Review

This manuscript presents a substantial and successful effort to employ 14CO2 measurements to disaggregate urban CO2 flux measurements into fossil and non fossil components. This is an important contribution to urban eddy covariance flux literature that is definitely worthy of publication. Eddy covariance flux work is prone to detailed debates over methodology.  I have no significant concerns, however, with the methods chosen by these investigators. They have documented their choices well and all appear defensible and reasonable.  I doubt that alternative choices would have a significant impact on their results.

My primary concerns are with the presentation of their results.  The results and conclusions, and the associated elements of the abstract, deserve, in my opinion, significant revision to ensure that this manuscript is as impactful as possible.  I will call this "major revisions" but they are mainly editorial revisions, not major revisions to the scientific content of the manuscript. If the authors do not find these suggestions to be helpful I would respect their decision.  And as always, apologies for any misunderstandings.  I look forward to your responses.

Major suggestions.

The results are not clearly written.  The text of the results wanders from methods to descriptions of the content of figures (which should be contained in the figure captions) to a description of the results. The main results are often buried near the end of the relevant sections and are not always stated strongly and clearly. I suggest that the authors pick out the main results - their primary findings - and reorganize the text of the results around these findings. Second, the portions of methods in the results that are unique (means of analysis of the flux measurements) should be moved to the methods section of the document. Redundant methodological text should be removed. The sharpened focus on the main results should be translated into the abstract. Finally, I suggest removing the summary of results from the conclusions and focusing the end of the manuscript on a strong set of take away messages, passing on the recommendations this work yields for how our community should move forward.  I believe that a revision with these guidelines (see more notes in the detailed comments) will significantly improve the clarity and impact of the manuscript.

I have a second overall suggestion.  The authors discuss the use of CO for disaggregating ff and nf CO2 fluxes and the benefits of using 14CO2. The results, however, do not revisit this issue at all.  Are CO measurements available?  Is it possible to add a discussion of the CO/CO2ff ratios implied by the 14CO2 measurements to this document?  It would improve the impact of the manuscript even further.  If this is beyond the scope of the authors for this manuscript, could the manuscript at minimum state the availability of CO measurements for this purpose?

A third major (but relatively minor note): The flagging approach is confusing.  I suggest trying to clarify this presentation.

1. Abstract.

I believe that demonstration of this methodology and its strengths and weaknesses is the primary contribution of the manuscript. The abstract primarily presents the somewhat limited information about the mixture of ffCO2 and nfCO2 fluxes that can be derived from the available observations without saying much at all about what the authors learned regarding how to make this methodology work.  I would put more focus in the abstract on the methodological progress and use these urban system flux results as examples of what can be learned from this methodology.

2. Line 4, “with mean ffCO2 contributions of about 80%.” 

Is this a magnitude?  How is the potential for the changing sign of biological fluxes (even in winter) taken into account?

3. Line 7.  “the observed nfCO2 fluxes were predominantly positive (∼50% of net summer fluxes),”

Could you be more precise about what this figure represents?  CO2 fluxes vary a lot with time. 

4. Line 38. “air is conditionally collected for one hour in an updraft or downdraft reservoir”

Air is separated into updraft and downdraft reservoirs, correct?  Each hour doesn’t result in either an updraft or a downdraft reservoir being filled. Please clarify / rewrite.

5. Line 70-72. “The 20 Hz CO2 measurements of the IRGASON were despiked by discarding measurements where the median absolute deviation of less than three consecutive observations was outside the upper and lower limits defined by Mauder et al. (2013).”

Apologies, I don’t understand this. 

6. Line 72. “were upsampled to 20 Hz”

What does it mean to be upsampled?

7. Line 75. “the time lag was determined via linear interpolation to correct clock drift”

Linear interpolation of what?  What clock drift?

8. Lines 93-95. “Since the CO2 concentration measurements of the MGA7 showed a better agreement

with the measured flask concentrations than the IRGASON measurements (Sect. 4.1), the fluxes calculated from the MGA7 measurements were used when available,”

The spectral response of the sensor is more important for flux measurements than the comparison to flasks.  Have you compared the spectra of the MGA and the IRGASON?

9. Line 106-110. 

Is there any directional dependence of the mean vertical velocities?  How sensitive are the results to the methods used to compute the deadband?

10. Line 126. “co-located high-frequency EC measurements of net CO2” 

Net CO2 flux?

