Review of “Using Open-Path Dual-Comb Spectroscopy to Monitor Methane Emissions from Simulated Grazing Cattle” by Weerasekara et al., submitted to Atmospheric Measurement Techniques.

General Comments

This manuscript examines the potential of a line-averaging gas sensor (based on dual comb spectroscopy, DCS) for use in calculating methane (CH₄) emissions from grazing cattle.  This is an important topic as grazing cattle are significant CH₄ emitters, and a sensitive and robust CH₄ sensor is needed to measure emissions in that environment.  The subject is appropriate for Atmospheric Measurement Techniques.  

The paper is interesting but there is a lot going on in the short manuscript.  What are the objectives of the paper: a modelling study to estimate CH₄ enhancement levels downwind of cattle; a gas release trial to determine the enhancement; an examination of DCS sensitivity relative to the cattle signal; an examination of the accuracy of the IDM technique for calculating emissions?  The manuscript touches on all of these, but generally without enough description and discussion to address each adequately.  The authors likely intend this work as giving a short-overview of the potential of DCS, but it becomes an overly simplified paper that tries to do too much.  My main recommendation to the authors is to revise the manuscript to follow a simpler objective(s), so that an expanded explanation and discussion can be added (for the more focused objective).  Along these lines:

1.    There should be a clear objectives statement in the introduction.   There is such a statement deep in the manuscript at line 168: “The main goal of this study is to determine if the DCS can detect small CH₄ concentration enhancements downwind from the area of interest, equivalent to those caused by a typical herd of cattle grazing on an extensive pasture.”  This is a reasonable objective, but it does not require using the DCS measurements to estimate emissions (using the IDM technique).  Other verification trials have shown that IDM can accurately give emissions when provided an accurate concentration measurement.  So the authors could drop the emission calculations to simplify the paper.  

2.    Despite the above comment, the paper is potentially more interesting when the DCS measurements are used to estimate emissions (with IDM). I would be OK making this task the main objective.  But the description of the tracer release verification needs improvement (comments below).

3.    In terms of dropping material … what is the value of the dispersion model calculations in the section “Sensitivity and precision required for grazing measurements”?  If it is to estimate concentration enhancement levels, can these instead be determined from the tracer release results (e.g., Fig 5, 7, 8)?  I prefer to see real-world measurements used for this task.  Another reason to drop the modelling work is that it is inadequately described (comments below).

Specific Comments

4.    Ln 57: “In typical IDM applications, open-path Fourier Transform Infrared (FTIR) sensors are setup upwind and downwind for the source of interest”. I would not refer to use of an FTIR as the “typical” IDM application -- line-averaging lasers (TDLAS) have been more commonly used.  I suggest substituting “line-averaging” for “FTIR” here.  The association of IDM with “FTIR” also wrongly suggests the McGinn et al. (2011) study used an FTIR (they used a laser).  

5.    Ln 63: “The limited path distance, bulky apparatus, multi-component retroreflectors (Bai et al., 2022) make employing open-path FTIR challenging in agricultural environments.”  Assuming the authors substitute “line-averaging” for “FTIR” as suggested above, they will have to either remove this list of FTIR disadvantages, or introduce the FTIR as a preferred (?) instrument in order to justify including this list.

6.    Ln 112: “A forward Lagrangian stochastic model (Windtrax, Thunderbeach Sci.) was used to simulate the concentration field downwind from a hypothetical herd of 20 head of beef cattle grazing in an area of 25 ha …”.  Give more details about the simulation: How were the point sources spatially distributed (randomly, evenly, moving)?  Where were the laser lines located relative to the sources (e.g., what does 45 m away from the herd mean)?  What was the height of the release?  How many model particles were used?  Clarify these details. 

7.    Ln 114: “Turbulence data for the simulations were measured … at a grazing unit adjacent to the measurement site. The wind dataset selected …”  Explain which wind parameters were used in the simulations.  Was the wind direction allowed to vary? 

8.    Ln 125: “The forward model predicted that a herd of 20 cattle grazing in an area of 25 ha would produce a CH₄ enhancement of 16 nmol mol^-1 above a 2000 nmol mol^-1 background for a beamline 45 m away from the …”.  The text implies the concentration enhancements were calculated from 30 days of wind measurements (line 115).  Over time the enhancement will change as the wind conditions vary.  So what does the single enhancement value of 16 nmol mol^-1 represent (average value, median)?  What do the enhancement values in Table 1 represent?  What about the variability in the enhancement?

