The paper provides a useful comparison of methane emission rate methods for rates lower than what has been studied frequently in previous publications. They conduct controlled release testing of four or five different methods. The paper is nicely written and provides interesting data. However, the authors should do more work on providing context for this work and when/where it’s applicable. Also, there are edits that are needed throughout the paper.

One concern with this paper is that it’s unclear who this work is applicable to. For example, for greenhouse gas emission inventories, there is always the emission factor approach, in contrast to what’s shown in Figure 4. The authors mention oil and gas wells but without providing much context into the type of oil and gas wells, which can explain how the flow rates studied were chosen.

Also, the authors caution against doing measurements at sites with hydrogen sulfide and aromatic hydrocarbons, which would bias samples for regional or national inventories. From a policy perspective, wells that emit hydrogen sulfide or aromatic hydrocarbons are often prioritized for mitigation, and it would be unfortunate if we can’t quantify the methane emissions being reduced through these efforts.  

The methods can give very different uncertainties depending on how the experiment is conducted. For example, there are many ways to implement the Gaussian Plume method, including how and where methane is analyzed and the micrometeorology measured. The same goes for the static and dynamic chambers. Overall, the authors need to add more detail on the methodology and experimental conditions (dates/times, exact equipment and supplies, etc.).

Below are some additional detailed comments:

line 2: what is considered a "small point source"?

line 7: how is a point source defined? At what scale?

line 13: not clear if static chambers are tested in this study.

line 16: why only 200 g/h? confusing because in the text, there are three flow rates mentioned.

line 40: "very small" sounds arbitrary. where does the 0.6 mg/hr come from? Kang et al. (2014) also measured negative emission rates. There also are a wide range of published measurements using this approach for much smaller fluxes found in natural environments.

line 41: replace "place" with "placed"

line 43: The static chamber method does not require a gas chromatograph. In El Hachem and Kang (2022) published in Science of the Total Environment, they do not use a gas chromatograph.

line 50: what is the source of this air? Is it background air?

line 53: what is the background methane concentrations in the air? And what air is the authors referring to?

line 56: how much power is required? What type of power source is needed?

line 62: what is the current commercial HiFlow sampler? I see in the next lines that you mention the Bacharach. But I've heard that it's been discontinued. Are there others that are currently commercially available? In the previous sentence, the authors write "typical rates are 300 l/min" but that implies there are multiple types of samplers. If there is just one, why not just report the on high flow rate?

line 87: the inputs to the bLS model appears to be the same as the GP model? What are the exact meteorology and micrometeorology parameters needed for the bLS model?

line 87: where is this gas concentration taken?

line 94: isn't complex topography and buildings also an issue for the GP model?

line 96: what kind of micrometeorology data is needed?

line 100: for higher emission rates, wouldn't it be easier to do downwind measurements such that site access is less of a concern?

line 100: what are the safety concerns here? just explosion risk due to high methane concentrations? What about H2S (See El Hachem and Kang, 2022)?

line 102: what is meant by "able to approach"? How close to the single point source in meters?

line 103: what are the emission rates considered? In the abstract, it was only for 200 g/h but there are three mentioned later. Need to be consistent throughout.

line 108: why the cut off at 200 g/h? There should be a paragraph on the literature for tests at >200 g/h and describe why those studies are not applicable here.

line 110: the exact dates and times in which each experiment took place needs to be provided.

line 111: it's unclear which exact experiments were done. it would be helpful if the authors could provide a spreadsheet with all the tests that were conducted in the supporting information.

line 115: how were the emission rates set? What is the type of flow controller? What are the gases that are used?

line 115-116: how do you define/determine what is safe or not?

line 116: why not lower than 40 g/hr?

line 125: it's unclear if a static chamber measurement was done?

line 127: what is the location and size of the fan inside the chamber selected? and how was this selected?

line 128: What is meant by "three further air samples? Further to what?

line 132: what where the shapes of the chambers? what is the aspect ratio (height to diameter)?

line 133: how was the quality of the ground seal determined?

line 135: is this experiment a copy of Kang et al (2014), Pihalatie et al (2013), or Collier et al (2014)? Which one took four samples? Kang et al (2014) took 7 to 8 samples.

