This work seeks to identify the role of  local event-scale variability – namely upwelling – in determining the regional air-sea carbon dioxide fluxes over the Scotian Shelf through the integration of several different data sets as well as the use of a regional numerical model. The paper features wonderful contextualization of previous flux estimates with observational limitations and integration of multiple kinds of data for this regional problem. The problem itself is quite timely as recent work has identified that the coastal ocean rates of change in carbon dioxide may always reflect the global changes. The manuscript requires additional details in the methods section – most notably about the regressions used to drive the initial and boundary conditions and river values, some issues with time surrounding the observations used and the simulation years, as well as the methods pertaining to evaluation of the model itself.  In additional, more attention needs to be paid to the role of the Revelle Factor in driving these interregional differences between the upwelling on this shelf and the CCS. Finally – an most importantly – the authors need to clarify how the upwelling event contributes to the shelf wide estimates more clearly. The paper would be publishable in Biogeosciences if these issues can be addressed by the author team. More specific comments follow.

Major Comments:

The main message appears to be that local processes are important for carbon content of the temperate Scotian shelf region. In the context of that message, the authors need to show how the localized upwelling event contributed to the overall regional flux somehow. One way might be to show this flux as a map. While there is quite a bit of information on the in situ observed location’s variability, there is very little about how that compares to the region as a whole – is it representative?.  For instance, where in Figure 1 does this upwelling occur (at the buoy and along the black line/transect?) – and how does the simulated flux at the surface of the entire region compare to this localized event? How fine of a resolution do we need to observe to get the shelf-based flux estimate direction right? Also, how does this flux compare with other regional/broader scale fluxes reported for the North Atlantic?

Secondly, it is critical to clarify time in this work. 2005 was the year when the warming started intensely on the east coast of North America. The model runs happen before that, but the comparisons are to data after that…. How does that impact the results? What about the time variability of carbon dioxide in the atmosphere over these various intervals?

The comparison to the California Current or other traditionally upwelling situations is not entirely accurate as the vertical gradient in DIC (presented in the figure here) is nearly half what it is in the CCS (Feely et al. 2004). The phytoplankton growth at the surface is quite efficient unless the winds blow too strongly and the phytoplankton can no longer grown in place. This aspect of the upwelling system is neglected in the text.  The signature of the phytoplankton drawdown can be seen very far offshore as it takes nearly a year for CO2 to equilibrate at the surface. In addition, the two systems likely experience very different temperature, salinity, and alkalinity parameter spaces – all of which are important to consider for the response of the carbon system.

The Revelle Factor influence on the differences between what is observed on the Scotian Shelf and in the CCS should be included – for an example described in more detail see here: https://www.sciencedirect.com/science/article/pii/S0278434317303643#f0005

Because the Revelle Factor is important to consider within the context of this issue, it would be important to evaluate DIC and TA with in situ observations locally, here within this manuscript. Please add DIC and TA evaluation of the model fields. Do observations of these fields exist for the simulated period?

The methods requires quite a bit more detail. Specifically, what is the model skillful in (Lines 112) from other studies? Was it evaluated mostly at the surface? Over annual timescales? Or events like in this work? The K1 and K2 constants chosen are not meant for regions that experience a lot of freshwater influence. Can you justify their choice in this region by discussing the salinity ranges that this region observes? What atmospheric carbon dioxide concentration was used?

Most importantly in the methods – the boundary condition DIC and TA relationships and river concentrations require additional documentation. In the case of the boundary conditions, they appear to rely solely on data from the winter months from an unspecified location. Can you add these relationships to supplemental? And describe the data that they rely on? Are they from a similar time period that was simulated? Were adjustments made for time in the DIC field if they were observed more than 5 years earlier/later than the simulations? There are existing hydrographic relationships in the region and globally that could be used instead (McGarry et al. 2021; Xu et al. 2020; CANYON; LIAR) – why generate a new one?