9.    Ln 162: This paragraph (or a variation of it) would be good in the introduction.  It gives important background information and contains a clear objectives statement: “The main goal of this study is to determine if the DCS can detect small CH₄ concentration enhancements downwind from the area of interest, equivalent to those caused by a typical herd of cattle grazing on an extensive pasture.”

10.    Ln 201: “The CH₄ tank was weighted in the beginning and end of the gas release campaigns and the mass of gas released was determined gravimetrically …”  For what purpose?  To verify the release rate given by the mass flow controller?  If so, did this confirm an accurate release rate?  Clarify. 

11.    Ln 220: “The WindTrax input data consisted of … and appropriate wind statistics.” Describe the wind statistics used.

12.    Ln 225: “The source area (Fig. 4) used by WindTrax to infer fluxes was set to match the 12.5 m² area of the CH₄ point sources.”  Was the area source at ground level?  

13.    Ln 233: “… where F is the flux, σ_F is flux error, σ²_𝑟𝑑 is downwind (background) mixing ratio error, σ²_𝑟𝑢 is upwind mixing ratio error, cov(d,u) is the covariance of the downwind and upwind errors.” Are the σ² values variances (of what variable)?  Reference a good description of Eq. (2).  

14.    Ln 241: “A CH₄ release at a rate of 3078 g day^-1 is shown in Fig. 5.”  It would be good to remind the reader of the significance of this release rate (e.g., it corresponds to X number of cattle)?  

15.    Ln 244: “This measurement demonstrates that the DCS system can detect small CH₄ enhancements equivalent to the ones caused by a small herd of cattle located at approximately 50 m from the downwind laser beamline.”  A) Can this conclusion be justified statistically?  B) From Fig. 5 it appears there is no difference between r_d and r_u for some periods. The magnitude of (r_d-r_u) will depend on windspeed (i.e., u*), and plotting (r_d-r_u) vs u* would presumably show the DCS system is not detecting downwind enhancement for higher winds (when there is more dilution of the tracer).  This is not a unique problem for DCS as any sensor would show similar trends with u*.

16.    Ln 246: “The two-DCS measurement path geometry is also capable of capturing and rejecting the temporal dynamics of the CH₄ background driven by changes in atmospheric boundary layer conditions.”  What does “rejecting the temporal dynamics” mean?

17.    Ln 252. “To determine if any mixing ratio biases exist between the North and South beamlines that may lead to incorrect flux values… The average CH₄ flux computed using WindTrax was ± 974 g day^-1 equivalent to approximately 5 head of cattle …”.  This paragraph is confusing and needs more explanation.  A) Remind the reader of what you are doing, e.g., using path concentrations during a period with no gas release in order to determine precision of the DCS+IDM calculation?  B) What is the difference between the given uncertainty in the average CH₄ flux (+/- 217) and uncertainty using WindTrax (+/- 974)?  

18.    Ln 258. Figure 6 caption (and Fig. 7, 8).  Explain what the error-bars represent … flux uncertainties due to DCS measurement uncertainty (and this does not include IDM uncertainty).

19.    Ln 266: “WindTrax computed average CH₄ flux was (4002 ± 1498) g day^-1, showing a good agreement to the actual release CH₄ flux of 3970 g day^-1.”  A) What does the ± value represent here?  B) The emission “recovery” fraction (4002/3970 = 1.008) is phenomenally good.  This is worth some commentary and context given other IDM tracer release studies (e.g., a compilation of verification studies is given in the appendix of Harper et al., 2010: The effect of biofuel production on swine farm methane and ammonia emissions.  J. Environ. Quality).

20.    Ln 290: “The agreement between the computed and actual CH₄ fluxes in this study shows that DCS can precisely measure the small concentration enhancements due to a herd of beef cattle in the field at distances up to 100 m from the source area.”  A) It would be good to add the caveat about the effect of windspeed on detectability (see comment 15 above).  B) I would like the authors to comment on the base level uncertainty (~ 5 cows), e.g., what does this imply about the minimum number of cattle that could be measured?  

21.    Ln 302: “For example, prior calibrations are often necessary when using multiple FTIR systems to perform multi-path gas concentration measurements.”  What is meant by “prior calibrations”?  Cross-calibration of different instruments?  This is not just an FTIR problem, but a problem with all types of concentration sensors.

