Review Ms GMD-2023-183 ‘’CAR36, a regional high-resolution ocean forecasting system for improving drift and beaching of Sargassum in the Caribbean Archipelago’ by Sylvain Cailleau et al..

The Ms present a new solution of IBI system based on NEMO ocean model at 1/36º resolution in the Caribbean Archipelago in order to forecast the drifting of Sargassum in the area. The system is forced at the open boundaries by daily solutions of the GLO12 and forced hourly by ECMWF.   

The Ms. present an interesting application of a high resolution model with undoubtedly interest for statistical studies of submesoscale processes in the Caribbean area. The ability of the system to reproduce the local dynamics is tested qualitatively and quantitatively against SSH, SST, sargassum detection, drifting buoys and eddy tracking. The different metrics presented and the comparison with data are in line with the usual techniques used to test model agreement although some effort in line with lagrangian diagnosis would be desirable. 

The paper is well written and provides a good introduction on the problem of drifting objects specifically in the area of interest. In my opinion testing the model with an eddy in the area, although interesting, is far from the objective of the system which as the authors state is the drifting and ‘beaching’ of Sargassum. Besides, the effects of waves, very important in the east side of the domain, have not been considered. 

I think the Ms can be published after some clarifications: 

Although being a research area that is very relevant, the present manuscript has two drawbacks ;

•  First, the effect of Stokes drift in the drifting of objects is never treated nor discussed. How does this component of the velocity influence the lagrangian dynamics in the area?. 
•  The second question is related to the objective of the paper.  Comparing the performance of te model with the detection of Sargassum in the Eulerian framework is fine but, why not using a lagrangian metric (for instance a better way to test the difference between both solutions would be the use of the lagrangian divergence or FSLE). This second issue would give the ms. a more robust significance.  

Specific comments: 

• Page 2 Ln 68. Define Caribbean Archipelago. 
• Section 2 (last paragraph). Why are you concerned specifically about eddies?. 
• Section 3.2. Ln 157 . Although this is true for the CAribbean and the west part of the domain this is not true for the east side. 
• Section 5.2. Ln 310. Is it possible to perform it?. 
• Section 5.2. Ln 319. It is very difficult to see from the L3 what the authors state (figure 8)
• Ln 353. <Can the authors discuss the disagreement respect the difference in the overestimation. Do you expect to improve the results using more eddies? (statistical relevance).
• Figure 4. I suggest changing the scale for the bias and RMSE to remark on the differences. 
• Figure 5 (the same comment). 

I am sure that the authors will provide answers to these comments given their expertise in the subject.

My only suggestion is that the authors consider a full check of the English. ChatGPT could probably do it.