, I see again for this paper the necessity of substantial revisions both from a methodological and a scientific point of view. This would warrant at least a request of major revisions or, if the handling editor sees fit, of rejection with encouragement to resubmit.
Major comments about the ambiguity in the type of waves considered in the Z1-5 range
1) The selected range of waves (in the wavenumber range Z1-5) is at times considered capturing climatological waves, sometimes as low-frequency, "planetary" waves, and sometimes as transient troughs and ridges driven by, e.g., Rossby wave packets. The focus on climatological waves is at the basis of the methodology, because trough and ridge tracking are performed only in precise ranges of longitude mapping into the known stationary ridges and troughs of the northern hemosphere, due to orography or land/sea contrast (e.g., the 180W-90W region upstream of the Rocky Mountains mapping). This explains why it is more likely to identify troughs and ridges in given longitudinal sectors (as depicted in Fig. 4), sectors denoted as the "original locations" of ridges and troughs in the reply document (see the second sentence of the reply to major comment 3). The Z1-5 waves, on the other hand, are treated as being part of transient Rossby wave packets when the results of this work are compared with the ones of Fragkoulidis et al., who specifically focused on transient Rossby waves (e.g., at lines 317-322). The presence of "planetary" waves, a third type distinct from the other two, is assumed implicitely in lines 120-124, where it is stated that Z1-5 waves at high latitudes might have wavelengths that are "too short for their characterization as planetary waves".
2) While distinguishing those types of waves might not be crucial from the perspective of the adopted methodology (which might be seen as "agnostic" about the nature of the waves found between n=1-5), the distinction is on the other hand very important to determine the scientific insights that can be gained from the analysis. For instance, interpreting North American cold spells as the result of the amplification of the climatological ridge upstream of the Rockies might be misleading. This is because the orographic wave dynamics explaining the presence of such a ridge is fundamentally different from the one explaining the genesis, amplification and polar excursion of the midlatitude upper-level ridges occurring downstream of extratropical cyclones (and upstream of cold spells). I see an example of such a (problematic) reasoning at lines 323-324, when it is concluded that "shifts in the location of waves relative to climatology can result in a cold air advection to the cold region": this formulation is quite ambiguous, because it might sound like the displacement of the orographically forced wave from its supposed "original location" is causing the cold spell. What would drive such a displacement? Probably not orographic wave dynamics, but rather the arrival of an upstream Rossby wave train leading to (thanks to constructive interference between different waves) an elongation and potentially a "shift" of the pre-existing orographic ridge: such a shift, however, would be fully due to extratropical Rossby wave dynamics.
3) In this regard I am also not fully sastisfied by the reply of the authors to my Major point 3. First of all, the references brought by the authors to support their statement that "planetary waves (also at mid-latitudes) can exhibit transient behavior" are also based on the same approach criticized here. Baggett and Lee (2015) define their long waves by filtering for wavenumbers between 1 and 3, and therefore using an analogous approach as the present study with the same problematic aspects (incidentally, they note that those waves feature "small propagation speeds"). Graversen and Burtu (2016) also define planetary waves using a wavenumber threshold (between 1 and 5, as in the current study). Incidentally, the conclusions that they reach that planetary waves are the main contributors to energy transport and high-latitude warming" might be fundamentally affected by the choice of defining planetary waves using a wavenumber-based threshold: at high latitdues, the length scales associated with transient baroclinic instability (which indeed is associated with net poleward heat flux) mostly project into that "planetary" range. Blackmon et al. 1994, on the other hand, state in their conclusions that "intermediate and long time scale fluctuations [...] feature little to no phase propagation", with no indication of transient behaviour of planetary waves at the day-to-day scale considered here.
4) The authors also perform an interesting analysis by restricting themselves to wavelenghts greater than 6000km to determine planetary waves. However, despite the limitation found especially at high latitudes (where three of the nine study regions are found), no modification has been brought to the manuscript to inform the reader.
Major comment about the methodology
I have a suggestion that would address most methodological comments. Why not performing again the tracking of troughs and ridges across the whole hemisphere, without being bound to the location of climatological waves? This would allow to focus on the characteristics of transient waves (and even moving "planetary" waves, in case they existed) during cold spells, simplifying the interpretation and solving some methodological issues such as the focus on six fixed areas for ridges and troughs. In case the approach needs such fixed longitude intervals, an alternative approach would be to define, for each region affected by cold spell, two fixed windows located 60° to the east and west of the central latitude and perform the tracking for each region separately. The interpretation of the result would then mention "ridges" and "troughs" in general around cold spells, and in my opinion align more closely with the proposed objective of targeting separately the properties of ridges and troughs around cold spells.
