Review of egusphere-2022-710:

High interannual surface pCO2 variability in the Southern Canadian Arctic Archipelago’s Kitikmeot Sea.

Richard P. Sims, Mohamed Ahmed, Brian J. Butterworth, Patrick J. Duke, Stephen F. Gonski, Samantha F. Jones, Kristina A. Brown,  Christopher J. Mundy, William J. Williams, Brent. G. T. Else

Overview and general recommendation

The study of Sims et al. presents recent (2016-2019) underway measurements of pCO₂ in the Kitikmeot Sea of the Southern Canadian Arctic Archipelago. By employing a suite of sensors in a custom-built setup onboard a smaller research vessel based in the region, they were able to survey pCO₂ shortly after ice breakup. They also surveyed less frequented shallow bay areas where few, if any, measurements were made previously. The authors estimated the CO₂ air-sea flux and found the region to be a net sink in summer, with substantial interannual and spatial variability. The authors discuss their results in the context of data from two nearby ocean observatories (a mooring and an eddy covariance tower) on local and regional scales. The authors also discuss interannual variability and large scale seasonal trends, putting their results in context of other recent studies from the region by the authors. One of the key findings of the study is that the surface pCO₂ values at the time of ice breakup and ice melt is important in constraining the magnitude of the air-sea flux throughout the summer ice-free season.

The presented datasets (supplemeneds and to be published at Zenodo) are from an extremely data-sparse region of the Arctic and, as is also stated by the authors, constitute an important new baseline for gaining a better understanding of the role the region plays in the uptake of atmospheric CO₂. The study is important, timely, and well motivated. The manuscript is well written overall, and the presented method is sound and descriptive. General issues need further attention from the authors. Addressing these comments below in a revision, I have no reservation for the manuscript to be published in Ocean Science. Please find my major comments below, followed by some minor comments that are referred to by the line numbers in the manuscript. 

Major comments

My main concern with the manuscript is the general lack of concretely identified procceses and controls that may explain the observed results, which makes the manuscript read more like a data descriptor report. Without further information from measurements of ancillary variables from the underway system or from CTD/Rosette systems in the vertical, I recognize that it is difficult to both identify and quantify controlling processes. However, I urge the authors to try to expand on this effort to make the study even more useful to the Ocean Science Community. It would be helpful if the results could be put in to context of different controlling processes, even if it means theoretical calculations and approximations. What is the expected response in pCO₂ when the temperature or salinity changes over the observed ranges? What effect would mixing of fresher waters with more saline waters have on the non-coservative behaviour of pCO₂? This exercise can be readily estimated from theoretical calculations (see minor comments). Did the authors consider applying any other models than the fitted relationship between  pCO₂ and weeks since ice breakup from Ahmed et al. (2019) to explain the observed values (see minor comments)? What I am trying to convay is that it would significantly strengthen the study if the results can be put in a much more clear and quantitative context, despite missing information from additional variables. My second major concern is that the results are mainly described by relative statements without actual numbers backing up the statements (although listed in Table 1), e.g., "...there was large interannual variability...", "...was generally lower than...", "...values were much lower...", "...highly undersaturated...".  This makes it difficult to digest the results in a meaningful way. Please consider including values/ranges/numbers and avoid relative statements.

Minor comments

Line 34: italicize p, subscript 2, for consistency

Line 35: Define CAA, preferably on line 21.

Lines 49-50: If possible, please provide an original reference (e.g., Jakobsson (2002)) to this areal statement as there are many different definitions around. Bates and Mathis (2009) do not include such a reference.

Line 179: "A made to order Sunburst..." reads awkward, please rewrite.

Line 189: Change "x" to the greek letter chi

Line 198: "processed following SOP 5 (Dickson et al., 2007)."

Lines 221-224: Any critical problems that warrants a notice in the main text?

Line 228: "...should be quite similar" Please avoid such relative statements. For example, compare observations from Barrow/Alert NOAA GML Carbon Cycle Cooperative Global Air Sampling Network. The difference between the two station means (1985-2021) is 3.8 ppm.

