General comment on “Climate change signal in the ocean circulation of the Tyrrhenian Sea” ms by Alba de la Vara et al

The manuscript addresses future changes that could take place in the Tyrrhenian Sea (TS) circulation by the end of the current century under the “business–as-usual” high-emission RCP8.5 scenario. Authors use the regionally-coupled climate model ROM (from REMO-OASIS-MPIOM) to conclude that surface mesoscale patterns in the basin are slightly modified, which authors ascribe to changes in water transport across Sardinia Strait and mechanical energy transfer from wind field to the ocean in the vicinity of Bonifacio Strait. More interesting is the conclusion that the flow from the TS into the Liguro-Provençal basin will weaken and the advected waters will be more stratified, speculating about the possibly hampering of winter deep water formation in the Gulf of Lions by the end of the century.

While the objective of the manuscript is of interest, a question that arises is why authors made up their mind to select the TS instead of the critical areas of intermediate and deep water formation, focusing on these important processes. Such a study would provide a deeper insight on the future evolution of the Mediterranean Sea circulation. I would like to have read a sentence explaining the reasons and the pros of their selection against the other alternatives.

In its present form, there are some points that should be considered and revised or completed. My objections refer to the model set-up, mainly the (boundary?) condition at Gibraltar, and to the validation. The review only addresses these two aspects and it does not include specific comments, technical corrections or typing errors.

MODEL SET-UP. One of the ms main conclusions is the future enhanced stratification in the TS, which is the joint result of the expected SST increase in a warming ocean and the freshening of the surface water. Whereas the SST increase is an undisputed fact, the freshening is not. Obviously, it must be the result of a fresher Atlantic inflow through the Strait of Gibraltar, which has consequences on the whole Mediterranean Sea, not only in the TS. And obviously again, all the results of the modelling (not in the TS uniquely) rely critically on the validity of this boundary condition. Except for a vague sentence on lines 104-105 ("the water exchange at Gibraltar and Dardanelles in ROM is not parametrized and Atlantic water properties are not relaxed towards climatological values in the areas adjacent to the straits"), nothing else is said about this critical point, as if the future freshening of the Atlantic inflow were a proven result or another indisputable fact. Does this result come from MPIOM global model? Authors must be much more explicit about this point, discuss its validity and assess how sensitive their conclusions are against small changes of Atlantic inflow salinity . Similar considerations can be extended to the strength and size of the exchanged flows through the Strait of Gibraltar, which should be notably increased if the density contrast between inflow (warmer and fresher) and outflow increases. Could authors show the ROM forecast for these variables in the “Model setup” section and extend the section by addressing the conditions that hold at the western boundary of the Mediterranean. It is necessary in order to support adequately the conclusions of the paper and to discuss how robust they are against changes of those conditions.

VALIDATION is addressed in a very light way in Section 3 (RESULTS), where it appears in different places along with the interpretation of the model outputs. AVISO geostrophic currents is the reference for validation, which is carried out by comparing the thirty-year period 1976-2005 averaged circulation from ROM with the thirteen-year (1993-2005) period of AVISO. From Figure 2, it is difficult to conclude that both patterns agree satisfactorily, partially because of the poor resolution of the figure. Rather, the agreement seems to be only moderate if not arguable. But, even if it was satisfactory, the question that arises is why different time periods are used to average both data when ROM can use exactly the same period as AVISO. Why this mismatch of periods? Considering that the longer the averaging period the smoother the resulting pattern, the, for instance, disagreement between the mesoscale-rich circulation summer pattern in the TS from AVISO (Figure 3D) and ROM (Figure 3E) could be partially explained. In any case, the right way of comparing results is to make use of identical periods if possible.

Reasons provided in lines 136-140 to focus on the surface geostrophic circulation for validation purposes are strange. Actually, using surface geostrophic circulation and/or sea surface height (SSH) from AVISO is out of necessity: they are the only available variables.

The mention to the in-preparation paper by Parras-Berrocal et al. in order to justify the realistic representation of the main features of the surface circulation (lines 156-157) should be removed. If ROM is successful in doing this, the results should be either shown in this paper or referred to an already published paper, not to an in-preparation work.

Review of the manuscript “Climate change signal in the ocean circulation of the Tyrrhenian Sea”, by Alba de la Vara et al.

