I remain torn on this contribution. As I said in my first review, it is thought provoking, presenting something new to the atmospheric dynamics and circulation community. I also appreciate the author’s changes to the text and efforts to be more quantitative. As detailed below, I still think significant changes are required. First, to moderate the implications, and second, to ensure that the schematic doesn’t imply detailed differences in the transport with height that may not be justified.
Major suggestions
1) My main concern from the first draft was the lack of quantitative analysis. This has improved in this draft, but the implications of the analysis are missing. In particular, the abstract suggests that the water vapor conducts “significant” mass transport and “influences the atmospheric momentum budget”. In terms of mass transport, “Significant” could be interpreted to mean large or to mean statistically distinguishable. The latter may be true, but not the former. The throughflow is 1000 times smaller than the net recirculation of the atmosphere associated with the transport of energy poleward. No evidence is provided to suggest that this 0.1% effect has a substantial impact on the momentum budget of the atmosphere.
To be specific, if the net moisture flux is of order 10 kg/m/s, then the total meridional mass transport is 2 \pi r_0 cos(lat)*10, or about 3 10^8 kg/s at 45 degrees. In comparison, the total transport of mass by the atmospheric circulation is approximately 1-2 10^11 kg/s (See Pauluis et al. Fig 2). Thus the latent cells are nearly 1000 times smaller. This scaling is consistent with the velocities, where I would say that meridional transport is of order 1 m/s, compared to the 1 mm/s for the latent cell.
Thus, to see the unclosed streamlines in a diagram of the meridional overturning, one would need extremely fine scale, accurate data to detect it. Speaking from experience, I think numerical errors in computing the meridional circulation will dwarf the error associated with neglecting the source sink flow, and it’s not a major issue that most past studies have assumed that the streamfunction has zero boundary conditions at the top and bottom.
Similarly, I think it's hard to argue that we should primarily think of the circulation as a latent cell, with familiar atmospheric features embedded within it. Rather, these familiar features are 1000 times larger than the latent cell. The latent cell can only be quantified with very careful analysis. The overturning circulation primarily moves mass poleward at upper levels / higher energy, with the return flow at lower levels / lower energy, and this is a 0.1 percent correction.
In the earlier draft, it was my understanding that the author argued that the required mass transport of inert species (specifically Ar and O2) by the throughflow was significant for their distribution. In the revised draft, am I correct to understand the primary influence on the distribution of these gases is dilution by evaporation, and the throughflow balances the diffusion down this gradient? (I include a specific comment about this below.)
All this is not to say that the latent cell isn’t interesting in its own right. It is transporting the same order of magnitude of mass of the upper branch of the Brewer-Dobson Circulation of the stratosphere. I just don’t think the latent cells need to be justified by implying there’s a significant error in the analysis of the meridional overturning circulation in text books, etc.
2) I regret that I didn’t pay close attention to the key Figure in the first revision. I would like to ask about a few features implied by the figure that are not discussed or justified in the text.
First, the tropical (red) cell is drawn in a way that appears to counteract the Hadley cell (e.g., counterclockwise in the northern hemisphere). Is there evidence that evaporated water actually moves *poleward* and upward before moving equatorward at height? I would have assumed that the moisture transport is primarily in the bottom branch of the Hadley cell, and where it primarily moves equatorward and slowly upward until it condenses somewhere in the tropical lower stratosphere. The dashed curves in the atmosphere indicate the return of liquid water, right? If so, should it be straight down in the atmosphere?
Second, is there any evidence of an equatorward transport of moisture in the extratropical cell, above the tropical cell that then shifts back poleward, as implied by the “S” shape of the blue cell in the subtropics? Naively, I would have thought that moisture is primarily transported slantwise (poleward and upward). As remarked above, I think the dashed lines shouldn’t extend further poleward than the solid (water vapor) lines, as there’s no mechanism for source sink flow past the point where the moisture rains out? If the dashed curves indicate rain, shouldn’t it fall straight down (at least on this scale)?