line 135: there is a "s" missing. it should be "four samples".

line 137: were there duplicates and blanks taken?

line 142: does this imply that emission rates were calculated for test even if only three samples were collected?

line 146: why was this chamber size selected? what is the shape and aspect ratio? How useful is this size for field measurements of oil and gas wells?

line 146: what type of plastic is used?

line 149: what is the detection limit of the HXG-2D? What are the methane concentrations observed inside the chamber?

line 150-151: blanks and duplicates taken?

line 155: how big is the hose end? Was the source enclosed by the Hi-Flow sampler?

line 167: any concerns with topography and large objects (e.g., buildings, trees, and other infrastructure)?

line 177: what is the minimum distance between the sources and the detector?

line 181: how different were the environments in which the experiments were conducted? Importantly, were experiments described in sections 2.1 to 2.5 all conducted on the same day. If they were all conducted on different days, then the uncertainties calculated cannot be directly compared.

line 190-191: The collection of gas vials is not a requirement of the static chamber methodology.

line 193: the static chamber can be used with a methane analyzer (e.g., El Hachem and Kang, 2022), overcoming the first and second shortcoming.

line 196-198: El Hachem and Kang (2022) conducted measurements from H2S-emitting wells using a self-contained breathing apparatus. There are many options available in industry to ensure safe working conditions when toxic gases are present.

line 198: what about for measuring low emitting sources?

line 200: why isn't "cost" italicized like the rest?

Table 1: I'm surprised that the HiFlow sampler is only $5k. Is this correct? The static and dynamic chamber measurements conducted here use a GC, which is around $50k. So it's definitely not free. Even just getting the gas concentrations analyzed elsewhere is not free.

Table 1: why is there no time for measurement and analysis for the static chamber? same for the accuracy.

line 209: there are other ways to reduce the potential of CH4 concentrations in the chamber reaching explosive levels when using static chambers.

line 210: the need for a power source is another important shortcoming of the dynamic chamber method.

line 266-267: what is the dynamic chamber the HiFLow more accurate than?

line 273: who is this decision-making paradigm for?

line 282: what are the conditions in this study? This needs to be better described to assess the applicability of the results.

Figure 4. Many estimates (e.g., the USEPA's GHGI) involve wells with no measurements. It's not possible for all wells to be measured. So if there is no trace gas analyzer, there is always the emission factor approach. But of course, that's not a good predictor of the emissions at a given well but over some large population, it may be representative. So this brings us back to the question of who this figure is for. This figure needs more context in the caption and the text.

General comments

The research presented in this paper reproduced and implemented in a controlled release experiment a portfolio of techniques dedicated to quantifying methane emissions from a point source. Through the source typology and chosen emission rate, the emphasis is on potential applications to fugitive ‘small’ (<=200 g/hr) emissions in the oil and gas sector. The research encompasses well established and already published methods including flux chambers and atmospheric downwind measurements.

The paper discusses the respective merits of each method, implemented from a very practical perspective, including an assessment of precision and accuracy.

The novelty of the paper lies in the joint implementation of all these methods in a single experiment focusing on low emission rates. The methods are already published elsewhere.

The paper presents several weaknesses in my opinion:

• The state of the art in the introduction doesn't acknowledge other available techniques. The selection of methods reproduced in the study is not explicit, and the reasons for ignoring/discarding other techniques is not clarified.
• There is a lack of context elaborating on the specific needs of the industry (e.g. buried sources are ignored or implicitly included as the paper seem to focus on aerial point sources), and possibly introducing some sort of statistical distribution of leak size would be useful.
• Finding that accuracy calculated from 3 estimates (Ar) improves compared to the discrepancy of a first, single estimate (A) is trivial but nevertheless occupies a significant part of the results (sect 3.1) and discussion: Sect. 4.1 is dedicated to this, even discussing a particular occurrence where 3-point accuracy is lower than the single point discrepancy. In general, if Ar is available, there is little statistical sense to discuss A at all.
• The “decision making paradigm” (Sect 4.2) is limited in scope by ignoring other techniques and situations that may representative of the industry, and it seems to operates in its own limited rationality, letting the reader ignore other works.