Finally, the point that the upwelling event signal leads to reduced outgassing compared to the rest of the shelf (Line 280-281) is not clearly shown and is related to the main point of the work. The reader is still considering (because none of the other fields were shown) that maybe the phytoplankton growth rate in relationship to the winds -documented in Evans et al. (2015) could also be contributing to this. What does the subsurface pool of pco2 look like prior to these events? Is that getting efficiently drawn down or is the biological response week and so the physical transport is the main control over the surface carbon concentration? See more discussion on the role of event based air-sea carbon fluxes in annual variability for a region here: https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1029/2010JC006625

Minor Comments:

Line 52-52: Please add the Feely et al. 2008 citation here (https://science.sciencemag.org/content/320/5882/1490).

The model gas transfer function chosen is Ho et al. (2006), which is different than the earlier Fennel model iterations. How does this choice (between all of the existing gas transfer functions available) influence your results?

Lines 213-214: Can you add statistics to support “good agreement” here?

Line 292: If you averaged your two regions together  - would your results be more in line with theirs?

Line 314: “ thermodynamic signal in pCO2 outweighs the influence of biological activity “ This is not clearly shown in this work.

Figure 2 - Add statistics (RMSE etc) directly to these plots. Is the smoothing of the model part of the issue? what about the time/spatial mismatch? Is the socat data being interpolated to the location of the mooring? was the model? how was that extracted? These details need to be added to the methods as well – evaluation methods.

Figure 3 – The summer gradient generated by the upwelling (observed) does not appear to be captured by model. Can you address this with respect to the localized mechanism that is the focus of this work? Please add some discussion of this to the text. Is the time period the same between simulated and observed?

Figure 4 – the longitudinal gradient in the observations does not appear to be well captured by the model. Is there additional evidence that the model simulates the upwelling in this area well?

Figure 6 - Highlight the “nearshore” region you mention in the text on this figure. The DIC gradient is not as severe as in the CCS. Consider putting it in this space:

https://www.sciencedirect.com/science/article/pii/S0278434317303643#f0005

Figure 7 – Please add other parameter time series to this plot including temperature, salinity and most important winds (both modeled and observed).

Figure 8- More detail needs to be added to methods about how these comparisons were made.

Figure 9 – Please add vandemark discussion to the text. What is the far right ”section”?

Finally, the title would be more informative if it were about the science question the paper is trying to address.

McGarry, K., Siedlecki, S. A., Salisbury, J., & Alin, S. R. (2021). Multiple linear regression models for reconstructing and exploring processes controlling the carbonate system of the northeast US from basic hydrographic data. Journal of Geophysical Research: Oceans, 126, e2020JC016480. https://doi.org/10.1029/2020JC016480

Xu, Y.‐Y., Cai, W.‐J., Wanninkhof, R., Salisbury, J., Reimer, J., & Chen, B. (2020). Long‐Term Changes of Carbonate Chemistry Variables Along the North American East Coast. Journal of Geophysical Research: Oceans, 125, e2019JC015982. https://doi.org/10.1029/2019JC015982

The authors use output from a regional oceanic biogeochemical model and mooring/ship-board observations to study the seasonal cycle of surface pCO2 and sea-air CO2 fluxes. The general findings are that the Scotian Shelf acts as a net annual source of CO2 to the atmosphere and that biological activity and temperature are the main drivers of the pCO2 variability. The authors also try to show that coastal upwelling is responsible for low near-shore surface pCO2 in summer. Overall, I find the manuscript well organized. However, I have several concerns (see below) that need to be addressed.

Line 14: Might be good to mention here by how much pCO2 changes due to this steep increase in temperature.

Lines 33-36: Since you specify the type of measurements that suggest that the Scotian Shelf is a net CO2 source, it would be interesting to know what type of data suggests that it is a net sink.

Line 63: I would be careful calling any model "accurate"! If the model has been evaluated properly (if the region has an adequate amount of observations), then I bet these studies identified some deficiencies. I would suggest to briefly summarizing the previous model evaluation here and state unknowns due to lack of data, if applicable.

Line 105: What are the possible implications of using a river climatology to foroce the model? Is anything known about interannual or longterm changes to the riverine input?