22.    Line 305: “Expected CH₄ horizontal gradients in grazing systems are often small, as demonstrated in this study, so small instrument biases can lead to large errors when inferring fluxes.”  Good point.

23.    Ln 310: “The driving rationale of this work is to quantify the net CH₄ fluxes produced by cattle grazing system …”  Are the authors suggesting a role for the DCS system in measuring soil fluxes (by either micrometeorological approaches or chambers)?  Is this realistic given the generally small magnitude of the soil fluxes or the logistics of pairing DCS with a chamber.

The manuscript describes the field deployment of an dual comb spectroscopy (DCS) system along two simultaneously operated open measurement paths with the goal to measure CH4 emissions from cattle. This capability is tested and demonstrated in a controlled release experiment simulating cattle emissions while measuring along approximately 200m long paths upwind and downwind of the release area. The paper describes the experimental setup, hardware, the data processing, and spectral analysis which in the end produce path averaged mixing ratios. It further describes the analysis to infer flux estimations from these mixing ratios and their differences. This work provides a contribution in the ongoing challenge of measuring methane emissions from ruminants and on the open question of how open-path measurements of greenhouse gases (GHGs) are best employed in practice. I recommend publishing this work after addressing some minor and some potential major comments below.

General remarks:

I think you do not clearly and transparently present the extend of your measurement campaign (i.e. line 153 ff.). For what timespan was the DCS setup deployed? How many release experiments did you perform and on which days? The data you present in Figures 5 to 8 spans at least 3 months, yet you do not show for example the emission estimates for the day in Figure 5 which you used to show the enhancements during the release. Did you operate the open-path system during the mentioned grazing period (May to mit July) and try to measure real cattle emissions? If so, what were the additional challenges compared to your controlled release experiment? Since your stated goal is demonstrating the capability of such a system to monitor emissions, I would encourage you to state transparently how often you had high quality results. In its current state, the manuscript generates the impression that the extend of the dataset is intentionally vague and potentially data presented very selectively.

Detailed remarks:

Line 23 f.: I do not see how the provided materials show that only optical power limits the measurement. The controlled release experiment had quite accurate knowledge of the release area and the manuscripts does not provide a systematic analysis of the impact of source distribution uncertainties and transport uncertainties, which typically contribute significantly to the uncertainties of fluxes estimated from concentration measurements.



Line 61 f.: To my knowledge, McGinn et al. (2011) did not use an FTIR system.



Line 71: I assume with "square-law photodetector" you mean a photodiode operated in a linear (power to current) regime. If so, calling it that way might make this more accessible to a wider audience. If not, I do not understand the point you are making here.



Line 82 f.: I appreciate you citing LMFIT but think, if you do it, the doi should be included in some form: https://dx.doi.org/10.5281/zenodo.11813



Line 83: You did not cite the most recent version of HITRAN (HITRAN 2020, Gordon et al. 2022). If you did not use the most recent version, you might want to mention which version you used and your reasoning behind that. You also might want to mention which information (i.e. line shape model) you used.



Line 108 f.: I think the concept of "molecular time" would be worth a one line explanation somewhere in the manuscript if you need it.



Line 114: What does "Turbulence data" include? Which parameters were measured and are available for analysis?



Line 138 f.: Your estimation of your measurement sensitivity of 5 nmol/mol is really interesting and I think it would be worth a bit more thorough explanation. I for my part found it challenging to follow you and am still not certain if I understood it correctly.



Line 190: It seems like you did not show or state your exact path lengths anywhere and not in Figure 4.



Line 208: How did you calculate your mole fractions? What did you use to estimate your total airmass?



Line 249: While this is not wrong in any way, I feel that using the combination of arrows and the convention to give the wind direction by where the wind is blowing from generates an unnecessary potential for misunderstanding.



Line 254 f.: I do not understand the difference between your estimate of 1 +- 217 g/day using the IDM and your WindTrax estimation of +-974 g/day. I thought, that WindTrax is your IDM (Line 220).



Figure 8 b-d: I would again encourage you to show a bit more context around the presented data. If you have the data from 13:00 to 14:00 available, your plots here would provide a better impression on how well the signal can be distinguished from the background.



Line 291 ff.: As in Line 23 f. I do not see where you presented data on the transmitted power over time and how this is limiting your performance.



Line 298 ff.: Again, I do not see where you discussed the technical aspects you list in point 1).



Line 335: You might want to mention how contributed to drafting/writing up the manuscript.