11. Line 137.  There is also turbulent sampling error in the vertical mole fraction differences.  I cannot, however, recommend readily accessible approaches from the literature for estimating this uncertainty.  But it would be worth noting that this exists. The upper limit could be estimated as the sum of random errors in each individual mole fraction measurement. Surface layer variances from similarity theory and turbulent sampling error theory (e.g. Lenschow and Stankov, 1986) could be used to estimate this uncertainty.

12. Line 147-149. “The ffCO2 contribution to the storage fluxes equals the ratio of the flux averages over the period during which CO2 accumulated below the measurement height, which does not necessarily equal the surface flux ratio during the measurement period.”

I do not object to the authors’ choice to exclude the storage flux from their flux estimates and to present data from low turbulence conditions separately but I do not understand this statement.  Storage flux represents accumulation in the column which occurred during the time period of the turbulent flux measurement.  I don’t understand how the time period for storage can be significantly different from the time period for the turbulent flux measurement.

13. Line 193-194. “Following Hensen et al. (2009) and Osterwalder et al. (2016), measurements

for which Eq. (3) does not provide reasonable ffCO2 fluxes were flagged according to β (Eq. 2).”

This criterion for flagging isn’t clear to me.  How are “reasonable” ffCO2 fluxes defined?  And what does it mean to flag a flux “according to beta”?

14. Line 194-196. “Measurements with large uncertainties due to the limited resolution of the 14CO2 differences between updraft and downdraft samples were flagged based on the signal-to-noise ratio SNR, defined as the minimum of the relative FffCO2 and FnfCO2 uncertainties.”

Similarly, this criteria isn’t entirely clear.  Is there a threshold SNR below which a flag is applied?  If so, what is that threshold?  Or is there a SNR estimate given to each flux measurement?

15. OK, I see that my questions are answered mostly by Table 1.  Please point out Table 1 before describing these flagging cases so that the reader can readily find the threshold values while reading the conditions.

16. Table 1.  I am a little slow, but the flagging scheme isn’t clear to me.  Conditions followed by “/“ mean the data are not considered further.  What does the & represent?  “The other flags were only assigned if all criteria were met”  I don’t really understand what that means.

17. Lines 214-217 are redundant to earlier text, no?  Please combine / condense the texts.

18. Lines 223-224. “along with the 10 - 80 % source areas for the well-mixed REA

measurements.”

Please specify the averaging interval used to create these source area estimates.

19. The methods section lacks the methods for analysis of the resulting flux measurements.  Please describe, briefly, your analysis methods.

20. Line 274.  “Regular” not “Reqular”

21. Line 274-277. “However, as the quality of the REA measurements varies depending on the micrometeorological conditions during sampling and the signal-to-noise ratios, the analyzed REA measurements were flagged as well-mixed measurements, low-turbulence and storage

measurements, or were not considered further (Sect. 4.1).”

Wasn’t this already described in the text associated with Table 1?  Please reduce redundancy in the text.

22. Line 274.  This paragraph is methods.  Please merge this into the methods section and remove this from the results.  These are not results.

23. Line 285-286. “the criteria finally considered, describing suitable well-mixed conditions”

This is verbose and ambiguous.  Do you mean just well-mixed conditions, or do these cases satisfy all of the criteria outlined in Table 1?  Please clarify.

24. Table 3 caption. “flagged as well-mixed, low-turbulence, storage or not considered further (Sect. 2.2.4).”

Apologies, but I don’t understand what this means.  Please make your data filtering choice easier for your readers to follow.  Perhaps a flow chart would be helpful, instead of a table, at some point in the methods.  I believe that Table 3 is showing a progression of filtering choices.

25. Table 3. I don’t see a row that presents how many flux measurements you DO consider.  I also can’t tell if the “”not considered” cases overlap, for example, with the well-mixed cases.  The number of cases don’t add up to the first row. I think that the first three rows add up to 100% of cases.  Maybe you could add the statement that all ffCO2 flux cases are “categorized either as well-mixed, low-turbulence/storage, or not considered due to quality control flags”.

26. Line 287-298. “In Paris, low-turbulence and storage measurements were mostly discarded.”

Did you use different filtering criteria for the Paris data?  Or is it just that the low-turbulence data in Paris were mostly filtered out by the quality control flags?

27. Line 314-315. “To illustrate the principle and to show an example of partitioning net fluxes of CO2 collected via EC into fossil and non-fossil CO2 flux components using REA, Fig. 2 presents data collected on 09 October 2024 in Munich. On that day, micrometeorological conditions were suitable for REA measurements and six flask pairs, sampled between 08:00 and 19:00 local time (UTC+2), were analyzed for 14CO2. The hour-long sampling periods are highlighted in Fig. 2 in light blue.”

