Review [December, 2021]:
Rohschenider et al., The Depth Scales of the AMOC on a decadal timescale
Overview
Revised manuscript is much more coherent, though still not easy to read. It argues that in a global ocean model, the Southern Ocean wind stress controls the Atlantic pycnocline depth, but that the Northern Hemisphere wind stress also influences the thickness and magnitude of the upper limb of AMOC in the northern hemisphere. I recommend one more set of modifications before publication, but I now believe it should be eventually publishable after one more revision.
Main Comments
A. Time Dependence. One essential revision is the need to characterize the time-dependence of their results. While extending the model runs may be prohibitively expensive due to the resolution, the paper should at least talk about the time-dependence for the 30 years of the run. Is there any evidence from the behavior of the overturning or stratification that the 20-year averages taken here would be similar if the next 20 years were used? Is there any evidence that the model is converging so that results may be similar if the run was extended another few centuries? While the paper explicitly says it is only talking about a particular time period, the theoretical framework of the paper is based on steady-state behavior, so its incomplete to not comment further on whether the results are relevant to the steady-state. And if the results are to be relevant to the transient behavior, then the time evolution has to be discussed.
B. Longer low-resolution runs. The paper would be better if an additional set of long runs were done with a non-eddy-resolving grid. This would tell us if the eddy resolution is important for getting the correct sensitivity to wind, and would give further insight into whether the results are indicative of long-term means. This is plausibly beyond the scope of the current paper, so I won’t insist on it, but at least a second paper to check this would be worth considering.
C. More on Dynamics. I simple way of looking at the effect of Southern Ocean wind stress is that it’s Ekman transport pushes water northward, and the resulting current joins the upper limb of the AMOC and returns with NADW. Shouldn’t the northern hemisphere wind do the opposite? Isn’t that what happens in previous studies? Ekman transport from strengthened westerlies goes southward, which would weaken the upper limb rather than strengthening it. Similarly, that same stronger wind in subpolar gyre would increase upwelling, which might counteract the downwelling associated with deep water formation. It would be helpful if the paper addressed this point. In addition, a key result is that we can think of a change in NH wind moving the depth of maximum streamfunction vertically, which moves a fixed vertical velocity shear vertically, which then determines the change in the overturning. But why should each of these facts (change in depth, constancy of velocity shear) be true? Answers to these questions would enhance the paper though they are not required for publication.
Other Comments
1. Improve Table 1. Make table less wordy and leave out information common to all 3 experiments, so that table looks something like this:
Abbreviation Name Description
1X Reference Observed wind stress
2XSH Double Southern Wind Double wind stress south of 30S
2X Double Wind Double wind stress at all latitudes
2. Improve Table 2. Separate and organize variables into groups, make separate “parameter name” and “parameter definition” columns. At the authors’ discretion, I suggest using lower case and upper case symbols to differentiate between variables and parameters, using a capital Z or H rather than lower-case η for depths, referring to “level of no motion” as “Streamfunction depth” or “upper-limb depth”, and referring to “pycnocline scale” as “pycnocline depth” (because it refers to a measurement of model behavior, unlike the advective depth scale which is an estimate calculated from forcing parameters.” I don’t see the need for defining derivatives of ψ in the table because (for instance) ∂ψ\/∂z obviously means the “vertical derivative of the overturning Streamfunction”. Something like this:
Symbol Name Definition
Variables
x,y,z Coordinates Zonal, meridional, and vertical distances
ψ(y,z) Streamfunction Based on zonal integrated meridional velocity
ρ(x,y,z) Potential density Standard definition
Streamfunction Scales
Ψ(y) Maximum Maximum at each latitude
Ψ_g (y) Max. Geostrophic Like Ψ but based on geostrophic streamfunct.
Ψ^* Max. Estimated from Z_ψ See text
Depth
Z_ψ (y) Streamfunction Depth Ψ(y,Z_ψ )=Ψ(y)
Z_ρ (y) Pycnocline Depth Integral depth scale of ρ
Z_w (y) Advective Depth Scale Estimate based on Ekman pumping
3. Clarify model spin-up (Sec 2.1). Paper says “we focus on the time-window 1991 to year 2010”. How long before that (if at all) was the model spun up? How close to steady state was it at this point? What does “change the monthly-mean climatology of the surface wind stress only” mean? Does run 1X use daily wind stress, and other runs use daily wind stress multiplied by a factor, or is there some more complicated procedure involving taking monthly means? Or are monthly mean wind stress used for all runs?
4. “Mid-depth” is confusing. In Abstract and elsewhere, I suggest replacing “mid-depth AMOC” with “upper limb of the upper AMOC cell” or “upper limb of the North Atlantic Deep Water (NADW) cell of AMOC”.