Specific comments
Lines 38-43: given that several studies have addressed the shortcomings of the Francis and Vavrus (2012) hypothesis, wouldn't it make sense to present this paragraph the other way around, giving more emphasis on recent work rather than to the initial, and in many aspects disproven hypothesis?
Lines 55-57: are there studies that claims otherwise? If yes, it would be great to point them out explicitly here so that the contribution of this study to the scientific debate becomes clear.
Line 78: the MEX corresponds to the squared standardized temperature anomaly and, thus, measures the overall temperature variability, regardless of the sign of the temperature anomaly. Thus, the top 1% would result in days with co-occurring cold AND warm extremes in the considered latitudinal band. Why not using standardized anomalies, then?
Line 81: follow-up on the previos comment: the fact that in Fig. S1 the chosen approach based on MEX results in composite patterns with low temperatures over the continents is presented by the authors as something obvious. Instead, hidden in those composites might also be days with specular temperature patterns, i.e., warm extremes over the continents (again, because the non-standardized MEX considers squared temperature anomalies regardless of the sign), potentially leading to confusing signals in the interpretation of the results. Have the authors excluded this potential problem?
Line 82: what is the advantage of using this two-step approach (through the MEX) in the definition of cold spells and of the study regions?
Line 95-96: this is an interesting statement, could you provide some more details on it? Or could this signal be due to cancellation between cold and warm spells during winter?
Lines 104-111: The rationale behind the choice of the regions is still puzzling to me. For instance, why is the region of the North American Great Lakes not emerging from your diagnostic, even though it is also a region anecdotically affected by cold spells? More in general, from the point of view of standardized anomalies sense, every longitude should be equally affected by cold spells: why only some end up selected?
Lines 116-117: In which parts of winter? And more i general, why do you think this is the case? Again, if the standardization is done properly, there should not be a preference for specific dates.
Line 123: the correction brought as "Planetary OR Rossby waves" is still ambiguous: which type of planetary-scale waves would not follow Rossby wave dynamics?
Lines 137-138: What if the effect of such a condition on the results? For instance, what happens if the condition is not fulfilled for a single time step? Are there many ridges and troughs that only last a few time steps? To check if this happens often, could you compute the average lifetime of the tracked troughs and ridges? Would the results change if, say, troughs and ridges with track duration lower than 24h were to be excluded?
Lines 139-140: This description sounds a bit "textbook" and not convincing. Phase speed corresponds indeed to the speed of individual waves and troughs, which emerge from the local combination of different wavenumbers (only in a mathematical sense, of course: the decomposition along a Fourier basis of circumglobal waves does not necessarily have physical meaning --see the example of a Rossby wave packet initiated locally by baroclinic instability, e.g., Teubler and Riemer 2016, https://doi.org/10.1175/JAS-D-15-0162.1). So, probably what we are looking at is quite similar to the phase speed: would the authors agree?
Lines 150-151: although I appreciate the clarification provided by the authors in the revision, it is still not clear to me how those wave features can do more than simply oscillate around their climatological positions dictated by, for example, large-scale orography (as it is the case of the WNA ridge).
Line 151: A follow-up to the previos comment: unless the tracking is rather done on the portion of transient, high-wavenumber waves in the range Z1-5: then, why limiting to specific longitudinal sectors?.
Lines 152-154: So, what happens once a trough in the ENA sector moves east of 30°W and then re-enters at 0°E the EMed sector? It is not the same trough anymore? Does tracking stop?
Line 157: this sentence is not so clear to me, does it mean that ridges and troughs form at high latitudes, then "persist longer" at mid-latitudes and move again towards high latitudes to decay? Has this behaviour been documented?
Lines 157-158: How is the EMed trough related to orography? Which orographical feature are the authors referring to?
Lines 158-159: This is also very interesting, as the genesis and decay rates are very low with respect to the climatological frequencies (I guess the authors refer to an occurrence frequency of 0.007, so to a 0.7% genesis rate per grid point --please double-check and adjust the legend of Fig. 4 accordingly). This lends further support to the hypothesis that (at least at mid-to-low latitudes) what is being actually tracked are the climatological, quasi-stationary features that mostly oscillate around their climatological positions, rather than propagate.
Line 164: is such a short life time over North Pacific related to transient storm-track activity, or to the rather zonal flow configuration indicating a reduced activity of "planetary-scale" waves thereby? If yes, it would be good to discuss why that region is "special".
