|Review 2 of:|
Andres and Tarasov: Towards understanding potential atmospheric contributions to abrupt climate change: Characterizing changes to the North Atlantic eddy-driven jet over the last deglaciation, Climate of the Past Discussion
The authors have made substantial improvements of the text from the first round of review, and I am glad to see that many of the suggestions and comments from the referees have been incorporated. I recommend that the manuscript is accepted, subject to a few minor edits.
My impression of this manuscript is generally positive, however I can’t help but come away with a feeling that we haven’t really learned a whole lot from this study that we didn’t already know before. I hope I am not too unfair. The main takeaway seems to be that the LGM was a different climate state than modern, but the transition between these two states appears to be both model and forcing dependent. As you have shown, Plasim and Trace agree on some aspects of this transition, but disagree on other aspects, and it is hard to say which simulation (if any) is more truthful. I certainly don’t think that Trace reflects reality despite using a comprehensive fully-coupled circulation model (the simulation has several known problems, many of which are pointed out in the text), but I am not convinced that the Plasim results are necessarily more accurate because of the simplified model code, general approach, and apparent issues reproducing the CMIP5 mean climate at the LGM. Although this study certainly takes a stride in the right direction, more studies of this kind are necessary, and we can only hope that both Palmod and iTrace will shed more light on this important issue in the not too distant future.
Response to rebuttal:
In the first round of review I raised a concern about the claim that the subtropical and eddy driven jets entered a merged state in the North Atlantic at the LGM. The authors doubled down on this claim in their rebuttal, and pointed out a number of examples to support this assertion (it should be said that they removed the references in the manuscript text to make it clear that this is their own interpretation). However, I maintain that the sentence on page 4, line 19 is problematic and should be removed from the manuscript because: (1) it is demonstrably wrong (see below); (2) it gives a false impression that may be perpetuated in other studies since there is a most unfortunate tendency nowadays of mechanically quoting earlier studies without looking up the original source.
First of all, the examples from the lower troposphere (everything below 500 hPa) can be dismissed right away, since the subtropical jet is by definition an upper tropospheric entity (it is primarily due to Coriolis acceleration in the poleward branch of the Hadley Circulation) and therefore has no signature (actually often negative u-wind) in the lower troposphere. Second, the cross sections pointed out in the rebuttal clearly show a wind maximum in the subtropics. For example Fig. 4 in Merz et al. (2015) shows a very distinct secondary wind maximum in the subtropics (as shown by the shading). The subtropical jet is perhaps weaker than in the PI simulation, but it is still there and is clearly separated from the eddy driven jet in midlatitudes. The other studies cited in the rebuttal show the same thing, and, incidentally, also Fig. 5 in this manuscript, even though the wind profile here looks a bit suspicious (probably explained by the small number of model levels and perhaps also the simplified modeling approach).
Page 1, line 5:
Should be “intermediate complexity model” instead of “intermediate resolution model”. T42 is a pretty low resolution by modern standards, but it is much more important to point up that Plasim is very simple compared to the majority of the PMIP models.
Page 3, line 18:
More spatially localized?
Page 4, line 19:
See comment above
Page 5, line 10:
I would say “simplified physics parameterizations” to be more precise
Page 6, line 1:
There is no such thing as wet primitive equations. The dynamical core of Plasim is identical to PUMA, which is a traditional primitive equation model (dry dynamics). Plasim also includes a simplified physics parameterization with equations describing water dynamics.
Page 12, line 12:
Fig. 5 shows the whole troposphere so low level must be a typo
Page 13, line 6:
Sentence starting “The wave...” is note clear. Equatorward propagating waves tend to break anticyclonically, which yields a poleward momentum flux convergence and thus a meridionally shift in the jet structure. However, since you don’t actually show any dynamics related to Rossby wave breaking, I would recommend staying away from this topic since it adds an unnecessary level of complexity that is: (1) speculative at best; and (2) only tangentially related to the topics discussed here.
Page 16, line 22:
What do you mean by “preferred latitude”? Explain
Page 16, line 32:
What? Why do you get a split jet long after, but not during the LGM? This peculiarity is opposite to what other studies have found and should be explained.
Page 21, line 3:
The wording here is clunky and hard to parse (I had to read this sentence several times before understanding your point). Simplify and say that you use modern topography with glacial surface albedo.
Page 23, line 21:
Point here is not clear. The abrupt circulation changes in Trace are directly linked to specific (i.e., key) changes in the planetary boundary conditions --- as shown here, and also discussed in earlier studies. However, since you use a different boundary condition, it is only natural to see circulation changes at different times than in Trace.
Page 23, line 29:
Well yes, although finite grid spacing is an inherent limitation of all models.
This argument is also halting because, as you pointed out earlier in the manuscript, the jet can transition smoothly between two distinctly different states, even on a comparatively coarse grid. If the jet maximum changes from being in one grid cell 100% of the time, to a different grid cell 100% of the time, but occupies both grid cells (or all cells in between if they are spatially separated) with some distribution (say 50% in each) for some period of time, we would perceive it as a smooth transition. On the other hand, if it changes from one grid cell to the next in a discrete jump, most people would perceive it as an abrupt transition. I don't think this is particularly controversial. Also, what constitutes a smooth or abrupt change is obviously context sensitive. A change over a few decades or a century can be viewed as abrupt in a 21,000-year simulation, but would probably be viewed as a more gradual change in a shorter simulation.
Page 25, line 2:
Obviously, it is meaningless to talk about wind inside the ice sheet... I am glad to see that you didn’t extrapolate the Plasim and Trace data where the surface topography intersects the pressure level of interest.
Page 25, line 6:
That is a reasonable hypothesis. However, I am not sure that the lower 1000 m are as important as the authors claim. Ice sheets generally have steep margins, which is something that climate models cannot resolve properly because of the finite grid spacing. I therefore think that this argument should be downplayed a little bit because it is contingent upon grid resolution, method used for discretizing the equations on the grid (horizontal and vertical), how sub-gridscale topography is incorporated in surface drag parameterizations, etc.
Page 25, line 15:
These are probably not mutually exclusive
Page 25, line 2:
Possibly, but the fact remains that there appears to be a key change in the boundary condition --- both in Plasim and Trace --- that yields a disproportionally large change in the circulation.
Page 26, line 4:
This information is important and should be moved up to the description of the simulations and how you analyze the results.
Page 26, line 9:
Perhaps an argument for using middle to upper tropospheric winds instead?
Page 26, line 25:
Trace was run with CCSM3, which includes both a dynamic ocean and sea-ice model...