Apologies to be an old professor, but allow me to make a suggestion.  I would tell my students that these sentences are a combination of a statement of methods and content suited for the caption of the figure. This is not appropriate for the topic sentence of a paragraph in the results section.  Move these sentences to the methods orto the figure caption. Please begin paragraphs in the results sections with topic sentences that summarize the results that you are presenting - tell us what you have learned. Use the paragraph to explain that topic sentence in more detail.  I believe that if you follow this advice your text will be more concise and focus more clearly on your primary results.

28. Line 327-329. “The continuous EC measurements (Fig. 2 d) show that the turbulent CO2 fluxes at 85 m height are approximately 10 μmol m−2 s−1 in the early morning, increase after sunrise, and reach a maximum of more than 60 μmol m−2 s−1 at noon, before they decrease again.”

Please excuse another old professor comment.  This text describes the contents of the figure.  The figure does that already.  Use the text to tell your readers the story of what you have learned from the figure.

29. Lines 349-351.  I believe this is what you are presenting as your main result.  I have two suggestions.  First, please begin this section with this statement.  Don’t bury it at the end of a lot of complex text.  Second, also please consider the significance of demonstrating the feasibility of extracting ffCO2 and nfCO2 fluxes using your techniques.  I believe this is also part of what you see as the significance of these results.  State that, please, unless I am mistaken.  I believe this is significant and a worthwhile result to state.  You may not have achieved everything you wish but you already present the caveats and limitations and you can explain, briefly, what you didn’t find that you might have wished for (e.g. traffic peaks in the ffCO2 fluxes?), and the reasons this hasn’t yet been obtained (e.g. sampling costs, SNR issues).

30. Line 356-358. “In this way, the ffCO2/CO2 flux ratios (RffCO2 ) as well as their temporal variability and

representativeness can be classified and differences or similarities between the three cities and measurement campaigns can be analyzed qualitatively.”

I don’t believe that this flux ratio should be presented as specific to each city.  These are specific to the site locations and sampling times for each flux measurement.  The flux footprint and sampling times are very local. 

31. Line 354-359.   Please move to the methods or figure captions.

32. Lines 360-364. “In the right panels of Fig. 3, the 1:1 line marks the case when the net CO2 flux equals the ffCO2 flux and the nfCO2 flux is approximately zero. Accordingly, measurements above the 1:1 line have a net positive nfCO2 flux component, while measurements below the line have a negative nfCO2 component, i.e., photosynthesis has dominated. The magnitude of the nfCO2 flux is indicated by the parallel dashed lines and the axes on the right, and the FnfCO2 uncertainties by the vertical error bars.”

Figure caption, please.

33. Lines 369-370.  “For comparison, the median fluxes of the continuous EC CO2 measurements are shown in the left panels of Fig. 3 for both summer and winter.”

Figure caption please.  What did you learn?  I can’t yet tell.  Start by stating your results - what you have learned.  Then explain how you have learned it.  Let the figure captions do the job of explaining the contents and structure of the figures.

34. Section 4.3.1 and the other results sections should be rewritten, in my opinion, with these principles in mind.  As written the text is a mixture of methods, figure captions and some of what you learned, but the “learning” (the results!) are difficult to extract.  Lead with what you have learned.

35. Figure 3.  I would strongly recommend showing the fluxes in local time, similar to Figure 2.  Human and ecosystem fluxes are best described as a function of local time, not UT.

36. Figure 3.  The error bars in the y-direction on the REA fluxes do not include the uncertainties estimated in the 14CO2 measurements?  The caption suggests that only random sampling error is included.  Why not, as noted in the methods, include the merged impact of both significant sources of random error?

37. Figure 3, caption.  What are the time periods covered by the continuous EC fluxes?  Is there any place where the number of EC data points per averaging interval is reported?  I don’t think this is essential to your story but it would be good to document.

38. Figure 3. Is it plausible that in Zurich and Paris the ffCO2 fluxes have a strong seasonal cycle?  In both cities the ffCO2 fluxes are noticeably larger in the winter.  This is not the case in Munich.  Perhaps Munich is more surprising.  Do the areas around the towers suggest that Munich, for example, is more dominated by traffic while heating is important to the ffCO2 fluxes at the other sites?

39. Line 383.  “Compared to the median CO2 fluxes, the fluxes during the selected REA sampling periods are often exceptionally high.” 

