This study describes the results from MethaneAir airborne campaigns across multiple basins in the U.S., with emphasis purely on attributed point sources above the MethaneAir detection limit. The study clearly presents the algorithms, the attribution, the results, and performs inter comparisons with other studies. I have a few comments that need to be addressed before I can recommend for publication. 

1. Although this study represents (to the authors' knowledge) the largest airborne methane survey performed over onshore oil&gas in the U.S., it wasn't entirely clear what new learning was gained from such effort. 

- (1a) One possibility is in their section on repeat sampling in Haynesville, where they show that point source frequency is possibly stable. 

- (1b) Another possibility is that non-oil&gas point sources make up an equal fraction of emissions as oil&gas point sources. That is very interesting, but the authors need to address detection limit issues if that's the central thrust of the paper. For example, certainly there are a lot of oil&gas point sources not detected by MethaneAir that were there (hence the need to filter results for comparison with Cusworth et al., 2022). For other sectors, if the general distribution of emissions is higher than the MethaneAir detection limit (e.g., underground coal vents), then those would be readily detected while some oil&gas sources wouldn't be detected, hence over-inflating the non-OG contribution to point sources. 

In any event, a presentation of this dataset itself is interesting, but some clarity on the general scientific questions this study is answering would be helpful to contextualize the results.

2. Line 85. A more recent MethaneAir paper cites 33-38 ppb for MethaneAir precision in the Permian Basin and Arizona. Why is there a discrepancy? Is 17-20 ppb consistent with what you calculated in this study? It would be nice to know the background standard deviation in each of the regions you surveyed as it contextualized the distributions of detections in this manuscript.

3. Lines 108-114. The DI and clumping technique you describe for plume detection uses many hard-coded values (e.g., 600-m pixel squares, 12 pixel clumps, etc.). Is this approach exactly the same as what was summarized in El Abbadi et al., 2024? In other words, how sensitive is the algorithm to these assumptions? 

4. Related to comment #2 - by selecting 600m windows, you are making a prior assumption on plume size which then could bias your detection algorithm. Put another way, if you assumed 100-m windows (i.e., smaller plume sizes), would this result in more plume detections? Have you tested this?

5. Line 261-264. This is conjecture. I don't think that conclusion is defensible without citation or actual operational data.

6. Line 353 and elsewhere. You cite the detection limit for MethaneAir to be ~200 kg/h repeatedly throughout the text, but then restrict most inter comparison to emissions above 550 kg/h. Other studies name this specifically with a probability of detection or an effective detection limit. You've shown in this study that effectively the detection limit for methane air is 550 kg/h, so it would be more accurate to update every mention of ~200 kg/h to "200-550 kg/h." 

This manuscript describes the results from a large MethaneAIR campaign, an aerial methane measurement method, across several U.S. onshore production basins. The results from this paper focus on the distribution of high-emitting methane point sources across different types of facilities, basin-level trends in point source behavior, and understanding the relative contributions of methane point sources across different sectors. I believe this manuscript provides data that will advance our understanding of methane emissions and distributions, but would benefit from additional analysis. It is not immediately clear what the novel contribution of this work is, and additional analyses could be conducted to make this manuscript more compelling. Some suggestions for additional investigation would be:

- Further explore the contribution of emissions from other (non oil and gas) sectors. Are there other previous studies from these sectors that could be included for comparison?

-Further exploration of the ultra-emitters (Line 213)

-For regions with multiple overpasses, how do they day-to-day statistics compare? Earlier in the manuscript (Line 137), it is mentioned that with large enough sampling you could assume representative sampling across space and time (ergodic hypothesis). Do you find this to be true in regions where there were multiple overpasses? Can anything be learned about minimum requirements for sampling, and the equivalency between spatial and temporal sampling?

In addition to the issue of scientific novelty, I also have specific comments related to the technical aspects of the work that I suggest be addressed before publication:

- The detection threshold is defined as 550 kg/hr, but data was included in the analyses and figures for emissions of ~200 kg/hr. Emissions below 150kg/hr were discarded. Please elaborate on this treatment of the data including the justification for excluding emissions below 150 kg/hr, while including emissions below 550 kg/hr. Is there detection testing that supports this? I would also consider the probability of detection as well, rather than assuming a binary detect/non-detect. What is the probability of detecting a 550kg/hr source vs a 200 kg/hr source? If a site that was previously detected at 200kg/hr is not detected again, how can you discern whether this is because of a low probability of detection or because the source stopped emitting?

- Line 260: Related to above comment: Could the intermittency of detection be due also to probability of detection? Is it fair to conclude that just because the sources were not detected, they were not there? The manuscript states that "Recurrence of site-level emissions was seen only from non-oil and gas facilities” - was there 0 recurrence at oil and gas facilities?

- Line 47:  "Methane point sources above 10kg/hr can make up 13-59% of total regional flux" Please clarify the meaning of this statement.  Are these point sources from all sources or just oil and gas sources? Is this just for oil and gas producing regions? What is the remaining 87-41% - is it from diffuse sources or source below 10kg/hr?

- Line 133: Please explain why the emission rate was divided by the number of overpasses. It might be more helpful to report the average emission rate for detected emissions and the number of non-detects separately. The number of "0" emission measurements, especially when there were previous emissions there, can be just as important as the measured emission rates. This also relates to the first comment of understanding whether non-detects/"0" measurements are a function of probability of detection or source intermittency. 

- Table 1: Does the % oil and gas relate to the count of detections or the emissions contributions

- Line 208: Is there a reason why the DJ basin has the highest frac of non-oil and gas in Table 1, but the lowest emission rates? 

- Figures 2 and 3: Blue/green colors are hard to distinguish from one another 

- Figure 4: It would be helpful to include n values here – for example, my understanding is that the emissions from the compression station facilities in Appalachian north came from 1 detection (n=1). I think the sample size is important to communicate here.

-Line 255: How many measurements of pipelines were taken? Are there less large point sources because there was less observation, or are they just less prevalent in pipelines?

- I would include more information on the comparison of these measurements with other measurement campaigns. I think this needs to be explored more, and to comment on whether the MethaneAIR distributions of high emission point sources agree with other data sources. I also think the information provided on Page 14 could be organized into a table or figure for more effective comparison.