This manuscript presents a 3D global ocean set-up within the SCHISM model and demonstrates the ability of the model to hindcast some aspects of the sea surface level, salinity and temperature distributions. As a main feature, the authors highlight the seamless (i.e. within a single grid) integration of global oceans and high resolution coastal features and estuaries in a 3D baroclinic model, which is new.

The modelling framework presented by the authors is impressive and results on SSH, salinity and temperature on the global level seem state-of-the-art, going by the numbers provided by the authors themselves. To me however, the manuscript reads like a progress report for the community already using SCHISM, not like a GMD paper. For it to become a GMD paper, I expect the authors to identify the unique ability of their model and then show it actually delivers on this ability. The authors identify the cross-scale feature as unique, but do not show this aspect actually delivers anything and why one would want this feature. Also, while results on the estuary scale are presented, it is unclear how accurate they are and hence unclear whether the model is usable for the estuary scale. I detail my comments below.

Overall I recommend major revisions.  

Main comments

1. Context and aim

The context now set in the introduction is that coastal and estuarine areas and baroclinic (3D) processes are of added value for evaluating the global ocean energy budget properly and that this SCHISM set up provides a new and unique environments for including these aspects. However, nowhere in the manuscript the added value of shallow areas or baroclinic processes to the global SSH or energy budget are actually shown, let alone that it is shown that this model is therefore superior or adds new understanding. In several places (ln 201-202, 303-306) it is claimed that the baroclinic processes simplify calibration and improve performance, but no evidence of this is given. The authors only compare their model on global M2 tides, SST and SSS but the added value of the shallows and baroclinicity to these results is not shown. 

This is a GMD paper, so no new scientific results are needed here, but I do expect the authors to say why I would want the unique ability of their model and then show that they can actually deliver that. Hence, if the context is, for example, the global energy budget, please compute the energy budget of the shallow areas and of the baroclinic component and compare to literature. The authors may respond to this comment by changing the context in the introduction and/or adding new results. Essential is that they demonstrate the model results can deliver on the sketched context and the specific novelty of this model.

2. Focus on certain processes and mentioning of other global ocean models

The manuscript to me hanging between two conflicting ideas. On the one hand, the authors state they want to focus on M2 tides and the addition of shallows, while they accept less focus on some other processes (e.g. eddies, sea ice). On the other hand, I did get the feeling from ln 37-43 and the manuscript overall that the authors are aiming at improving global forecasts of tides, SST and SSS. This is followed by a great deal of apologies for barely resolving eddies (e.g. Ln 173, 235) and sea ice (154, 257, 272, 278, sect 4.1) and no focus on any improvement compared to other models. These apologies are not so nice to read and feel disappointing. I’d advice the authors to clearly state their assumptions and then briefly mention several global models (with names and references) and some of their unique abilities in terms of resolution, dimension, eddies, sea ice etc. Then they can follow on the first idea of focussing on shallows only and accepting their assumptions. The missing eddies and sea ice then only need to be mentioned briefly to explain some of the discrepancies in the results, but no more is needed. Results on the specific  added value of shallows and baroclinic processes is needed. Such an overview of models and their abilities helps readers to set the results in context and select the model that they need for their purpose. 

3. Estuaries

Section 4.2 about estuaries should play an essential role in this manuscript, as it is the main section that focuses on the cross-scale feature that is the main novelty of this model. However, I don’t know what to learn from this section. It is shown that some degree of stratification and region-of-fresh-water-influence (ROFI) is produced by the model. However, since there is no systematic comparison of salinity with observations or dedicated models, I have no idea if results are accurate and for what purpose these estuary-scale results could be used. As the authors state themselves, more local calibration will be needed in estuaries and this is not efficient to do in a global model. So then why would an estuarine scientist want this? I don’t think the authors can state with any confidence that this global model can be reliably and practically (i.e. time-wise) be used to say anything about estuaries. Hence, I think this section needs to be removed, possibly to consider for follow-up research. Instead, the authors may choose to focus on water levels only. In this case, results and observations further up-estuary would be appreciated and I would like to know why I want this over a local model.

In their own introduction, the authors never claim the model is suitable on the estuary level. They only claim that the addition of estuaries has important back effects on the global SSH and energy budget. In line with my comments above, I therefore recommend the authors to focus on this aspect instead.

4. Context of SCHISM

SCHISM is not new and the authors add several references to the model. It remained unclear to me if the model presented here adds any new aspects to the software or if this is a new global grid set-up within the existing software. I recommend the authors to add a short subsection stating what is actually new about this model and the set-up and list a few cases where the model was used previously, also to demonstrate the novelty of this manuscript.

5. Reproducability

Please add the version number/date of SCHISM used. Is the grid set-up and forcing data all available online? 

Specific comments

Ln 97: salt intrusion processes? (1) this is new and was not mentioned in the introduction at all. Given the goals, it seems unfit to discuss this. 2) even models focussed on estuaries alone struggle to properly resolve salt intrusion dynamically, so even if you have the right resolution does not mean you can actually resolve it well. 

Ln 107: where are the small numbers of layers used. Baroclinic effects in shallow areas can be important for friction estimates and hence the local energy budget. Also, it seems you don’t use such few layers in all shallow areas. A bit more elaboration on your choices would be useful here.

Ln 111: no friction at large depth is probably a reasonable approximation for many purposes, but is this not a huge error compared to the energy dissipation in an estuary? When reading this I’m still on the line that you want to improve global energy budgets. In that context it seems weird to model details of estuaries but be so crude about friction in the global ocean. A much improved context in the introduction should help to resolve this. 

Internal tides (e.g. Ln 116): is its dissipation in the model not because of reduction of number of layers towards the shallow? Did you check numerical accuracy of any of the baroclinic action in the global model?

Ln 201-202 ‘the more complete physics ...’ This conclusion is not supported at all by results. Please remove.

Again in ln 303-306. The more complete physics does not generally simplify calibration. I believe you have a specific benchmark model in mind that includes parameters for the baroclinic processes that you don’t need. Also, you choose not to add any additional calibration parameters related to the baroclinic part (you could choose to add some parameters to the turbulence closure for internal wave breaking and such). Please make this explicit. What benchmark model(s) are you comparing to? What specific parameters do you not have (or are simpler compared to that/those models)?

Ln 270 and fig 8: indeed, river plumes seem to be hugely overestimated. Based on this, I’d say that your model cannot be used to look closer at any coastal areas or estuaries specifically (hence, section 4.2 should be removed). Also, I’m not sure if your addition of shallow areas is actually having the correct back-effects on the global SSH or energy budget. Substantiation of this in numbers and showing these effects of coastal areas on the global values are appropriate is really needed.  

Ln 344-347: you claim here that your model better captures the NTR compared to 2D models. However, I have not seen the NTR in fig 13 and it is not compared to 2D models. This claim is unsupported. I think it should be removed.

Fig 14-16 look nice, but do not substantiate anything if there is no comparison to observations and no indication in the text on how one should want to use this model for estuary-scales.

Ln 365: adequate: what is that? I don’t see comparison to observations or other dedicated models

Ln 367-370: this is speculation without substantiation. I don’t think this fits here. 

Eq 2 seems to have a printing error

Ln 148: What is SAL?

Eq 5: integral should be over Omega not A. This is a form of the average, not integrated (ln 192)

Ln 332: edge -> wedge