Agreed. Do you have an interpretation?  If so, please lead the paragraph with that interpretation.

40. I’m not sure that Table 4 adds much relative to Figure 3.

41. Lines 411-414. “As emissions from point sources are generally not representative of the average fluxes in a city, and the comparison of measured and modeled point source emissions on an hourly basis is

limited by uncertainties in the emissions inventory and transport models, we attempted to identify the REA measurements which were potentially influenced by emissions from the district heating plant.”

I would argue that the city is a collection of point sources and no single measurements are “representative of the average fluxes in a city.”  I do not believe that you should try to represent average city fluxes with these measurements.  I suggest that you drop the apologetic text.  I fully support the effort to interpret your measurements with a flux footprint model and knowledge of likely large point sources.  It is of course true that, “comparison of measured and modeled point source emissions on an hourly basis is

limited by uncertainties in the emissions inventory and transport models.”  But isn’t that part of what we are trying to test and learn - that is, the limits of our understanding of these fluxes and quantification systems? Embrace your effort.  Don’t apologize because the tools and methods are not perfect.  All models are wrong.

42. Section 4.3.2.  There are lots of methods in this text that is supposed to be the results.  Please move the analysis methods to the methods section of the manuscript.  Focus here on the results, please.

43. The discussion of the attempt to use 13C is interesting but, in the end, not used.   This is worth reporting for others who might try similar methods but perhaps this could be an aside?

44. Lines 443-444.  “The results thus emphasize that tall-tower measurements in urban environments can often be affected by individual point sources.”

I would argue that these results show that your measurements are doing what you hope they will do.  Of course large point sources have an impact - and you detect this impact!  - even when this is not fossil emissions and you have to use your decomposition to tell the difference.  I would change the focus of this analysis to one of verification of the efficacy of your methodology, rather than an apology for having measurements “contaminated” by point sources.  Cities are complex.  They are not homogeneous.  If we are going to make measurements in cities we must be able to deal with sources that are mixed in space, time and type.  You are demonstrating progress toward these objectives and scientific needs.  Embrace your progress and mission.  You do state this (later in this paragraph, someone buried).  I suggest that you move these statements to the fore - including the main message summarized in the abstract.

45. Lines 528-542. This entire paragraph is methods.  Please move it to the methods section.

46. The authors have sprinkled discussion into the results, but my intuition is that a brief discussion section would be helpful.

47. Lines 563-565.

I suggest that the conclusions should not waste text summarizing the results.  Summarize the manuscript in the abstract.  Provide the take home messages for the community in the conclusions.

48. Lines 569-570. “The Munich measurements show that with an improved technical setup and an adapted flask sampling and selection strategy, average nfCO2 fluxes of the order of 10 % or 3 μmol m−2 s−1 can be identified with a reasonable number of measurements (50 to 100).”

I don’t recall this being demonstrated in the results.  I do think this is a valuable statement and I’m not disagreeing with this finding but I would make this clear in the results.

49. Lines 573-574. “Situations with large fluxes are therefore favorable for the uncertainty-limited

REA measurements and were preferentially selected for sample analysis.”

Variances in the atmospheric surface layer decrease rapidly with altitude above ground (see decades-old similarity theory - Kamal and Finnigan’s text is where I go for a concise summary).  Lower altitude measurements would improve the SNR. It is true that you might be hesitant to do this in urban areas with large buildings but there are many urban settings with moderate building heights.  This would be a suitable topic for discussion - what are the strengths and limits of the approach you have chosen?  You could follow this by a brief conclusion section that provides recommendations for what you recommend for future investigations. At present, lines 571-577 are mostly summary statements.

50. Lines 579-587.

This text  is a summary of results.  I would make sure this is all clearly stated in the results, delete this text from the conclusions, and present your “take home” messages.  How should the community deal with point sources in cities?  How can the research community deal with the mixed fluxes typical of urban systems?The following paragraph is more suited for conclusions.

51. Lines 597-598. “This again highlights the fundamental challenges of extrapolating local observations to derive emissions at the scale of an entire city.”

I would not suggest direct extrapolation as an objective of urban EC flux measurements.  EC measurements are well suited to testing process-based models.  These models are the correct tools for extrapolation. 

52. Lines 599-604.

This isn’t compelling text for the conclusions.  This restates the results section and should be fairly well understood.  The take home message here, “be careful selecting your background,” is reasonable but vague. 

53. I suggest building on the “outlook” section for a sharper, targeted conclusions section of the document.