Line 234: The scaling factor (SF=0.24) is superfluous as it is inherently included in the calculations when the flux is given in mmol m-2 d-1, based on the given units for the gas tranfser velocity, solibility, and partial pressure difference. Suggest to omit SF as it may be confused with a scaling factor for sea-ice cover, although the study concerns open water.

Line 235: Why Nightingale et al. (2006) and not Wanninkhof (2014)/Ho et al. (2006)? Please motivate.

Figure 4: Please consider changing scale of the the y-axes for the different years. I recognize the benefit of having the same scale for all years, but at the same time it is very difficult to make out any fine-scale patterns between the different variables.

Lines 308-309: Please mark the locations of the ONC mooring and Qikirtaarjuk Island observatory also in Figure 2. 

Figure 5: Please put labels of a), b), c) in the figure.

Line 335: How is "good agreement" defined? There is no "good agreement" bewtween the EC tower and the ONC mooring October 2017?

Line 377: "Dilution by low pCO_2(sw) ice meltwater", remove the subcripted "(sw)". Please consider undertaking the exercise of theoretical calculations on the non-conseravtive behavior of pCO₂ during the mixing of "fresh" and saline water, following Figure 11 in Meire et al. (2015). This could be useful in the discussion on how much a salinity change could/would lower the pCO₂ during mixing of waters of different salinities.

Line 381: Please clarify at which depths the Freshwater Creek plume is typically found.

Lines 387-389: The sentence starting with "On the 17th August 2017..." is very long and reads somewhat awkward. Suggest to break it up and rewrite the part "...this would point to this being due to something only happening in the Bay..."

Line 422: change to "...Ahmed et al. (2019) did..."

Line 437: Suggest changing "oversaturation" to "supersaturation" throughout the text.

Line 460: Would it be helpful to derive your own similarly fitted model (pCO₂ vs. weeks since ice breakup) for Kitikmeot Sea? Did you consider applying a different model, like the one (Figure 3) by DeGrandpre et al. (2020), to try and explain some of the observed results?

Line 468: Subscript 2

Line 499: "air-sea flux"

References

Jakobsson (2002):  https://doi.org/10.1029/2001GC000302

Meire et al. (2015): https://doi.org/10.5194/bg-12-2347-2015

DeGrandpre et al. (2020): https://doi.org/10.1029/2020GL088051

Review of: High interannual surface pCO2 variability in the Southern Canadian Arctic Archipelago’s Kitikmeot Sea by Sims et al

Overview: The MS by Sims et al presents new data for the Canadian Arctic Archipelago in the area of the Kitikmeot Sea. The observations are largely collected from a ship-of-opportunity, but also data from a seafloor observatory platform and an eddy covariance flux tower are also presented. Four years of summer observations are used to define this region as a sink for atmospheric CO2. Notable spatial differences in the data were highlighted, as were differences from the seafloor platform and the flux tower. My impression with this study is that the data are unique but I wonder about the discussions around the comparison to the flux tower and the seafloor node, and I can’t help but think about the missed opportunity to compare these new results to data in the SOCAT holdings. For instance, a quick check of SOCAT reveals there are underway surface measurements in this area from Mike DeGrandpre for the years of 2017, 2019, and 2020, not to mention the earlier data from Tim Papakyriakou. It isn’t clear to me that the ONC data is directly comparable to the surface data given they are measurements from the sea floor, and without understanding the size of the “footprint” of the EC tower-determined surface pCO2, that isn’t a directly obvious comparison point either. Perhaps linking to the SOCAT holdings could benefit various sections of the discussion, as well as lead to a discussion about trends beyond inter-annual variability. Discussion of the drivers of the spatial and temporal variability is pretty limited and could be strengthened as well. There are also some corrections that need to be made in the presentation of the methods. I urge the authors to consider these points in addition to my detailed comments below. This paper certainly is worthy of publication in Ocean Science after the authors address these comments, and I really liked seeing the setup on the R/V Martin Bergmann. Best of luck, Wiley Evans.