Line 102: Again, a brief summary of deficiencies and skills of the model would be good.

Line123: Why is it drifting and how does the nudging impact the actual model skill. I was surprised that so much nudging was done for a relatively small model domain. Are the nudged areas not used in the analysis? And if these areas are used, how do you deal with them? Would be helpful to show the nudged areas in Figure 1.

Line 131: Model spin-up of a biogeochemical model usually takes 6-10 years. Can you show that 1 year is enough and the model won’t drift anymore? For example run the model for 10 years perpetually, using the 2000 conditions. Does DIC remain relatively stable, without drifting?

Line 147: Need to label the location of the Halifax and Deep Panuke gas platform.

Line 163: “from top to bottom…” belongs into caption and not into main text. Also, describe method you used to temperature normalize pCO2 in caption.

Line 164: To me it is confusing to talk about days and months. I would just stick to months, since days are less obvious – The reader would have to first convert to the month before understanding what time of the year you are referring to. I don’t see how pCo2 is relatively constant between day 0 to 75. Are you referring to the temp normalized pCO2? But even temp normalized pCO2 is increasing during this time. Might be better to give a range here?

Line178: add “buoy” to “… at the low end of the buoy observations

Line 182: I don’t think the word “consistent” is appropriate here? The model seems to underestimate the DIC drawdown due to primary production compared to both types of observations (temp norm. pCO2). 

Line 189: verb is missing.

Line 209: Figure 4 shows how the model struggles to simulate the spatial variability, which should be pointed out.

Line 212: add east or west to longitude description

Line 216: I don’t think these events are all that obvious in the observations. There were only a total 3 inner shelf observations during this time period, two of which  are actually higher than an outer shelf observation point (also the only one during this period). I agree, that this is obvious in the model, but would be more careful with this statement for the observations. I just don’t think that the observations can be interpreted that way… Im also not conviced by the proposed mechanism that leads to low pCO2, despite high DIC. What does the salinity profile look like?  I think this section needs something like a Taylor decomposition to show that what is responsible for the low pCO2 (see details in 

Line 264: “Accurate” means: “correct in all details; exact”- as mentioned earlier, I yet have to see a model that can be described as “accurate”. I would tone it down… especially because you start the sentence with “This limitation aside…”

Line 270: would be nice to calculate how much the temperature change affects pCO2 and how much DIC increases affect pCO2….

Figures – I really like the color choices of the figures!

Figure 1: It would be nice to give the reader a better understanding of where the Scotian Shelf is located. Maybe a zoomed-out map as an insert? Label all location names you are mentioning in the paper e.g. Halifax Harbor. What are bin 1 and bin 2? Please describe in caption. Also, LAt and Lon labels are missing, including whether it is north or south, and east or west. This should be adjusted for all figures throughout.

Figure 2: Correct “Glider observations”

Identify grey band in legend for consistency.

What are the two different x-axis?

Figure 4: What are these inserts? Zoom in? Does not seem to show what you see in the smaller box below. This figure is kind of confusing. What are we actually looking at? Are there 365/5 lines total per figure?

Figure 5: On the left, there is no top and lower panel... please adjust accordingly. Also, maybe identify "thick black line" as "vertical black line"

Figure 6:Please identify the variable that goes with each unit next to the colorbar. Always good to specify units of all variables in the caption too. Also, define abbreviations in all figures e.g. dissolved inorganic carbon (DIC). Figures and captions should be readable without reading the manuscript. Since you refer from figure 5 to this figure, you should mention here that July 11, 2000 is indicated in figure 5. Please show the transect line again in the map and label it with lat and lon. Why not also show a profile of pCO2 here to make the point that pCO2 decreases during upwelling event.

Figure 7: Add “the” to …the values from the nearshore bin…

Figure 8: What do the error bars mean? What are they based on? What are the numbers behind +/-? 1 STD? Needs to be defined in caption.

Figure 9: Why are some bars faded? Are all ingassing bars faded? Needs to be defined.