Yoshikai et al present a novel implementation of a mangrove root model in the ROMS data package that has the potential to be very useful for the wider scientific community. They show that the model predicts flow velocities and turbulent kinetic energy in a more refined way in the vertical and that this matches qualitatively better with the presented data from a flume experiment and the field. Although I think that the model is very useful and can advance the scientific field, I have major concerns with the application of the model to the sediment transport predictions that links directly to large parts of the discussion. In addition, I missed more detail on the data that has been used. Finally, to help lift this contribution from a presentation of the model, I suggest to apply the model to a case study with available sediment transport rates and potentially morphodynamic change to show that those predictions are improved using the presented model. I outline my concerns below:

1) To understand where the data comes from I suggest to add maps of both the study area and the model grid with flow velocities as well as the set-up of the flume experiments/model iof the flume experiments. It is unclear where the measurements have been taken (unvegetated vs. within the root system, close to tidal channels etc) and how exactly the model looks to allow to understand the results and to reproduce the study. In addition the data used for validation should be presented in the supplementary.

2) The sediment transport computations seem arbitrary from the choice of parameters. Although it is interesting to compare sedimentation rates for the different model parameterizations, the rates need to be validated by data to be able to say that they are realistic. Especially the choice of just one setting seems very limited, depending on the types of grain sizes and parameterization the rates can be very different and there is now no indication that the model can predict realistic rates or that the new model predicts these rates "better". I suggest to add a validation here and test a wide range of sediment parameters to be able to identify trends.

3) Based on the analyses presented, some parts of the discussion overstate the outcome of the study, for example :

line 434: "The good performance of the model in both the model- and real-Rhizophora mangrove forests having a range of vegetation complexity (Fig. 3) suggests the model’s general applicability to Rhizophora mangrove forests worldwide. "- I don't think you can state that the model improves predictions for any other systems than the one studied here. To be able to upscale to other systems, more analyeses are needed.

line 458: "For the practical use of the model, we proposed a model framework (Fig. 1) leveraging an empirical model for the Rhizophora root system (Rhroot model) with parameterization of subgrid-scale tree variations (Fig. 2), which we implemented in COAWST."- as far as I understand you implemented the already existing theoretical model of root area, so more careful phrasing here since the novel part here is the implementation.

line 472: "This study thus offers the first framework of numerical modeling which can be readily applied to Rhizophora mangrove forests in

the field." - again, I think you with this work you provide a good implementation of the model

4) The discussion on the sediment transport would need to be removed or revised in case new analyses are added

Minor comments:

1) the paragraph on carbon in the introduction seems a bit far from what is presented in the study

2) I am not sure you need the reference runs in your graphs since you are comparing the new root structures with static vegetation. Why not add the model in Xie et al (2020) to compare with another "more realistic" representation of roots?

line 91: the reference seems to be the data of the paper. I would like to know what is different between the implemented model and the model you are referring to here

line 181: are you defining tree sizes as a distribution?

table 2: please make more clear that these are the measurements by linking them to the map

line 259: please present the sensitivity runs in the supplementary

fig 4: maybe remind the reader in the caption what is HRmax

Fig 5: what absolute water levels and time-steps throughout the tidal cycle are presented here? I am not sure what the difference is between actual and modeled aroot. Is one the data and one the predictions of the implemented model? Add R^2 values to quantify the error

I hope that the authors can extent their analyses and revise the manuscript as I believe this will be a very useful contribution.

General Comments

This manuscript presents a new approach to modeling the flow of water within Rhizophora mangroves. The key improvements to the COAWST vegetation package are: (1) allowing the vertical varying projected area density (frontal area per unit plan area), (2) using the root and stem length-scales in the turbulence dissipation terms, (3) implementing the Rhizophora module which can calculate projected area density from easily obtainable field measurements. These improvements allow the field to move beyond the conventional cylinder assumption, and are generally applicable to all hydrodynamically rough environments which aren’t well described by cylinders.

I liked the approach and theme of this paper, and think with some revisions it would be a nice contribution. I also really appreciated the detail of the supplemental information.

Specific Comments

1. The no vegetation case shown in Fig. 5 and Fig S1.
   1. While including this test case in the manuscript is interesting to see how the hydrodynamics are changed if the mangroves were removed from the ecosystem, it seems to change direction from the rest of the paper. I believe the message of this paper is comparing how this new approach of accounting for the roughness of mangroves is different from past approaches (cylinder arrays or enhanced z0 values). I think that if the z0 value for the no-vegetation case is increased, maybe similarly to the Zhang 2012 mentioned on line 65, this would fit with the theme of figures 5 and 6 which contrast the new approach to past approaches.
2. Does the sparse cylinder model (line 289) have an equivalent frontal area to the field data? Similarly to how the cylinder models used in EXP1 and EXP2 have equivalent frontal area to the Rh model. If so, then this section would nicely flow from the lab section where frontal area was conserved. If not, then there is a jump in the methods being used to create the cylinder arrays in the lab vs in the field section. For best transition between the sections I think the method for generating the cylinder arrays should be consistent between the lab and field parts of the paper.
3. I would love to see a figure that shows the difference accounting for 2 length-scales in the turbulence routines makes. This is mentioned on lines 444-445 and lines 424-426. I haven’t seen a figure that does this yet, and the changes to the code are already made so I think this could be a nice addition to the paper.
4. I would appreciate if the cylinder metrics (diameter, density, height) mentioned in EXP1,EXP2, the sparse cylinder array and the dense cylinder array could be compiled into a table and attached as supplemental information.
   1. I believe it is important to state the height of these cylinder arrays. Based on the velocity profiles shown in Figures 4 and 5, I believe all the cylinder arrays span the entire water column. However there are other papers which use cylinder arrays which span a fraction of the water column, so I think it is important to state.
   2. The widths are also important to state because of they are used in the turbulence dissipation term. The diameters for EXP1 and EXP2 are already stated, but I couldn’t find diameters for the sparce and dense cylinder arrays.
5. Lines 200-211, I think that this section and table 1 can be removed from the paper or moved to supplemental information. This section might be useful as a guide for someone using your code, but I don’t think it adds much value towards understanding the content of the manuscript.
6. Generally it seems the core modifications to the code are in the drag term and the turbulence routines, and these change the flow field which then might change the sedimentation rates. I think the sediment parts of this paper are a case study of how these model changes can affect a variable of interest (like sedimentation rates), but I don’t think the focus of the paper should be on the changes in sedimentation rates.

Technical corrections/ Typing errors

1. Line 308 “run” should be “ran”
2. Line 214, could you please consider stating the vertical resolution and/or the number of sigma levels used? This is mentioned in line 241, but I think it would be nice to have all of the domain characteristics mentioned in the same place.