The paper “Great Lakes Waves Forecast System on High-Resolution Unstructured Meshes” by Abdolali et al presents the Great Lakes Wave Forecast System, from the history of the first models in the area, to the present operational system and the future developments. The paper is well structured and written. However, sections are quite concise as the topics treated are often just mentioned. Also, the paper deals with a complex forecasting system in a complex geographical environment, which would require more detailed descriptions of the different elements involved (the forcings, the details of the modelling workflow, just to mention a couple) and of the figure and results presented. My recommendation is to revise the paper expanding the presentation, in particular of the topics of Section 2.

Specific comments:

• line 77: the forecasting schedule is not clearly enough explained in my opinion, please make it more clear to the reader
• line 94: it would be useful if authors repeat here the resolution of the Great Lakes grid
• Figure 2: acronyms should be added in caption; also, details of the forcings are quite concise and might be better explained
• Section 2.5: I do not see the reason for having a section without any result. This can be added in the conclusions or somewhere else
• lines 205-215: the point the authors are raising here is quite important and they have already commented it well. However, I would stress a bit more the benefits of the high coastal resolution even in absence of a sea truth to prove them, by showing some examples. 
• in the introduction a recent GLWU implementation which incorporates the implicit scheme and the current/water level forcing is mentioned. Then, the version 2.0 presently operational is presented, based on the explicit solver and no current/level forcing. And those two points are finally listed as the future developments (the first is well elaborated, the second only mentioned). In my opinion this generates confusion, so please, fix it.

Ms. Ref. No.: gmd-2023-167

Title: Great Lakes Waves Forecast System on High-Resolution Unstructured Meshes.

Ali Abdolali, Saeideh Banihashemi, Jose Henrique Alves, Aron Roland, Tyler J. Hesser, Mary Anderson Bryant, and Jane McKee Smith

Overview:

In this paper “Great Lakes Waves Forecast System on High-Resolution Unstructured Meshes” by Abdolali and coauthors, the Great Lakes Wave Forecast system is described. Its evolution from first-generation parametric model to a third generation full spectral model with an unstructured mesh (GLWUv2.0). The operational implementation of the GLWUv2.0 is described: its meshes, forcing fields and workflow. A validation was done comparing the operational model output against in-situ measurements from 25 buoys. The numerical system was also run using some key features that WAVEWATCH III, which is the core numerical model, already has and will be used in future operational implementation of the wave forecast system. The output for the later runs was compare to measurements from 6 buoys.

General recommendation

The paper has a value for the wave modeling community, it is well structured. However there is a lack of connection between objectives, the proposed experiments, the results and their conclusions. The paper needs a major revision. I recommend to revise the paper and resubmit it.

Major comments

1. The abstract should include briefly the key conclusions of the study.
2. For numerical wave modelling on high geographical-resolution and shallow waters, high-quality bathymetry is essential, however there is a lack of information about the source of the bathymetry and its quality. This must be addressed since in some numerical experiments the wave system has a resolution up to 5 meters.
3. Section 2.5 is totally disconnected from the rest of the article. “Dangerous Seas”, this topic does not appear in any other section; from introduction to conclusions. There are not numerical experiments presented in this paper related to this topic. It could be mentioned in a couple of lines in “Future Implementation” section, but it doesn’t deserve a full subsection.
4. The Conclusions section offer, again, a description of the system, description of the unstructured meshes, describe the need for high resolution meshes and coupling with other Earth systems, etc. It does not offer conclusion of all the implementation, validation and statistics done.
5. The values of statistical parameters, when comparing G0, G1 and G2, do not show a “noteworthy improvement” as it is stated (line #209), as a matter of fact the authors declares that “due to the lack of coastal observations (line #220) … the three simulations show nearly equal performance (line #221)”.
6. The implicit scheme makes the forecast system to finish faster but the price we have to pay is the numerical diffusion. Something should be mentioned about this topic and I guess a time step must be provided when using the DD scheme. What were the time steps used?
7. The authors mentioned that for the winter simulations there were not buoys measuring waves and “…only qualitative checks were performed”. Those checks are not shown in the paper. How good or bad the forecast system was qualitatively?
8. Lines 207-208. Figures AA1 and AB1 are mentioned but the results were not described nor discussed and they were not used to conclude anything. Any description or conclusion is left to the reader. In this case those figures do not add any value to the paper.

Minor comments

1. Line 3. Instead of “is successfully tackle by” could be “is successfully tackle in part”. There is a need for implementation of more accurate physics, as it is mentioned in lines 37-8.
2. Line 8. “Our results describe the development…” The results section should not be used to describe the development of the wave forecast system.
3. Lines 13-14. In the US population living in the Great Lakes region the entire states population is taking into account, however for Canada the population is taking only as a part of Ontario, please review the literature on how many people lives in Ontario, and set the percentage related to Canada, as it was for the US.
4. Lines 31-32. “Two years later in 2006”, to years later compared to what? There is not a reference to the year 2004.
5. Line 50. “allowing very large meshes”, CD allows very large meshes as well, but what is the difference? “allowing very large meshes to run in short time”?
6. Line 66. “Section 3” should be Section 2”.
7. Line 81. “The WW3 model” should be “The GLWUv2.0”, as WW3 can have a different values for the parameters, but the values provided there are specifically for the Forecast System.
8. Line 83. Need space between et al. and (2010).
9. Line 101. “a stationary ice concentration at the initialization time step”, then, what is provided after the initial time step? A non-stationary ice concentration? The ice field is keep constant in time or there is a forecast system for the ice concentration?
10. Lines 109-110. A resolution for the HRRR winds is provided but no for GFS winds.
11. Line 117. No need to repeat the list of the Great Lakes.
12. Line 123. “In case the current cycle is not available” should be “In case the forcing for the current cycle is not available”.
13. Line 124. “If the ice field is not provided, the previous forecast cycle ice field is used” So, is there a forecast system that provides forecasted ice fields? Or are those analyzed fields which are provided by NIC and they are kept constant in time for the whole forecast window? This is not clear.
14. Line 139. What is the running time for the long (or short) cycle for Lake Champlain?
15. Line 149. “25 locations, shown” instead of “25 locations as shown”.
16. Line 151. “Which was one of the criteria”, where there other criteria used? Which ones?
17. Figure 10. In the caption, instead of “normalized by frequency and directional resolution” should be “normalized by the number of frequencies and directions” as indicated in the x-axis.