5. Density Difference Figure. I don’t understand what is gained by looking at density differences in Fig 3ab. Since the reason for examining density is connected to measurements of isopycnals depth, why not just look at isopycnals? Also, rather than using the normalized density (black curves in Fig 3ab), it would be better to use a measure that is closer to the one used to calculate integral depth scale
r(y,z)=(∫_z^0▒(ρ-ρ_r )dz)/(∫_(z_T)^0▒(ρ-ρ_r )dz)
Which by definition gives r=0 at the surface and r=1 at z=z_T. Can then show separate panels for r contours for each experiment.
6. Focus on latitude-band averages for depth. Figure 6 shows averages over latitudes 10-30o in both hemispheres. Extend this to several quantities. Instead of Fig 3c and Fig 4 showing latitude dependence, just show averages for each of the 3 depth scales in each hemisphere. Each panel would contain depths (y axis of panel) for 3 runs (x axis of panel) for northern (upward-pointing triangle) and southern hemisphere (downward-pointing triangle) for a single quantity (streamfunction depth, pycnocline depth, advective depth scale). This would emphasize how each quantity depends on the wind, rather than current version which emphasizes complicated latitude dependence which text does not comment on much. Also Fig. 4 currently is very busy with 6 different curves and one has to concentrate to see the point about the wind different wind dependence in NH and SH. The latitude dependence of η_w is not an appropriate value to plot, since η_w is a scale quantity representing the pycnocline depth for a given gyre, not the detailed geographical variation of pycnocline depth within the gyre.
7. Why is advective depth scale included? Currently fig 3. Plots η_w (y) with different values of g'. The values seem arbitrary, and I don’t understand why these alternate calculations are graphed. The only significance I can see of η_w is that it depends on √τ. Therefore, maybe just compare η_ρ variations to √τ in the plot I suggest in (6) above.
8. What is the significance of the geostrophic transport? Below the Ekman layer (top 50 m or less?), shouldn’t velocity be geostrophic? Or does nonlinearity from the eddies add an important term? The max geostrophic streamfunction shown in Fig 5 is some kind of perturbation due to the Ekman transport? How is it relevant to the discussion of the overturning? If it isn’t, why is it discussed?
9. Maybe separate transport and depth data. Since I think the depths should be plotted as a function of experiment, maybe the NH and SH transports should be plotted that way as well rather than plotted against depth as in Fig 6. Then again, the plot does do a good job showing that volume transport varies with η_ψ, so I wouldn’t object to keeping it anyway. The dashed lines are distracting though; if graph kept as-is, eliminate them and perhaps use a more distinct symbol if geostrophic data is retained in revision – perhaps open symbol instead of lighter symbol. The transport estimated from shear and η_ψ, currently shown as a function of latitude in Fig 9, could also be included in the figure (10-30o average for each hemisphere as a function of run).
10. Which part of Fig 7 is new? Significant parts of the discussion around Fig 7 seem to just be remarking on the fact that changing the wind stress changes the subtropical cells, which follows from the nature of the cells. If there is something else that the discussion is trying to say about the system, this should be made clearer and more distinct.
11. Abstract Clarity. The Abstract is okay as written, but has a number of awkward elements. Here I list those elements and give an alternative text for the first 2/3 of the Abstract. The authors can use all, part, or none of the alternative text at their discretion.
“wind forcing dependencies” is a little vague
“level of no motion as the depth of maximum overturning” is trying to say that the 1st phrase = 2nd phrase, but readers may be confused by “as the”
“interplay of nonlocal and local” also kind of vague – at this point reader still doesn’t really know what abstract is talking about
“downwelling region where Ekman pumping takes place” Actually the wind is changed over entire hemisphere, so not clear that it’s the Ekman pumping location that is key
In my rewrite, I try to give the reader a bit more context first, and to describe the issues and experiments in a more concrete way.
The strength of the North Atlantic Deep Water (NADW) cell of the Atlantic Meridional Overturning Circulation (AMOC) is commonly linked to the strength of Southern Ocean zonal wind stress, which influences global pycnocline thickness. Here we study the separate influence of wind stress in the southern and northern hemispheres by running an ocean general circulation model (OGCM) with different values of wind stress. The experiments are 30-year simulations with an eddy-resolving version of the Max Planck Institute Ocean Model (MPIOM). While southern hemisphere wind strengthens overturning and thickens the pycnocline throughout the Atlantic, increased northern hemisphere wind strengthens and deepens the upper limb of the NADW cell in the northern hemisphere without strengthening it in the southern hemisphere or changing the pycnocline thickness in either hemisphere. Thus, North Atlantic overturning is affected by both remote forcing from the southern hemisphere and local wind. The thickness of the upper limb is a better proxy for overturning strength than pycnocline thickness. |