Lastly, I’m not an oceanographer, but is it true that the oceanic transport goes deeper in the tropics?
To be constructive, it might help to include a sketch of the tropopause. At first I was puzzled why the tropical cell was shallower than the extratropical cell, but I think this is meant to imply that the tropical cell is primarily shallow (thought tropical convection is deep:I suspect some condensation in the tropics occurs higher than in the extratropics). This could be emphasized by sketching the tropopause in the plot, to give a sense of the vertical structure.
In addition, I appreciate that this is meant to be a schematic, so perhaps making it more idealized would be helpful. If there is no evidence of the S-curve of moisture in the blue extratropical cell, perhaps just draw an idealized half cell, e.g., the top half of an oval, which connects with the return flow in the ocean below to form an oval.
In terms of the ocean I would only only show changes in depth of the return flow if there is evidence that there is definite vertical structure to the flow.
3) A final suggestion. It might be good to note that while the latent cell does show the net transport of water vapor by the atmosphere, it may not indicate what actually happens to a given water molecule. I suspect that moisture is continually being evaporated, rained out, and re-evaporated along the march poleward. The majority of this transport is effected by the baroclinic eddies (the diffusive transport), and so-called atmospheric rivers (the moist, poleward transport on the flank of a low pressure systems). These streamlines are thus not well captured by the Eulerian mean circulation, but by the TEM circulation (or the circulation in isentropic coordinates, as explored by Pauluis et al.).
I think my problem might be confusion over what is meant by diffusion vs. through flow. I think that water vapor is moved by eddies e.g., “diffusive” transport at line 26. Other inert gases (Ar, O2) but with net transport up the gradient created by dilution, which is balanced by diffusion of the dry constituents backwards. Don’t both directions involve diffusive transport by the nomenclature introduced in the paper,. Please pardon me if this reflects my confusion, and this comment is meant to help the author explain what is meant.
Small comments by line number
12 I do not feel that evidence has been shown that latent cells alter the momentum budget of the atmosphere. The meaning of “significant” mass transport is unclear, but given that it’s 1000 times smaller than the transport by the atmosphere, I’m unsure this is a fair characterization.
23 I am not sure what is meant by “bulk flow” here. Given correlations between temperature and velocity, the eddies move mass, as characterized by the TEM circulation. The key is that they are not random in the sense that poleward flow is associated with more energetic (warmer/moister) air, while equatorward flow is associated with colder/drier air. It is this correlation that allows them to transport mass and energy.
29-30 Could you emphasize that these flows possess no net *meridional* momentum when averaged vertically? I think this is what you mean.
Note that the Hadley cell does transport zonal momentum because the air in the poleward flow aloft moves eastward relative to the low level equatorward flow. I know the author didn’t mean this, but in the theory for the Hadley cell, the field normally focuses on zonal momentum.
36 I don’t mean to dwell on semantics, but while the latent cells are hemispheric in total scale, I suspect a lot of this mass flux is driven by baroclinic eddies. As I suggested in major comment 3, water is likely evaporated and condensed, and reevaporated a few times on the way from the subtropics to the tropics.
63 I am not sure if this is a good analogy for the atmosphere, where the other species (N2, O2, etc.) cannot spill out. Only water can be injected and removed. As the atmosphere doesn’t have a lid to spill, in this case it would just increase in mass until there came a pathway for the water to get out.
The mixed case is a better analogy, but again it’s imperfect in that only water is allowed to fall out (condense) at the top.
138-9 I agree that assuming the streamfunction vanishes at the top and bottom is incorrect, but as discussed in point 1, the error is 0.1 percent, and thus it’s a very good assumption relative to other uncertainties (e.g., how well we know the circulation in a reanalysis product).
196 Does the author mean the “net transport” of water vapour transport? This poleward transport of water vapor could be effected by baroclinic eddies, which the author has termed a diffusive transport. See my confusion as expressed in major comment 3.
