Adcock et al present a 12 year time series of O2/N2, CO2, and the tracer APO from a coastal background site in the UK. This is a unique, high quality dataset highly worthy of publication in ESSD. I recommend publication after some revisions. My main issue with this paper is that it is overly long without being detailed enough. At least half of the paper is devoted to an analysis of the dataset, which per the ESSD guidelines is not supposed to be included. This seems to come at the expense of detailed information on changes in the measurement system and technical issues, which are not fully described and could be of interest to anyone trying to actually use the data. 

My suggestion is to generate a complete list of calibration, target, and zero tank changes with IDs and assigned values (where applicable), and a complete change log or README type file where the major alterations to the measurement system are fully described, with exact dates. This shouldn't be too difficult to generate given the figures presented (i.e. the authors must have this information in hand to create the figures). There is a long paragraph of significant changes with approximate dates which would already form the basis of such a table. I am asking for this because in 10 or 20 years, someone may wish to analyze the time series but not have enough information as presented to understand whether a given feature is a real signal or an artifact.

The paper also contains a lot of verbalization of data which is already in a table. It's a very long paper, and cutting this redundant text will make for less text to sift through.

My last major comment is that the correction to the APO record is speculative at best. Forcing the WAO data to fit the CBA record is a creative approach, but the impact this has on the WAO data was not demonstrated. As I understand it it relies on the assumption that fractionation through a torn diaphragm was constant over shorter time scales, but not over several months. Is there any basis for making this determination? I understand the desire to salvage a significant chunk of data, and I think this is generally an OK approach, but the authors could have done more to convince me that it was reasonable. Could they perform the same baseline shift exercise on data which was not impacted at WAO and see how the residuals compare--do they have any structure? And what do the corrected vs uncorrected residuals look like? At the very least, this data should be better flagged in the data file--right now it is flagged as "2", which means "contact data provider". A separate flag for "corrected" or "baseline shifted" should be implemented. 

Minor comments:

General comment: The use of O2 mole fraction throughout is confusing. It is a term with a specific meaning, and I don't find it more convenient to alter its meaning for this particular paper.

General comment: The figures are not color blind safe, please use different symbols in addition to the colors selected, or use a different pallette.

L16: APO is not a tracer for terrestrial biosphere processes, this would be better phrased as "insensitive to terrestrial biosphere fluxes". 

L27-47: This is a fine introduction, but the paper is 47 pages long and its goal is simply to describe the data set. These two paragraphs could be consolidated into a sentence or two, pointing to some key references. I don't think it's necessary to list all of the ORs for different fuel types, for instance.

L52: A good place to include the equation for O2/N2

L58-59: I don't understand, per meg is used throughout the paper...?

L60-70: Suggest to cut this paragraph, it's not necessary to explain the data being presented.

L71: "Calculated term" is awkward, suggest "tracer" or "data-derived tracer"

L75: this is not what I understand a "conservative tracer" to mean

L125: How long is the inlet line? The mast height is given, but not the total distance from the inlet to the instruments.

L126: Could you briefly describe the inlet? Dimensions, how and how much it is aspirated, etc. What is the flow rate in the sample line?

L298-314: This paragraph could be shortened or cut by putting the data in a table. It's not necessary to state after each result whether it is smaller or larger than 2 or 10, the reader can do this on their own. The statement that "...any systematic drift over time, indicating long-term stability of the WAO O2 and CO2 calibration scales" is quite misleading. Both the CCL scale and WAO scale could be drifting together, and the drift does not necessarily have to be systematic. Scales can drift over multiple time scales for many different reasons, which can appear to be scatter when sampled sparsely.

Fig 4: Isn't the target tank a measure of the repeatability, not the compatibility? If so, the shaded bands should be half the compatibility goal as pointed out on L263.

Fig 5: The drift in the cylinders is not linear (except for maybe the last one), so I question the usefulness of showing linear fits to them. If the tanks are desorbing/fractionation with pressure, one wouldn't expect it to drift linearly anyway, unless it was leaking badly. Finally, this figure doesn't show any information not already contained in Fig 4. I recommend cutting it, or at least consolidating with Fig 4 by adding the fits to the data.

L395-419: This section could also be cut down, given that all of this information is in the table.

L448: These are per meg values, not mole fractions...maybe this is what is meant in L58? If so, it is confusing and looks like an error. "Value" would work fine here.

L474-479: There is a cumbersome emphasis here on the compatibility goal, usually with no added discussion. There is also a repeated pattern of verbally describing the values in a table within the text. I think all of this can be cut.

L533: Calibration cylinders should be remeasured to account for cylinder drift. I think the authors should try to constrain how much this contributes to the uncertainty. Also, From Fig 8 it looks like the changes in the slope are small due to changing of calibration tanks, but it should be shown that this is a small effect. Have the authors indicated whether the calibration coefficients are interpolated between calibrations, or applied as step changes? I may have missed this.

L673-688: The lengthy text on technical issues should be at least separated from the text on missing APO data. I think this information needs to be formatted into a table or change log of some kind, with exact dates, spanning the whole dataset. As it is only approximate dates are given and the user would have to guess if subtle features in the dataset might be artifacts pertaining to such changes. There is also not enough detail--for instance, "pneumatic valves" and "solenoid valves". Could the authors reference the plumbing diagram directly in a way which is unambiguous?

L687: "was not so it had to be removed from the dataset." -- missing word or typo?

