Review of “MESMAR v1: A new regional coupled climate model for downscaling, predictability, and data assimilitation studies in the Mediterranean region” by Storto et al.(2023)

I have completed the review of the manuscript entitled “MESMAR v1: A new regional coupled climate model for downscaling, predictability, and data assimilitation studies in the Mediterranean region” by Storto et al. (2023)”. Here the Authors analyzes a newly coupled model developed for the Mediterranean region, MESMAR v1. In general, I found the outcomes of the manuscript interesting and the work falling in the scope of the journal. However, I found several issues in the manuscript that I require the Authors to address before the publication of the work. Thus, I ask in this case major revision for the manuscript.

Abstract

 Line 17: I would report the horizontal resolution of the models described rather than using the adjective “moderate” or “coarser”

Line 21: Are you planning to use the model to produce climate projections? Please specify: The Authors use several times the word “predictions” but never “projections”

Line 23: “Intense Mediterranean cyclones “rather than “intense mid-latitude cyclones”

Introduction

 Line 38: Usually RCM refers to Regional climate models and thus atmosphere not to coupled systems (RCSMs for example, Reale et al., 2022). Please correct that in the text to avoid confusion

Line 38: Please update Giorgi, (1990) that is a bit old as reference

Line 51: “Coupled…unexplored”..see Sevault et al., 2014. The coupled CNRM model uses the spectral nudging.

The Authors use several times the word “predictions” but never “projections”. Are you planning to use your model also to produce projections? Please specify

Line 72: Be cautious since hurricanes are (from some points of view) very different from medicanes. See Flaounas et al., 2022. Please reformulate this sentence

In the introduction it is missing a clear description of the present state concerning the coupled models in the MED region and at which extent the new modeling system described by the Authors is different or eventually represent an advancement with respect the preexisting modeling tools.

2.Earth system model configuration

 Line 94: What do you mean with stationary geophysical fields? Is the topography a geophysical field? Please explain

 Line 101: Do you refer to the width of the sponge layer? Please specify

Line 110-115: As far as I understand you are using shortwave radiation as forcing that attenuates along the water column according to a water attenuation coefficient plus the chl-a concentration: since you are using the satellite chl-a (first 10 m as far as I remember) how do you quantify the attenuation effect led by chl-a below 10 m?

Line 117-120: Are you using an open boundary in the Atlantic? Are you applying also the sea surface height at boundary?

Line 126-128: Is the Nile missing from the numerical settings? It is not mentioned in the text.

Line 138: Is “25” resolution dependent? Please explain

Line 143: Do you mean that all the model components exchange field every 30 min? please specify

Line 148-150: Not very clear. Please reformulate. As far as I understand WRF transfers surface and subsurface runoff to HD that is remapped on the ocean grid and passed to NEMO after HD runs. Does HD pass only the river discharge to NEMO? Do you need to locate the river mouths on the ocean model domain?

Line 152-157: Why do not use CMEMS MYOCEAN reanalysis specifically tuned for the MED sea instead of GLORYS? Moreover, there are no information about the spin up of your model for pot temperature and salinity. Did you perform the spin up at least for the long run? I would also include additional information concerning the computational performances of the model (number of cores, computational times etc.). It would be useful having a table summarizing the main settings of the model at least in the most important experiments.

3.Sensitivity experiments

 3.1 Impact of the interactive river discharge

 I do understand the idea of the Authors to show the importance of river online on the simulate salinity. However, I think that 2 years of runs are too small to assess the importance of river inflow on salinity (in particular along the water column where the signal need sometimes to penetrate) in absence of information about the spin up. I would change Fig.3 and Fig 4 adding the comparison with observations (instead of the comparison between the two configurations) as You did in Figure 6. Moreover, why did you use EN4 instead CMEMS or JRA55 instead of Ludwig et al., 2009.

3.2 Nemo vertical physics

 There are no information about the length of tests. Moreover, there is no quantification of biases (smaller etc is too generic). Please quantify the biases

3.3 WRF configuration

 Why have your tests run only for the period 1993-2021 (also in the reference simulation) instead of covering the entire ERA5-ORAS5 period? Please specify

What do not you investigate also the behavior of wind and precipitation on the ocean domain since their importance for E-P and mixing?

4 Reference simulation

 Line 235: I would add also Soto-Navarro et al., (2020) and Reale et al., (2022) that addressed this issue.

Soto-Navarro, J., Jordá, G., Amores, A., Cabos, W., Somot, S., Sevault, F., et al. (2020). Evolution of Mediterranean Sea water properties under climate change scenarios in the Med-CORDEX ensemble. Clim. Dyn. 54, 2135–2165. doi: 10.1007/s00382-019-05105-4

Reale, M., Cossarini, G., Lazzari, P., Lovato, T., Bolzon, G., Masina, S., ... & Salon, S. (2022). Acidification, deoxygenation, and nutrient and biomass declines in a warming Mediterranean Sea.Biogeosciences, 19(17), 4035-4065.