Specific comments

L16: not only for 200 g/h.

L29: why is this 200g/h threshold important? Is there a scientific rationale? Is it specifically representative of situations or technical challenges in the industry?

L31-38: A number of techniques and approaches (FLIR, mass balance, tracer release, remote sensing…) are ignored in this study. Their existence and their absence here should be acknowledged and thoroughly commented. They may not be included in this study for some (presumably good) reasons?

L37: “despite the interest in developing methods”: unclear

L78: what is “total reflection of CH4 at the surface”?

L79: I would argue that mass balance is the simplest method, rather than GP.

L81: perfect “gaussian” plumes are indeed seldom met in nature. But also it is rare to have lonely ‘weak’ plumes in an industrial setting, so the GP approach needs to account somehow for multiple sources.

L98: is it the same 38% uncertainty applicable for 4ug/h and 3kg/h?

L102: if the 200g/h limit is for safety/practical reason, is it still useful in real life?

L108: can you please then comment on what was done at leak rates higher than 200 g/h? What are the limitations to transfer these conclusions to smaller leak rates? Why should we care about leak rates below 200g/h?

L110: it seems that the section 2 repeat the method description from the introduction, or at least lists again each method. The text would be more fluid and logical if all method description (including Eqs 1-5) are into the Method section and the introduction then focuses more strongly on state of the art, context and research questions.

L117: what are the uncertainties associated to the release rate (for example from the mass flow controllers)? How long does the releases last? What is the shape of the injection exhaust/outlet? Is there any attempt to reproduce a ‘diffusive’ exhaust? Or to control the gas exhaust velocity?

L122: how long is the calibration? How precise is the gas standard?

L127: how well do we expect the air inside the chamber to be mixed with the fan?

L128 how is the air sample drawn?

L134: why zero wind condition and not “wind speed below X m/s”? can you elaborate on this serious limitation?

L165: at what height is the gas scooter measurement made? Was any attempt made to measure CH4 across the plume (cross wind) to confirm the gaussian shape of the plume? If not, why?

L178: What is the impact of very small distance (5m) on the accuracy – does the model have a lower limit? Also measurements at 5m downwind suggest that relatively close access (and at the same height AGL) is possible, this should be acknowledged (also in Table 1).

L195 and following: why not automate chamber opening to avoid explosive limit? This should be easily done in a commercial context, and acknowledged in the discussion.

L227: the accuracy of the single “snapshot” becomes irrelevant once 3 repeat measurements are available for the release. Accuracy and precision derived from 3 points supersede the single measurement discrepancy as informative numbers.

L232: abs(A) “decreases”: this is not confirmed in Fig 2b. can you please explain where this comes from?

L242: please quantify “generally”

L245: Ar>A : to me this is meaningless. It just means that there was some luck in the first value, and therefore it carries little sense to report A once you have Ar.

L251: This is expected and may be seen as trivial. On the opposite, what is surprising is when it is not the case. Why does SD for some techniques increase with increasing emission rate?

Fig.3 in my opinion would deserve further comments and discussion.

L265: Which technique has the lowest A is relatively unimportant when Ar is available. Ar>A is also fairly trivial from a very general perspective.

L280-283: These sentences are vague, complicated. There seems to be a confusion between the site scale uncertainty and the single GP uncertainty. Could you please clarify in terms of separating biases and random errors? (and all methods here seem to have consistent biases). Did you make any attempt to look at uncertainty budget in the different methods, the GP in particular, including the choice of stability class? Does it match with the 3-point accuracy?

L293-294: Can you please explain the “meaning and balance” and provide examples of studies that “present unexpected findings”?

Fig. 4: other parameters/selectors would be useful but are ignored in this study: is the source buried? What is the source intensity? Did you perform a source detection prior to quantification? Some techniques are missing; as such, the value of this diagram is very poor for decision making, although the idea is good.

Editorial comments

L21: GP: expand acronym

L103: “copy” I suggest “reproduce”

L177-178: the cited papers are not listed in the References section

Fig 3: what is y axis unit?