L703-713: This is analysis of the dataset, per the "Aims and Scope" of ESSD: "Any interpretation of data is outside the scope of regular articles." 

Section 4.2: Same comment as previous, also noting the ESSD "Aims and Scope" statement: "Any comparison to other methods is beyond the scope of regular articles".

L754: "there is no discernible marine influence on the CO2 seasonal cycle" -- This is not correct, there is a significant component to the seasonal cycle of atmospheric CO2 from air-sea fluxes. If the authors mean no discernible contribution at WAO, this would be surprising given its location, and needs to be shown.

L805: APO has a diurnal cycle in many locations, better to specify "APO at WAO" to avoid confusion. One would in fact expect a strong diurnal cycle in APO at a coastal site like WAO due to land/sea breezes. It is also not correct that the boundary layer effects cancel. The changing boundary layer height dilutes or concentrates a flux signal from the surface with background air from higher up in the troposphere. If you had constant outgassing of APO from the ocean surface over 24 hours, for instance, you would still see a diurnal cycle from this effect.

Review's comments

Manuscript Number: essd-2023-129

Title: 12 years of continuous atmospheric O₂, CO₂ and APO data from Weybourne Atmospheric Observatory in the United Kingdom

Authors: K. E. Adcock, P. A. Pickers, A. C. Manning, G. L. Forster, L. S. Fleming, T. Barningham, P. A. Wilson, E. A. Kozlova, M. Hewitt, A. J. Etchells, and A. J. Macdonald

General comments:

The authors of this study present high-quality record of the atmospheric CO₂, O₂, and APO data observed at Weybourne Atmospheric Observatory (WAO) in UK for decadal period between May 2010 and December 2021. They carefully assess the stability of CO₂ and O₂ scales and the repeatability and compatibility based on the measurements of variety of cylinders including intercomparison round robin cylinders, Target Tanks, Zero Tanks, Working Secondary Standards and so on. These results reveal that the data at WAO have high quality and significantly reliable. They also investigate the characteristic features of the trend, seasonal cycles, and diurnal variations of CO₂, O₂, and APO. The data at WAO would contribute to various studies including the global carbon cycle, air-sea gas exchanges and so on. I found that the paper is well written and contains material that should be published in Earth System Science Data. I highly recommend the manuscript to be published with the minor corrections as suggested below.

Specific comments:

Page 2, line 51: The authors described that a standard with a known O₂/N₂ ratio is used to report the change in atmospheric O₂/N₂ ratio. Is it possible to show the exact number of the O₂/N₂ ratio of the standard scale of this study?

Page 2, line 52: I think the sentence “O₂ and N₂ mole fractions are affected by changes in trace gases” is a little misleading. The major atmospheric components like O₂ and N₂ are affected by the change in the total amount of the air caused by changes in any atmospheric components, which is called as a dilution effect. Therefore, O₂ mole fraction is affected not only by trace gases, such as CO₂, but also O₂ itself. The dilution effect is, however, negligible for the trace gases. Therefore, the direct comparison between O₂ and CO₂ concentrations is rather confusing when they are expressed as mole fractions.

Page 2, line 56-57: As far as I know, a mass spectrometric method, which is adopted by many laboratories, directly measure the O₂/N₂ ratio.

Page 2, line 58-59: The authors describe that O₂ variations are refer to as O₂ mole fraction changes rather than δ(O₂/N₂) ratio changes in this manuscript. But δ(O₂/N₂) ratios are used in the most of this manuscript.

Page 5, Figure 2: I think it would be better to add an aspirator and a differential pressure transducer in the legend.

Page 6, line 147-150: I’m curious about how to balance the pressures and flow rates between the sample air and WT air streams. In the manuscript, the authors described that the balance is manually achieved by adjusting the two needle valves. Is it possible to keep the balance for long period? In the Figure 2, the differential pressure transducer and the solenoid vale are connected to the “MKS” differential pressure gauge via green lines. Does it mean that the solenoid valve is automatically controlled to achieve the balance of the pressures between the sample air and WT air streams?

Page 6, line 149-150: Is “the two manual needles valves” a typo?

Page 6, line 152: Does “A solenoid valve” correspond to “4-way switching valve” in Figure 2? Are those same things?

Page 6: I think it would be better to clarify the flow rates of the sample air and WT air in this section of “Analytical set up”. I know the flow rate (about 100 ml/min) is mentioned in in line 599, but it would be better to mention it here too.

Page 8, line 189-190: Don’t the authors use the interpolated calibration coefficients from the bracketing calibrations?

Page 13, Figure 3: The shade of ±10 per meg range is unclear.

Page 14, line 340 (Figure 4 caption): “Target Tank (TT) measurements of CO₂ (top panel) and O₂ (bottom panel) at …”

Page 15, line 351-352: “… with slopes (in ppm year^-1 and per meg per year^-1 for CO₂ and O₂, respectively) …” “…each TT, for CO₂ (top panel) and O₂ (bottom panel) …”

Page 27, line 618-619: “Manning, 2001” is not listed in References.

Page 33, line 725-726: It would be better to clarify what the ranges in the parentheses mean. Are they 95% confidence intervals?

Page 35, line 767-768: I think that the effect derived from seasonal and/or diurnal covariance between surface fluxes and atmospheric transport including PBL dynamics is termed as rectification effect. The seasonal cycle of PBL height itself isn’t termed as the rectification effect.

Page 46, line 1036-1037: “Stephens, B. B., …, 2000” has been already listed in line 1033-1035.