|Second review of “Sensitivity of deep ocean biases to horizontal resolution in prototype CMIP6 simulations with AWI-CM1.0” by Rackow and Co-authors|
Although the authors addressed the majority of the comments and suggestions of both reviewers, I think there are still a few lingering items that should be addressed before publication. The first concerns consideration of other metrics to access the overall impacts of going to higher resolution. The authors indicate that they added performance indices (PIs) for some ocean fields. However, as far as I can tell from the explanation in Appendix B, this effort only concerns 3D temperature and salinity fields. In my request, I specifically asked for AMOC, and northward heat and salt transports from all the cases. Perhaps one should also add the Antarctic Circumpolar Circulation (ACC) transport at Drake Passage to this list. I would like to see spatial distributions of AMOC from all the cases as well as line plots of northward heat and salt transports. Plus a table of ACC transports. Based on Fig. 12, I am expecting that AMOC will be rather weak in MR. Second, it is clear from Table 3 that the improvements are not really monotonic with increasing resolution. Specifically, going to HR and MR0 do not improve the solutions as far as the metrics in PI are concerned. So, statements like “We show that increasing ocean resolution locally to resolve ocean eddies leads to a major reduction in deep ocean biases.” are not quite correct without a qualifier. A third item concerns the temperature and salinity biases depicted in Fig. 6 for the Strait of Gibraltar region. Specifically, I suspect that those biases are simply reflecting a vertical shift / change of the Mediterranean Outflow waters by a few hundred meters. Because these plots are at a constant depth of 1000 m, such shifts are not captured. I suggest adding vertical profiles of regionally-averaged potential temperature and salinity to capture such shifts. Although these items are not necessarily onerous, I would like to classify this revision as major because they can impact the conclusions.
Fig. 1 and related discussion: Indicate how many CMIP5 models are included. Explain why DJF was chosen. Both here and elsewhere, change depth axis values to be positive because depth is positive downwards and negative values indicate a vertical coordinate that is positive upwards.
p.3, l.2: Please explain what “mapping” refers to.
p.3, l.33: Clearly state that the simulations are 100 years long.
Table 1 caption: “…. at 1000 m depth.”
p.4, l.11-12: This sentence is not clear. Please rephrase.
Figure 6: Either use K or degree C throughout the manuscript. This figure uses degree C, elsewhere it is K.
p.14, l.11-12: This is true for a 30-year mean. What about seasonal and inter-annual excursions? If there are such excursions to deeper levels, then this general statement is not correct which can change your conclusions.
p.14, l.26-20: Can be deleted as already discussed earlier.
Fig. 9 caption: Is there a missing URL?
p.18, last line: “…. climatology as the circulation and eddy fields are ….”
Fig. 11 caption: These are not maps of along-isopycnal fields, but rather fields on a constant isopycnal.
p.20, l.4: “is systematically reduced when moving to successively higher resolution ….” Not justified.