|The authors have made some improvements to the previous version of the manuscript and answered satisfactorily to some of my previous comments, but the bigger comments from last comment round were not taken into account properly. There are still four major issues (in addition to few smaller ones) that require addressing before publication in AMT.|
1) The manuscript and its analysis would really benefit from statitistical tests used to check wether the fluxes measured by different techniques really differ or not. The amount of chamber measurements is probably not enough for this purpose, but fluxes from EC, IDM and gradient flux methods could certainly be used for such tests. You should also test if the nighttime and daytime CH4 fluxes really are statistically the same or not.
2) About spectral corrections of the flux measurements. The authors responded to the request of spectral corrections that the high frequency spectral correction is not important at a measurement height of 18m and changed the average flux by 0.8%. While it is true that the importance of smaller eddies decreases at higher measurement heights, they still cannot be neglected at 18m (not high enough and smaller eddies probably still exist). In addition, the importance of low frequency spectral correction increases with higher measurements, and should be accounted for in the analysis. Why are the spectral corrections still not included in the measurement description? What kind of spectral correction method(s) did you use? You should check the power spectra of your measurements to check if high- and/or low frequency spectral corrections are needed. I am not conviced by only comparing the average flux, without low frequency correction. If you still are omitting the spectral corrections, you should justify it in the text very clearly and include the power spectrum and cospectrum that proves it.
3) The fact that the gradient flux measurements are relying heavily on EC CH4 flux measurements is still not discussed with the flux comparison results. I well understand that the gradient flux was calculated based on a fit to CH4 EC fluxes and not directly to the fluxes, but there is still a strong link. What would make it a bit more reliable is to test the methods I recommended in the previous comments or make the same fit you have now done but using CO2 fluxes instead of CH4, and then calculate the gradient fluxes in similar fashion. In any case, it should be discussed in the results how your calculation method affects the comparison!
4) Last but not least, many of my previous comments were either ignored (e.g. text is not well organized: in methods section EC and gradient instrumentation descriptions are not well separated, in the results on-pond and off-pond results are mixed and the reader easily gets lost, spectral corrections are still ignored, gradient-EC (in)dependence still not discussed...), not applied everywere in the manuscript (for example eddy covariance → EC, including medians together with means, some clarifications, e.g. about tube dimension), or were answered in the author response, but not applied in the revised manuscript (e.g. discussion on shifting winds affecting gradient fluxes are also affecting EC fluxes, medians not included in the text even when they were in the response). Many of the comments listed below are the same as before, as they were not implemented in the revised manuscript. Overall, it seems not much attention was given to the revision of the manuscript (e.g. a figure is referred that does not exist).
Detailed comments (line numbers refer to line numbers in the “track changes” version of the manuscript):
Table 2: What are BDL and NA (mention in caption)
Fig 2: Write the correlation coefficient in the plot....
Fig 3: What I meant by my previous comment, was that the bins are not of equal size in terms of number of data points included in each bin. And that is why the fit does not really follow the original datapoints. How does the fit change, if you use equal number of datapoints for each bin? What is the fit equation?
Fig 5: I still advice to use a uniform color scheme here, as the present one is highlighting differences in the range of 1.9-3 ppm, which are not really as dramatic as the ones from 1.9-9 ppm. My suggestion is to use a uniform color scheme, e.g. similar colors as you have used in scales 3-9 ppm. I recommend to read a recent Nature Communications article about colormap choises (https://doi.org/10.1038/s41467-020-19160-7). Adding the radius lines (0.2, 0.4, 0.6, 0.8, 1) on top of the wind rose would help the reader a lot (same goes for Fig S1).
Fig 6: The Xtick labels are too close to each other. Widen their distance or make the labels e.g. in 45 degree angle. You can also leave out the hours, just put dates in each xtick and specify in the caption that date tick represents midnight. Why is the lowest panel missing horizontal grids, when other panels have them?
Fig 7: The shaded areas are not the same that you mention in the text!
Fig S5: Somehow mark in the figure daytime/nighttime.
Add the footprints of stable/unstable/neutral conditions to the supplement.
L18-20: Larger footprint together with frequent sampling.
L29: oil sands
L56: Not only emissions, but also uptake, can be measured with the technique. Change to "gas fluxes" or "surface-atmosphere exchange"
L59-60: The sentence still reads that the fluxes are well defined spatially. Reformulate the sentence.
L74: 1/2" inner or outer diameter?