Specific comments:

ONC seafloor platform depth is reported to be 7 and 9 m on pages 3 and 19, respectively? Which is correct? Seems like a potentially big difference in the stratified Arctic.

Page 7, please report the scale used to present salinity observations.

Line 185, page 9, the need for water vapor correction stems from the fact that drying removes water vapor and this impacts the partial pressure. For instance, unadjusted pCO2 from a GO8050 (that has drying components) would a “dry air” value that needs to be “corrected” to 100% humidity using SST and salinity to compute vapor pressure and adjust pCO2 to a “wet air” value. If an analytical system does not dry, then there is no need for a water vapor correction. Therefor this statement and the application of a water vapor correction to the data is in error.

Line 199, page 9, the LI-840 does not measure pCO2. The measurement is CO2 absorption that is linearized to produce CO2 mole fractions over a broad range. See: https://www.licor.com/env/support/LI-840A/topics/theory.html. Raw xCO2 from the instrument would be calibrated using multiple reference gases, and this calibration function should be a linear fit between the reference gas concentrations and the raw xCO2– not a piece-wise linear fit. The first point here is an easy correction to the text, the second point needs addressing at the data processing level.

Page 9, the authors state in situ temperature and salinity were adjusted for “ubiquitous” skin effects when calculating “interfacial” pCO2. I believe this means pCO2@equilT was adjusted to pCO2@skinT, but is presented as pCO2@SW. Was the relationship from Takahashi et al also used for the salinity adjustment?

Page 9, The Wanninkhof 2014 relationship is used for Schmidt number but Nightingale et al 2000 was used for the gas transfer rate, why is that? Why not use Wanninkhof 2014? Also, was there good agreement between the reanalysis and locally observed wind speeds?

Page 10, expressing pCO2 uncertainty as an absolute value is a bit misleading as certainly the uncertainty is less than 8 uatm at 200 uatm and likely more than 8 at 600 uatm. Instead of “final” could say “average”? Suggest sticking to expressing uncertainty as a percentage.

Page 10, somewhere the size of the footprint of the EC tower needs to be defined. Also, what are the uncertainties in SW pCO2 determined for the EC tower? For instance, the authors use temperature and salinity from 13 m (i.e. not surface and deeper than the ONC platform) to compute Schmidt number and CO2 solubility. Does this, in addition to the spatially integrative nature of the EC tower determined SW pCO2, contribute to the reported differences from the underway pCO2 measurements. That is in addition to the surface skin effects? Given underway pCO2 was adjusted for median surface skin effects (

Page 13, Lines 281-294, 2016 doesn’t look to be “close to atmospheric equilibrium” in Figure 4, though the areal averages in Table 1 indicate conditions were closer to atmospheric levels than during the other years. I was surprised by the degree of variability during 2016 relative to the other years. Maybe this is a point that could be built on RE drivers?

Table 1 legend: the cautionary note seems a bit odd since the comparison between years is done in Discussion section 4.3. Maybe remove this statement?

Figure 5 and section 4.1: the legend states “surface pCO2 from across the Kitikmeot Sea” which I think means all the data in Figure 4. But it doesn’t look like all the data are shown. Suggest to use only data from within the EC tower footprint, whatever that is, so as to be more directly comparable. This might help clarify this section and better support the statement on Lines 330-332. The additional SOCAT data might also help in this section as well.

I don’t’ follow the comparison to the seafloor platform without some understanding of how temperature and salinity also compare. Could this be added to Figure 5?

Sections 4.2 and 4.3 would benefit from comparison with the SOCAT data holdings.

Section 4.4 title, suggest replace “sink for pCO2” with “sink for atmospheric CO2”

Data availability: while I appreciate that the authors are making their data available through Zenodo, and I applaud them for the effort, these data would be a bigger benefit to the community if they are submitted to SOCAT and NCEI. I strongly suggest the authors consider submitting these data to SOCAT.