Review of: Long-term Variability and Trends of Agulhas Leakage and its Impacts on the Global Overturning”, by Großelindemann et al.

In this paper, the authors analyze Agulhas Leakage and its drivers and impacts in a suite of eddy-resolving coupled climate simulations. They find that Agulhas Leakage is well represented by the model, according to several metrics. They find positive correlations between Agulhas Leakage and several metrics of wind stress, and find that this model projects an increase in Agulhas Leakage in response to an aggressive forcing scenario, despite a weakening of the Agulhas Current itself.

I found the paper very well written, and a pleasure to read. The methodology is clearly described, the analysis is convincing, and the results are interesting and relevant. I do have a list of minor comments, which will require only minor revisions.

Minor Comments

l. 125: employes -> employs

Fig. A1: I think it would be better to make subplots for the piControl and FOSI simulations. It is hard to see the individual curves representing the FOSI simulations. Besides, the variability in the two sets should not be expected to be identical, so combining them in the same plot does not make much sense. This plot is important, as it allows the reader to judge the accuracy of the method applied. Despite the muddled mess, the three FOSI curves don’t seem to track each other very well. 

Fig A2: This figure does not show a black rectangle, as the caption claims. 

L. 158: plausible -> would accurate be a better word here?

ll. 207-208, 233-235: Duplicative. 

l. 227 and elsewhere: Fov is the freshwater flux /induced by the overturning circulation/. 

l. 244: Uptream -> Upstream

l. 265, 285, 276: Unless you can find a way to make a more robust significance estimate of this spectral peak at 14 years, I would not put that much emphasis on it; especially if it is only one estimate that sticks out, instead of a few adjacent estimates. At 95% confidence level, one is to expect 5 ‘false positives’ for every 100 measurements.

l. 273: So minimum zonal wind stress represents the easterlies in the subtropical belt?

l. 283: Filtering might also play a role in spreading out the signal. 

Figure 6: I find the correlations between Fov and AL suspicious, as they are significant stronger than -0.4 for lags between +/- 7 years. What does that mean? Do both time series have decorrelation times of several decades? The paper says that the time series have been detrended, can you confirm?

l. 294 and beyond: I’m wondering if this argument could be taken one step further by actually calculating an appropriate east-west gradient (upper ocean pressure, SSH, or maybe even the depth of the 10 degree isotherm) and comparing that to AMOC strength at 34S.

309: stable -> stabilizing, negative?

l. 321: I think it is important to mention the trend over that same period for the control simulation, since I suspect it is not much smaller than that of the historical + future simulations. If even the control has a significant trend over that period, then I suspect that that will modify the conclusion.

l. 362: are -> is. Essential for what?

l. 358: It may be worth noting that the Fov as a metric of the salt-advection feedback would asymptote to zero (from a positive value) upon a decreasing AMOC. The fact that it crosses zero suggests changes in the stratification at 34S, giving credence to the conclusion here. Is there a way to quantitatively compare Fov with the salt flux induced by AL?

l. 393: Correct parentheses around reference. 

l. 397: This conclusion is more or less unless contradicted a few lines later, where it is claimed that no distinction can be made between Rossby waves or propagating rings. I don’t see it as a problem that we don’t quite know the dynamical character of these propagating signals (maybe they are one and the same!), but it would be good to be consistent. Also, l. 303 acknowledges a potential role for winds, which is missing here.

l. 403: It may be semantics, but I’m not sure if Agulhas Rings can be classified as a mesoscale eddies. In my mind, they have a different character and dynamical origin, not in the least because of the barotropic nature of rings that contrasts with the baroclinic character of eddies.

l. 427: conductive -> conducive. Correct double ‘of’.

ll. 468-469, 16-17, 450, and other places: I’m uncomfortable with the strong statements that are made regarding the links between Agulhas Leakage and the stability of the AMOC and potential for collapse. Even though I think that the link between Fov and AMOC (bi-)stability is a compelling theory, there is still a lot of work to do to confirm this theory (for instance, by demonstrating it in an eddy-resolving climate model). I personally would not go beyond a statement along the lines of ‘with potential implications for the stability of the AMOC’.