This is an interesting analysis of the variability of the MOC expressed in sigma coordinates, separating a velocity-driven and a volume driven component. ALl my comments/questions are 'minor', and only provided to clarify the already very well presented analysis and results.

The discussion is made along the Ovide line, but could as well have been made on more zonal sections such as OSNAP-E, or further south. The AMOC strength is calculated from the surface downward, which for the CORA and EN4 estimates gives a very strong weight to the altimetric product (the same one for the two products which might also explain partially why the time series seem rather similar between the two products despite large differences in the density field in key areas such as near the western boundary). Although this is not a core issue for this paper, I am wondering how accurate the altimetric product is for the currents close to Greenland, where one could expect larger errors in the geoid time changes, during this period of often large (at least seasonal) south Greenland ice melt (but with decadal/interannual changes), and whether this could have some impact on the reconstructed AMOC variability in some of the analyses (my guess is possibly not that much, as this is an AMOC in sigma coordinate, thus with shallow upper layer in the western part of the section). The period chosen in 1993-2021 starting thus with the advent of more precise altimetric sea level products, but before the Argo array (or its predecessors in the North Atlantic) .

The choice of the different products compared is relevant, and spans a wide range of approaches. In particular, I find interesting the differences between ENA4 and CORA, which basically use the same data with some objective mapping, the largest differences been in the scales retained, and what is done near bathymetry and shelves. The two products could also differ more before 1997 and the ACCE sampling in this region (and even up to 2001, before the Argo float network became fully operational in this region). It could be interesting to mention that ECCO V4R4 as far as I know is a state estimate everywhere, except the Arctic, as well as the year of beginning and end of that run (as this is a state estimate, the adjustment is done over the whole period of the run, and meridional boundary conditions play a role).

The decomposition is interesting and suggests a difference of contributions between seasonal variability and decadal variability (different in ECCO). Interestingly, for seasonal variability (driven by the sigma change component), it seems that it is the component of the EGC ( a decrease in winter) which is the main contribution (due to denser waters there in winter, thus less volume transported southward). I am wondering how sensitive this is to the resolution of the mapped analyses. Are we sure that they do a good job separating what happens on spatial scales typically of 100 km for the seasonal time (typically for EN4 and CORA, the mapping scales are larger than this 100 km: does this play a role, in particular for EGC?).

On decadal scales, the main result is a strong contribution of psi’v to the variability except for ECCO. However, I wonder whether its strong contribution relative to the sigma one before the early 20002 (Figure 5) could be for a period less constrained in the analyses by the density profile data, thus at least in CORA, an underestimation of the density-induced variability. The anti-correlation between the two contributions is also well explained, and expected due to the distribution of where the large density changes occur. ECCO on the other hand exhibits a rather different structure of density anomalies. For me, this clearly points to a deficiency of the state estimation of the density field, and maybe also of the circulation in ECCO (the subduction of the density anomalies in the thermocline to the southeast).

Other 'minor' comments:

1. 193. As important as reporting the mean value of the mean transport would be reporting the seasonal or decadal variability of this net transport which is small for ECCO, but seems to me rather large in GloSea5. Does this net transport variability have a large impact in the decadal variability discussed later.
2. 263: ‘…, the MOC strengths (-0.23, 0.23)’. I suspect these differences are estimated at the time of the Ovide sections, and then averaged. It could also be interesting to report the uncertainty on these average differences (I guess roughly (rms(MOC) with hydro)/sqrt(11)). It seems that large differences might stem from the section in 2006. Are these differences within the error bars of the psi(hydro) estimate for that year? (or is there something special that might explain that apparently larger difference with all other products (less so, for ECCO, except if one corrects for the time series mean difference)
3. 274: ‘largely similar data sets…’, I agree although CORA and EN4 also use Ago float data, which will have some weight, even at the time of the Ovide Cruise. Altimetry and hydrographic profiles are also used in ECCO and Glosea5, but apparently leaving much larger biases with the Hydro estimate (nicely discussed in the next paragraph).
4. 283: Based on Figure 2, it also seems that ECCO does not have an ERCC (which is or is not a cause for a lower NAC, and in itself would contribute the other way for AMOC)
5. 291 (in Figure 1’s caption) ‘Biases with hydrography’ should probably be ‘Biases with respect to hydrography’
6. 299: ‘Positive velocities are directed northwards’. I thought that in Daniault et al’s paper the velocities were the component normal to the section (thus usually not northward, except close to Portugal).
7. 459: what does ASTE refers to?
8. 497: the sentence ‘NADW belongs to…’. This sentence is not very informative, and somehow does not deliver a point in the discussion? (just that it is not in what is estimated here for the upper limb?)
9. 518-519: ‘We note, however that…’. Yes, but Chafik et al (2022) include the density anomalies up to 1000m. Although the lower part is not directly related to the anomalies of transport in the upper limb (locally), its contrast with what happens further east would be important to that depth… It could be interesting to remind what is the period investigated and on which time scales were the main results of Chafik et al (2022) established. Is the result of Chafik et al (2022) (no direct relation to local wind forcing …) relevant and why does it seem to differ from the conclusion of this paper on the role of local forcing (although not shown, just argued for…)?
10. 538: maybe ‘to average conditions’ instead of ‘to the mean’
11. 540 (and figure 11 caption): ‘Our time series’? To which time series does this refer to (also to be mentioned in Figure 11 caption; maybe also clarify that on lines l. 523, as it is not crystal-clear which product and time series are considered; it might be ‘all’ (the four products), in which case, maybe do not use ‘our’; also, is figure 11 an average of the four time series, or one in particular) . This might have been mentioned earlier, but the reader might lose track of it.