Biogeochemical modeling is increasingly important for both understanding biogeochemical processes and predicting rapidly changing ocean conditions due to ongoing climate change. The authors have developed a fully coupled NWA model that integrates ocean, ice, and biogeochemical components, meeting the needs of many researchers working in this region. Sensitivity analyses enhance understanding of the roles of tides and sea ice in model biases.

This manuscript is well written. The authors use RMSE, RMSD, and other statistical measures to evaluate the performance of the model runs, which is appropriate. While it is known that the Northwest Atlantic Ocean has been undergoing significant changes, temporal changes in the ocean can serve as important indicators of the impact of climate change. Although the short simulation period of 4 years may not be useful for climate research, I suggest the authors select specific locations to demonstrate whether the models can properly simulate the evolution of key variables (monthly mean or annual mean). Numerous observed sea surface temperature data, bottom temperature data (summertime), current meter data, and other types of data are available for this region, which could be used to further investigate model performance. This would enhance the confidence of readers and future users in the model. Additionally, winter convection events in the Labrador Sea are believed to be a significant driver of circulation in the North Atlantic Ocean, impacting water mass properties in this region. It is worth reporting whether the ROMS model can well simulate these winter convection events during the study period.

Below is a list of specific comments.

• Line 34 “up to 7”, could “up to 7 psu” be better”?
• Lines 153-154. Why are the eddy viscosity and diffusivity set to zero? Given that you employ high-order accuracy schemes (3rd and 4th order), it’s unclear why these values would be zero. Could you provide a rationale for this choice?
• Lines 172-174. Are the inter-annually variable GLORYS12 data used at open boundaries or the daily climatological ones? Please specify.
• Lines 182-183. I am curious why it is through the bottom. River discharge is through top layers.
• Lines 298-299. Glorys12 uses data assimilation (4d var), and in theory, this can eliminate those biases from the non-inclusion of tides.
• Lines 310-311. Same as above. You may want to find some literature to support this statement.
• Lines 369-378. This seems an indication of numerical scheme issue or the horizontal mixing issue (zero is used).
• Lines 409-425. Data from Drinkwater (1988) were for the year of 1982, and the data your model and GLORYS12 are for the recent years. You need to mention the probable existence of decadal or even longer variability for the current in this area.
• Figure 13. I cannot see any dots there. Could the quality of the figure need improving?

The manuscript presents results from implementing and validating a coupled ocean-sea ice-biochemical model of the North Atlantic. The authors developed an advanced coupling between the ROMS and CICE ice models. Model results are compared with surface and in situ ocean observations, and the validation encompasses various setups, including model forecasts, simulations constrained by observed temperature and salinity, and scenarios without tides or sea ice.

The manuscript is well-written, and the model validation results are robust. The authors’ conclusions are well-supported by various statistical measures of model error. The model validation demonstrates the model's significant potential for future diagnostic and prognostic climate change simulations in the Western North Atlantic.

However, I recommend adding further discussion on several aspects of the model setup and simulations:

    • Lines 55-56: Is deep convection an additional or dominant component of CO₂ removal in the Labrador Sea, as Tian et al. (2004) discussed? Clarifying this point would strengthen the manuscript.

    • Line 104: The resolution of "O(km)" is mentioned. Currently, eddy-resolving circulation models of the North Atlantic typically have resolutions between 1 and 10 km. A more specific definition of the model resolution would be helpful for readers.

    • Line 119: Is the freshwater flux separated into solid and liquid components? If not, could the authors provide a rationale for this choice?

    • Lines 153-154: The model diffusion/viscosity is zero. What boundary condition is used for the tangential velocity component, and does this formulation accurately represent the friction in the lateral boundary Ekman layer?

    • Lines 348-350: The salinity model error may be related to boundary conditions in this region of the model domain. Do the GLORYS simulations provide adequate boundary conditions? The higher horizontal resolution of your model suggests that the GLORYS model might underestimate horizontal advective transport through the open boundary of the St. Lawrence Estuary.

    • Lines 550-556: The impact of tides on temperature and salinity in the Bay of Fundy appears minimal (Fig. 19). Could the authors discuss why the tidal effect is relatively small in this area? I would expect them to be more significant.

Thank you for considering these suggestions to enhance the manuscript.