This work is devoted to the analysis of the future evolution of the circulation in the Tyrrhenian Sea (TS) under the pessimistic RCP 8.5 emission scenario. Using the outputs of the regional atmospheric-ocean coupled model ROM, the authors study the projected changes in the seasonal circulation patterns and the exchange through the main straits connecting the TS with the rest of the Western Mediterranean: the Sardinia Strait (SS) and the Corsica Channel (CS). The results show a weakening of the TS main cyclonic circulation and an enhancement of the mesoscale structures of this sub-basin. Authors attribute this changes to a reduced inflow of Modified Atlantic Waters across the SS and to an increase of the kinetic energy transferred by the wind. In addition, the results also show an increase of the water column stratification and a reduction in the water transport to the Liguro-Provençal basoin across the CS, which may impact the deep water formation process at the Gulf of Lions.

For my review, I’ll take advantage of the journal’s format that allows me to read the comments of my fellow reviewer. I agree with him/her in the interest of this work, but I’m also surprised that the authors have chosen the TS as area of study. Given the potential of the model to reproduce complex dynamics processes in the whole Mediterranean, I coincide with reviewer #1’s (R1) opinion that extending the region analyzed to, at least, include key processes like the deep water formation would have substantially increased the impact of the paper. Clarifying the reasons of this choice is important in my view.

Nonetheless, the results obtained are relevant and well presented, and the conclusions are well-argued. However, I have some concerns about both formal and content aspects of the current MS.

Formal aspects:

My main recommendation here is that the differences between the future and present climates would be much easier to understand if they are directly represented in the figures. Instead of only plotting the average for the present and future, if you include a panel with their differences with a positive-negative color bar the reader will be able to rapidly identify the regions where the different magnitudes increase/decrease. I understand that this is tricky for the velocity vectors, but for the scalar magnitudes it would be of great help.    

Contents:

Model set-up: In general, I agree with R1 that the description of the model should be extended, more considering that one of the key results strongly depends on the modification of the MAW properties. It is important to understand how the Atlantic Waters are imported through the Strait of Gibraltar and which are the conditioning factors for their lower salinity. For instance, is the salinity of the GCM used as boundary condition for ROM, MPI-ESM-MR, driving the properties of the inflowing waters?

Validation: Following my previous comment, Parras-Berrocal et al. (2020) show that for a hindcast of ROM forced by ERA-Interin in the present climate the average SSS in the Mediterranean is between 1 and 2 psu higher than MPI-ESM. This result should be commented here or in the previous section to contextualize the results. It means that, for the present climate, the GCM could be underestimating the salinity. This doesn’t necessarily mean that the projected freshening of the surface layer is wrong, but it is an important information in order to interpret the results. Comparing with the SSS projected by other regional models of the Mediterranean, particularly if they are forced by different GCMs, would also give more context to the results.

In the validation of the geostrophic currents using AVISO altimetry data, I also agree with R1 that the results shown in figure 2 are not as conclusive as the authors claim, and that the comparison should be made using the same periods for the average. Here a plot showing the difference between model and altimetry data would also be of great help to identify similarities/differences. A quantitative analysis, perhaps computing the spatial correlation in the area of study, would also help.  

Another point to be considered is extending the validation to include the seasonal cycle and the interannual variability, not only mean state for winter and summer, and including the SSS (as carried out by Parras-Berrocal et al. (2020) for the SST in the whole Mediterranean). This way if there is a bias in the SSS variability of the simulation in the present it could be identified and considered for the discussion of the results.

Finally, the AVISO interpolated products in the Mediterranean have strong limitations representing mesoscale structures (see Amores et al. 2018, JGR https://doi.org/10.1029/2018JC014140). I would suggest complementing the validation using a reanalysis dataset.

Results, discussion and conclusions: : In my view these sections are complete. Results are clearly explained and argued. My only suggestion is to include some of the referenced results, particularly those related to the previous work of Parral-Berrocal et al. (2020) and to the reduction of the deep water formation in the Gulf of Lions (I agree with R1 that a work in preparation shouldn’t be references). Also, in the discussion section, I miss a comparison of the results for the future evolution of the different variables with previous work for the region. Are they consistent with other modelling studies?