212-222 Do I understand correctly that the concentration of Argon is largely determined by the dilution: when water evaporates into dry air, it must displace some of the dry air, so that the concentration of Ar (in kg Argon/kg air) is reduced. This creates a gradient in Argon, which will diffuse from cold/dry air to warm/moist air. Am I correct to say that this diffusion is balanced by the net return of Argon associated with the latent circulation through flow?
If this is the case, I don’t understand how one can say that the throughflow dominates the net meridional transport. In the climatological mean, there is no net transport of Argon: the down gradient diffusion must be balanced by the through flow.
214 Consider omitting “strong”, as the strength of this diffusion is never quantified. Strong relative to what?
225-8 Oxygen is diluted on the order of 0.1% (i.e., thousands of parts per million) by evaporation. This causes a diffusion downgradient, which must be balanced by the throughflow back.
I understand that the gradient of O2 in kg/kg is larger than the gradient of CO2 because there is more O2, but the relative effect of dilution on CO2 is the same, right? I suspect CO2 is likely more influenced by biology (and human activity, burning fossil fuels), which induce sinks and sources that are large relative to the dilution effect of water vapor.
234-242 Please eliminate these paragraphs. As the transport associated with the latent cells is a thousand times less than that associated with the TEM / isentropic circulation, I think its contribution to the momentum budget is insignificant. Uncertainties in surface drag and convective momentum transport would dwarf any errors associated with throughflow.
Conclusions section: Please see my first comment. I think the idea of latent cells is quite interesting and this could be a useful contribution to the literature. But I don’t think it’s fair to say that the “conventional” view of the atmospheric circulation (the Eulerian mean and Lagrangian mean features) are embedded in these latent cells. They are moving one thousand times the mass on the same hemispheric scales. |
The paper notes that there is net (water) mass transport by atmospheric circulation and argues that such a throughflow should be accounted for. While the fundamental observation is correct and the author raises some interesting issues regarding the distribution of trace gases in the atmosphere, the proposal to address it through a 'latent' cell is rather unclear.
Main comments:
1. Discussion of the atmospheric circulation:
The description of the atmospheric circulation as consisting of three cells superimposed on an eddy "diffusive" flux is rather dated. Multiple authors have provided insight into the global circulation and, in particular, on the role of the hydrological cycles (see Pauluis et al. Science 2008, and Lalilberte etal Science 2015, among others). There is also extensive literature on 'freshwater transport' by the ocean circulation (see Schanze et al., Journal of Marine Research, 2010).
2. What exactly is the author's proposal for a throughflow?
While everyone would agree that there is indeed net transport of (water) mass by the atmospheric circulation, this transport is small when compared to overall mass transport. The author proposed addressing this by adding a throughflow. Less clear is how the author would estimate it, or whether it would have a significant impact on circulation itself. (Abeit the author seems to concede that the impact of the throughflow would be minor, "the direct contribution of atmospheric throughflow to the meridional transport of water vapour itself is minor" on line 180.)
The issue here is that there is a simple fix to the problem raised by the author: Given that there is no mass transport of 'dry air', one solution is to compute the circulation in terms of dry-air mass transport. This circulation would be 99% similar to the circulation obtained using the 'moist air transport', and would not involve any throughflow . In this context, the paper would benefit from some modicum of computations. Reanalysis datasets are widely available and should be used to provide preliminary results.
3. The discussion of trace gas distribution is welcome and could be more detailed.
The more interesting part of the paper is the discussion of trace gas distributions at the end of section 4. The discussion is fairly informal and lacks a clear testable hypothesis that would motivate the use of a thorough flow. At one point, the author claims that "Throughflow therefore dominates the net meridional transport budget of argon in the midlatitudes, even though it is not readily apparent in conventional circulation diagnostics". While this is an interesting claim, it is not backed by any data or supporting evidence.