Fig. 4: Why those sharp edges at longitudes 60°E, 0°E, and 180°E in Fig. 4a, and 30°E in Fig. 4b? The transition between climatological ridge and trough probability is much smoother and not bound by longitudinal sectors, on the other hand, see for instance over North America (as one would expect). Why do such patterns emerge?
Lines 179-180: does this still hold for regions at low latitudes (35N-45N)?
Line 188 and following: Please clarify in this sub-section which part corresponds to the Cattiaux et al. (2016) approach and which parts are the improvements/changes suggested by the authors, so that can be emphasized.
Line 200: which hysoipse? Please give it a name or use some symbols consistently across the article and the schematics, otherwise it is at times very difficult to follow the explanation.
Line 203: Does this mean that the threshold now is 90 and not 180 degrees? Is this an improvement with respect to Cattiaux et al. (2016)? If yes, please specify it (see previous comments).
Line 210-211: From reading the paragraph at lines 182-187 I thought the chosen approach would have circumvented this problem. How are cut-offs treated in the novel formulation?
Line 214: in their reply to Major point 3 the authors said that they "avoided the terminology of planetary waves" but they are still mentioned here. Please reframe to avoid confusion by the reader.
Lines 215-216: this is not clear because MEX conflates together cold and warm spells.
Fig. 5: Would it be possible to use the same example in Fig. 5 and Fig. 1 to improve the clarity of the explanation? Furthermore, would it be possible to compare the geopotential pattern shown here to the unfiltered pattern of Z300?
Line 220-221: here the authors are describing the shift of a climatological feature to the West. Which process is causing it, if we are not talking about Rossby wave packets?
Lines 216-235: How can the authors be sure that what is being diagnosed here is a shift of the "planetary" waves and not a projection of synoptic variability on those scales, especially at higher latitudes?
Lines 232-234: which processes governs those barotropic shifts? Could it be that what is being diagnosed here is atmospheric blocking, that features an equivalent barotropic structure and is associated with cold spells and other types of extreme weather during winter?
Lines 234-235: is this formulation a convoluted way to mention the process of "advection", or is there more to it?
Lines 259-260: It is very difficult to portray how such a propagation of the ridge at low latitudes is supposed to happen... could the authors provide for review a case study, among the chosen cold spells, exemplifying this behaviour?
Caption of Fig. 6: do the authors mean 0.2, 0.2% or 20%?
Line 270-273: I think the motivation provided here by the authors is a bit weak. By which process could a surface cold spell could cause the amplification and slowdown of an upper-level Rossby wave? The opposite chain of processes, on the other hand, is much more obvious through the process of cold air advection. Maybe what the analysis contributes to is to evaluate separately the different contribution of ridges and troughs? Please explain or consider reformulating for clarity.
Line 280-281: Has something like that already been discussed before in the context of cold spells? Otherwise this might sound a bit hand-wavy...
Lines 283-284: This theoretical connection seems a bit hand-wavy, as it cannot be tested. Why the need of disturbing finite-amplitude wave-activity flux theory, when one can rely on simpler explanations such as the PV-view of baroclinic instability as in Hoskins et al. (1986) that also involves a deceleration of the upper-level wave pattern?
Lines 285-286: I am still not convinced about the novelty of this statement and encourage the authors to be more precise: if the added value of the approach is to separate the contribution of ridges and troughs, then make this explicit here rather than generally talking about "waves".
Lines 290-295: I would be very cautious with such statements about the 850hPa propagation, because low-level flow evolution might be substantially affected by orography and many of the study regions involve large-scale orography (e.g., the first three regions are all very close to the Rocky Mountains) that can divert the low-level flow and locally alter the propagation of waves. How can the authors be sure that these effects are not due to orographical features? Could they separate regions with orography reaching 850hPa from the others and re-do the analysis separately?
Line 293: It might be inappropriate to use "somewhat baroclinic" just by judging from a 1-day lag. Also, what would be the physical meaning and the implications of such a difference in wave character between upper and lower levels?
Lines 296-310: these two paragraphs are not organized and overly speculative, with little references to previous research to back up the hypotheses. Please consider rewrite and delete parts that are too speculative (e.g., the connection with stratospheric activity).
Lines 315-316: the authors should pinpoint in the Introduction previous studies that raise such a question about the direction of causality between these two, so that the novelty of this work can be easily appreciated.
Line 319: cold spells in Europe (missing "in")
Lines 323-324: the sentence "shifting the location of waves relative to climatology can result in a cold air advection to the cold region." suggests that the waves considered here are the "climatological" waves, stationary waves due to orography, land/sea contrast, storm track activity. Please reframe also considering the major comments above. |
Thank you for preparing this manuscript. Please find the comments attached.