L77: H2O not defined. Mention that these are used for EC measurements
L78: Even with the numbers and formula you gave in the response, I get the Reynolds number ~1600. I don't understand how you get 4500.
L73-78: Lots of confusion about sampling tube lengths. In the original manuscript gas concentrations were measured through 45 m tube and EC through 40 m tube. In the response you mention 40 m for gradient and 30 m for EC, in the revised manuscript gradients are measured with 45m and EC with 30 m. This should be quite straightforward and well documented...
L89: "Friction velocity (u * ) can also be calculated from measured u, v, and w.". There is no point to this sentence if you don't present the equation, or tell how it is calculated.
L116: EC has already been defined....
L120: "which in this study limits the method to sensible and latent heat (H2O) fluxes, momentum, CO2 and CH4 fluxes." Since you don't show or describe other measurements, this sentence is unnecessary
L121: eddy covariace -> EC! Check this everywhere in the manuscript!
L121: replace "..the eddy covariance method simply calculates the flux by averaging.." by "in the EC method, flux is calculated by averaging…"
L126-130: Mention that you assume linear concentration profile for storage change flux calculation
L132: Mention the average time lag in the text.
L130-135: Mention that each wind sector was processed individually and how you took into account the different roughness elements of different wind sectors.
L190: Again, shifting wind directions are also a problem for EC, not just gradient fluxes! So these occasions are not probably due to shifting wind!
L200-203: Mention the units you have used!
L218: I looked the report cited here and still did not find how the chamber fluxes are calculated. Please provide a formula for flux calculation. I understand this may not be the focus of this paper, but chamber fluxes are quite sensible to the calculation method used, and thus I think it is valuable information for the reader.
L219: N2O not defined
L220-223: Again, I still find this section quite confusing. Is this then the final flux you use in the flux comparison, or something else? Which flux measurements are used for this? EC, gradient, chamber, IDM? Provide more details.
L303-308: Again, it is not clear wether you are now describing the off-pond or pond fluxes. For pond fluxes, wind should play at least some role, enhancing the turbulent transport of gases from the pond. Specify in the text if you are focusing on pond or off-pond fluxes in this discussion, since in the previous paragraphs you are describing both. As this indeed is from pond direction only (as you state in the response), discuss why they should not play any role, i.e., why more mixing would not bring up methane produced deeper in the pond. Looking at Fig. S3, even though there is no clear linear relationship, it is still clear that lowest fluxes are not measured at high wind speed and highest fluxes are not measured at lowest wind speeds. So there is some kind of relation to wind.
L324: Report also the median flux differences, as requested before.
L327-338: You are still not discussing the relation to EC fluxes. Yes, the Kc was determined from a fit made to EC fluxes, and yes, almost all gradient flux methods require some input of the EC system. But how would the results look if you determine Kc from CO2 flux instead of CH4 flux? That would lead to at least a bit more independent comparison to EC CH4 flux. By minimum, you should discuss how the derivation of Kc from EC CH4 flux is affecting your comparison, or justify why it is not affecting at all.
L335: Which studies? Add references to the sentence.
L340: Again, you cannot really say they are lower when the medians are almost the same, and also meand within the confidence intervals. It would be good to try some statistical tests to see if they really differ or not. Add it to the discussion.
L351: There is no Fig 9... You might mean Fig. 8b.
L366: Again about the bubbling zones.
Response: “The locations of the 15 flux chamber measurements are marked in the revised Figure 1. As can be seen, most of them fall within the EC footprint. It is possible that the sudden bursts of CH4 could invalidate the flux chamber calculation and lead to an underestimation of flux, as discussed in Zhang et al. (2019). We wrote this in line 334-337. Integrated over the footprint of micrometeorological flux measurements, the intermittent nature of ebullition will have a minimal effect.”
I did not find any discussion on lines 334-337 on the subject (on the previous manuscript, on the revised manuscript, or the author response).
L416: As a side note, Pond 2/3 is a great name for a pond!
L459: Abbreviation CO2eq is (still) not defined
Somewhere in the discussion, mention how high these CH4 emissions are compared to natural (wetland/lake/pond) emissions from other studies, just to give some idea of the flux magnitude.
L462: Mention the methods used.
L466: "in 2017" is not needed here
L467-468: "micrometerological flux measurements"
L468: "larger footprint together with high temporal resolution"
L469: To be accurate, the measurements are still not representing the whole pond. But they are representing most of the pond area. Reformulate the sentence.
L471: Further studies of what? EC, chamber, IDM, gradient, temperature, wind, what? Flux measurements in general? At what time resolution?