Review of the article “Dense water formation in the Eastern Mediterranean under global warming scenario” by Parras-Berrocal et al.

In this paper the authors study the evolution of the dense water formation in the Eastern Mediterranean (EMED) along the 21^st century, under RCP 8.5 greenhouse gases emission scenario, using the ROM Regional Climate Model (RCM). This RCM have been previously validated and used in several climate studies for the Mediterranean basin and sub-basins. They find a significant reduction of the deep water formation (DWF), between 75% and 85%, in the three regions were deep and intermediate convection take place (Adriatic, Aegean and Levantine basin) by the end of the century. The authors identify the increase in the water column stratification, due to the projected warming and salinization caused by the global warming, as the main factor driving this DWF reduction. They also predict a shift in the main Eastern Mediterranean Deep Water (EMDW) formation region, from the Adriatic to the Aegean Sea, similar to what occurred in the mid-90s with the so called Eastern Mediterranean Transient (EMT).

I find the paper very interesting. As the authors mention, this topic has not been studied in depth and it is very important to understand the expected changes in the Mediterranean thermohaline circulation as a consequence of the climate change. Although the use of a single model limits the robustness of the results, as the authors themselves point out, the results presented and the analysis of the mechanisms behind them are very relevant to the climate modeling community of the Mediterranean region. The manuscript is well written and organized, all the ideas are concisely and clearly stated, and well referenced. There are only a couple aspects that, in my opinion, need to be clarified before its publication in OS.

The first and more important one is the selection of the density of reference used to estimate the DWF rate in the model. As the authors themselves explain, the potential density they use in the model for the newly formed deep water in the Adriatic, Aegean and Levantine basins is slightly lower than the observed and reported in the literature. This is due to the lower salinity, and hence density, of the model respect to the observations in these regions (figure S1). Adjusting the reference density to the ‘model reality’ is a sound methodology, so this shouldn’t be a problem. It would be interesting, though, to identify these references (both for observations and model) in the profiles of figure S1. The selected reference corresponds with approximately 650 m depth in the historical period. However, the authors maintain the same reference densities in the future to compute the DWF rate evolution, which could have led to an underestimation. The profiles for the projections seem to show a general reduction of the density in the whole water column for the Adriatic and the Aegean regions (figures S2,3). This mean that the density of reference would correspond with a deeper layer, and that future deep water might be lighter than the present one. It is difficult to identify these differences in the figures, and very likely there will be no significant variations in the results, but in my opinion the authors should clarify this point in in the results or discussion sections. Would the DWF rate increase if a different density of reference is used for the future? Maybe including a third set of panels with the evolution of the density in figure 4 would be also of help.  

My second concern is that the authors did not describe the limits of the regions of each sub-basin used to estimate the average profiles shown in the figures and the DWF rates. As they point out in the introduction, the areas of the Aegean and Adriatic where the deep convection take place are very specific. Are the profiles and DWF rates estimated in these specific areas or in the whole sub-basins? When computing the volume of deep water formed every year, do they account for all the volume of water with densities higher than the reference in a specific region, in the whole sub-basin or in the whole EMED (maybe considering the possible spread)? The region selected could also modify the results, so I think this should also be clarified in the MS. Maybe you could include the basins limits in figure 1 (the color scale is also missing).

The authors investigated the variations of dense water formation (DWF) in the Eastern Mediterranean (EMed) through the twenty-first century under the RCP8.5 emission scenario for understanding the impacts of climate changes on the Mediterranean overturning circulation. Their results indicated that the dominant source of Eastern Mediterranean Deep Water (EMDM) shifts from the Adriatic Sea to the Aegean Sea during the 2005-2040 period. By the end of the century, DWF for the Adriatic Sea, the Aegean Sean, and the Levantine Sea all perform a pronounced decrease by 75%, 84%, and 83%, respectively, which is a result of hydrographic changes of surface and intermediate water and the associated strengthening water column stratification under the RCP8.5 emission scenario. The results shown are impressive and, as was pointed out in the manuscript, also fill in the gap of the DWF study in the EMed providing a more quantitative assessment than previous studies. The manuscript was also well-written and easy to follow. But some improvements may be needed before the publication.

1) The coverages of the Adriatic Sea, the Aegean, and the Levantine Sea should be specified and shown in a figure as results and discussions of this study focus on the DWF from these regions. Thus, it is important to provide the spatial extent of these basins, which can also help readers to understand the studied area better. In addition, as it was stated that the horizontal resolution of the model varies from 7 km to 25 km (which is a big difference, I think), it is better to show the computational grid as well.

2) Statistics analysis and parameters are needed. Firstly, the authors may need to provide p values for every correlation coefficient as they are important to illustrate the significance. Secondly, the 2040s was regarded as a time point around which sharp changes in SI and DWF (Figure 3) were observed. However, the author may need to provide a more convincing way to address this time point not just by the naked eye but using some statistical tools, like the non-parametric change-point Pettitt test (Pettitt, A.N. A non-parametric approach to the change-point problem. Appl. Stat. 1979, 28, 126–135).

3) Could you double-check the unit “Sv yr” which first appears on Line 113? If my understanding of the unit “Sv” is correct, Line 113 should be rewritten as:

During 1981-1999, ROM_P0 produces a total of 5.45 Sv of newly waters denser than 29.0 kg/m3 corresponding to an annual formation rate of 0.29 Sv…

4) Lines 202-211. Although the authors provided descriptions of SI for different periods, I am still not quite sure how the authors calculated the percentage contributions of different water bodies to the temporal changes in SI. Could you please provide some descriptions or equations to further address the calculation?

5) Lines 268-270. It may be a jump to conclude that the increasing potential density is caused by the increasing salinity over the upper 100 m, as the authors only compared the salinity changes and density changes (Figure S7) but ignored the contributions of temperature changes. As shown in Figure 4 subsurface (0-100 m) temperature seemly performs an increase in the ADR (Figure 4a) from the period of 2006-2020 to the period of 2020-2040 but fluctuates in the AEG (Figure 4c) and LEV (Figure 4e). Thus, the author may need to quantify both contributions of the changes in temperature and salinity to the changes in density.

6) Lines 270-271. The authors may need to provide more evidence in addressing the causes of the changes in the upper ocean circulation, like correlations between changes in salinity or temperature and changes in circulation patterns. Or to provide some mechanistic explanations on how the changes in salinity or temperature would lead to changes in circulation. Or to provide results of previous studies here that may have such discussions.