Review for Cai et al., 2025 – NAAC (v1.0)

This paper describes a long-term model simulation that investigates saltwater intrusion along almost the entirety of the United States East Coast over the course of a 20-year long simulation with remarkable spatial resolution using SCHISM. Overall, the paper demonstrates an impressive application of ocean modeling tools to multiple coastal ecosystems, allowing for novel regional comparisons that are not typically simultaneously simulated together with this level of detail. The methods described are appropriate for this task, and most of the model validation presented here supports the use of SCHISM for this application. However, many details could be elaborated upon a bit further, particularly in the discussion section which is quite light and could benefit from a reorganization of section titles with more comprehensive references to other literature and the figures presented in this paper. Some conclusions drawn in the discussion section are difficult for the reader to pick out from the presented results and some may be better supported by other results that are not shown (or are not shown in great detail).

Specific Comments

Line 32: small typo, “… has been an effective tool…”

Line 116-122: Typos in “… Ocean Observing System…”, “… Chesapeake Bay Program…”, and “… data from the Neuse…”

Line 170 (Figure 3): In panel b, are the light blue bars meant to show the annual mean discharge for each of these rivers? Could use a little more explanation in the caption or text.

Line 217-219: I’m a little confused by the explanation of the model bias, which mostly just restates the definition of a bias? I’d also add that SCHISM appears to have higher variability on shorter timescales than DRBC observational data, although I’m not sure if this is just due to temporal sampling limitations of the upper saltwater limit?

Line 233-234: I don’t find this degraded model skill in temperature in shallow tributaries that suprising, NARR tends to overestimate shortwave/longwave radiation and air temperatures in coastal water bodies when compared to observations and other reanalysis products like ERA5. I don’t think that this warrants any major changes to the paper, but it’s something that could be considered in the future.

Line 247-248: And the model appears to slightly overestimate current velocities at station ‘cb1001’? I’d be curious to know if this overestimate and underestimate are temporally correlated or if the bias patterns are unique to the different stations.

Line 272-274: This sentence reads a little awkwardly, I believe that you are saying that water resources on the Delmarva peninsula are at risk from SWI?

Line 275-278: Change “as not reached” to “has not reached” and suggest changing second half of sentence to, “… from DRBC may tend to be underestimated by interpolation constraints due to relatively few observational stations.”

Line 282: Suggest rephrasing as “share a drainage source from the same…”

Line 289-290: Some awkward phrasing, also seems to suggest that Sanford et al. (1992) was focused on coastal carbon cycling? A better reference for the second clause of the sentence would be something like Najjar et al. (2022) - https://doi.org/10.1002/2017GB005790

Line 297-300: What figures is this point referencing? I assume a combination of Figures 3 and possibly 8? Same question for the remainder of the paragraph.

Line 305-314: To me it seems like this wetlands application of the model got significantly more description in the methodology section than is warranted by the relatively brief summary of the results presented here in the discussion section. If there is not much more to add, I would suggest potentially moving this paragraph to the results section or describe some additional implications of highly resolved wetlands processes here.

This paper describes a regional model with downscaling capabilities, such that the model resolves a large domain including the mid-Atlantic bight and Gulf of Maine, while also resolving a tidal wetland in detail. This 'seamless' approach is an important advancement in regional modeling, and mirrors recent work by the ICON team ('Seamless Integration of the Coastal Ocean in Global Marine Carbon Cycle Modeling' by Mathis et al., JAMES, 2022). I would recommend the authors acknowledge this related work. While I am very impressed with this effort, I think that there are some issues that need to be addressed and acknowledged (if not fixed) before this work can be published. My concerns are centered around model fidelity at the smallest and largest scales, and the connections between these scales, which I don't think has been adequately demonstrated.

First, at the smallest scales, I am worried that the large timestep reduces the model accuracy. The authors state "Notably, the model's scheme bypasses the constraints of the Courant-Friedrichs-Lewy condition (Zhang et al., 2016), allowing for the use of relatively large time steps even with high-resolution grids that capture detailed geomorphic features." Model stability is not the same as model accuracy, and running at timesteps greater than CFL prescribes can lead to model degradation, even if the model remains stable. The authors need to discuss the accuracy of their numerical schemes at the smallest scales, as these are what typically set the timestep.

At the largest scales, it is not clear to me that the model reproduces large scale shelf features well. A single snapshot of the Gulf Stream surface currents is not sufficient evidence that the key features on the shelf are reproduced. This is important because it is the whole point of the 'seamless' downscaling approach -- that the larger scale (shelf) processes that influence the smaller scale (bay/wetland) processes are all reasonably reproduced.

Finally, given this last point, it would be good to see some evidence that there is indeed a connection between the shelf and the wetlands. I would prefer to see some relationship in terms of tracers (temperature or salinity) rather than tidal flows that at least qualitatively demonstrates how shelf processes might influence the wetland region.

In summary, I'm very impressed by the model application development, but I am not yet convinced of its usefulness. 

This manuscript is about the implementation of a modelling system based on SCHISM covering the Gulf of Maine, the Mid-Atlantic Bight, and much of the South-Atlantic Bight (SAB) along the North American Atlantic Coast. In this study, the authors conducted a 20-year simulation from 2001 to 2020, focusing on tidal elevation, salinity, temperature, and velocity. The system was evaluated by model-observation comparisons over the studied regions. For the first time, they extended a regional continental-scale ocean model to the tidal wetlands to include the compound flooding process. They also conducted a single-year simulation with comprehensive tidal marsh coverage for sensitivity analysis. The model lays a foundation for future applications and is worthy of publication. I suggest a major revision before publication.

The labels in Figure 1 are confusing, with repetitive labels of (b), (c), (d), (e), (g). I understand that the orange boxes are defining the regions covered by (b), (c), (d), (e) subplots. However, the way to present is confusing. I suggest you change the orange labels to the name of the regions, for example, change orange (b) to Gulf of Maine, orange (c) to Long Island Sound.

P125: was used to validate creek and wetland simulations -> were used to validate creek and wetland simulations

P235: The authors stated that “The model effectively represents the Gulf Stream, as indicated by sea surface temperature (not shown)”. I am wondering why they are not showing the SST comparisons? It is important to demonstrate the representation of the Gulf Stream.

For the currents comparison in Figure 10, a metric is needed for a quantitative comparison of both magnitude and direction.

P275: “by the salt front estimation data provided from DRBC (Fig. 10)”. It is referring to the wrong figure, since Figure 10 is not related to salt front.

It is not clear how salt front distance is defined/calculated.

In section 4.1.2, the authors stated that “…modeled tidal flow fluxes at the gate or major inlets of the three significant waterbodies at mid-Atlantic – Delaware Bay, Chesapeake Bay, and the APES. While the net fluxes are always associated with the riverine discharge as denoted in Fig. 3, the flood and ebb tidal fluxes have distinct patterns among these three systems” However, this discussion lacks the detailed analysis one would expect from such a statement. The section provides no calculations, additional figures, or quantitative comparisons of these "distinct patterns." Without supporting data or deeper analysis, this discussion feels superficial and insufficient to warrant its own section.