 Being a long term simulation, it would be interested having the analysis of long term timeseries of temperature and salinity with respect to observational datasets to have an idea about the behavior of the thermohaline properties at different depths.

It would be important also to assess the source of the error in the total net heat flux. Is it an underestimation of the net shortwave? Latent or sensible heat fluxes? Please think to include this analysis

Figure 11 Why is the area outside Gibraltar colored in the upper panels and not in the bottom panels?

5 Data assimilation

 I found the data assimilation an interesting new advancement in the coupled model. However, I do not understand the reason for relaxing at the surface when already both atmosphere and ocean assimilate at high frequency data to correct errors in the simulated field. Could you please explain better that?

Moreover, I would ask the Authors to better discuss why with fully data assimilation active (and high frequency of assimilation) the improvements in the biases is relatively small. Low quality of input data? Please infer on that.

5 Mediterranean hurricanes

 How do you track and reconstruct the temporal evolution of the systems analyzed in the manuscript? Please infer on that.

6 Discussion and conclusions

 There is a lack of the comparison with respect to previous modeling systems. Why should a potential user use your modeling tool instead of another one? Do you expect that increasing the horizontal resolution should improve the performances of your model or should make it slower?

This study utilizes a newly developed regional coupled model (atmosphere-ocean-hydrology) to investigate the basic air-sea interactions in the Mediterranean region. Especially, different configurations for each module, such as turbulent parameterization, microphysics and so on, have been tested and the best configuration has been suggested. Further, assimilation experiments were conducted and results related to extreme events were presented.

This manuscript is generally well-written and well-structured. The result of this work is important to the regional climate research and prediction community. I, therefore, recommend Minor revision after a few modifications. The detailed comment is in the attached PDF file.

The paper is a well-written description of a new regional modelling system, MESMAR1, which includes atmosphere, ocean, hydrology and data assimilation components. The latter is a focus point of the paper. The paper topic is very suitable for GMD.  I only have minor comments on the paper.

Minor comments.

In agreement with another reviewer, I would like some more discussion (in the Intro.)  on why the system is different from previous regional models, and its specific advantages. There is a brief discussion at lines 379-383, but I would like it more explicit in the Introduction. For example, to what extent do other systems have data assimilation?

Line 17. It would help to specify the grid spacing for ocean and atmosphere (not just ocean) using the same units.

The paper has a slight bias towards the oceans (with less discussion of atmosphere processes). This is OK, but lines 39-43 ignore the fact that enhanced resolution may improve representation of atmosphere dynamics, not just ocean. Is one of the aims of your paper to identify when full coupling provides better results? You could make this explicit.

Lines 56-57 and 66-67 both introduce medicanes. I suggest to delete the first example.

Line 101. Are the lateral bcs applied as some kind of nudging?

Line 108 "partial bottom steps"

Lines 114-116. Is this the same as the GOTM (https://gotm.net/portfolio/) ?

Line 177. 2 years seems a bit short for this comparison.

Line 183. For Fig. 4, add a map of model minus EN4 surface salinity (probably annual mean).

Line 193 introduces the TKE scheme in the wrong place. It should be described more fully in section 2.2.

Lines 205-210. Fig. 5 shows a strong seasonal dependence of bias, but Fig. 6 includes all seasons. Consider making the equivalent of Fig. 6 for winter and summer.

Lines 270-275.  Does the net air-sea heat flux difference between MESMAR and reanalysis relate to SST difference and turbulent fluxes, or to radiative fluxes?

Line 288. "climatological anomly"-> "anomalies from climatology"?

Line 305. More definition of full-field nudging - I assume this means no length-scale filter, and similar timescales?

Line 320. "SST skill scores"    - SST is nudged to analysis on 15-day timescales, so i don't know how to interpret "SST skill"!

Section 5.2. I assume that the verification data was not assimilated. (Please state this explicitly.)

Fig. 14. As above, the system is nudged to SST, so how to interpret MESMAR minus satellite SST?

Line 357. CTL has Medicanes due to bcs?

Lines 362, 375 quote percentage improvements for ocean assimilation, how do they compare to atmosphere assimilation? I understand that we expect ocean assimilation to have a lesser effect than atmosphere assimilation for atmospheric weather systems, but it might be useful to compare.

Fig. 11, consider adding two more panels of the differences: model minus reanalysis.

Fig. 12 - I think a lot of lines overly each other, e.g. BIAS of humidity, perhaps mention this in the text. i.e. the lack of sensitivity to changing configuration.