|The author’s made improvements to the article but many of the reviewer questions remain unanswered/unclear. This is a valuable dataset but the analysis and manuscript could still use improvement in my opinion. I completely understand that there will be future papers coming from this unique dataset, but it would be very helpful to the readers to know what is slated for future work to help put this analysis into better context.|
Line 99-102: Hecobian specifically compares the emissions of distinct operation phases of natural gas extraction. The analysis presented here would be greatly improved by doing a more direct comparison of the emissions at each stage of well production with Hecobian (e.g., for lines 276-281). Both datasets are incredibly rare and valuable, so it would be really helpful to see if they compare for gas fields in different regions of the U.S. and would further help to put Orak measurements into context.
The addition of the trailer location to Figure 1 is very helpful. This combined with Figure 2 highlights the fact that the wind was rarely from the SE sector where the majority of the drilling equipment, and presumably the drilling activity, was occurring. It would be really helpful to show what the air composition was as a function of wind direction. Was methane, ethane, etc. higher when wind was coming from the SE sector? Does Factor 1 of the PMF results have the highest contribution to the ambient measurements when wind is from this direction? This was one of the unaddressed questions (Reviewer 1, Q1). You should have a timeseries of each factor of your final PMF model that can then be analyzed as a function of any other variable that was NOT included in the PMF such as wind direction. If the Natural Gas factor is most prevalent in the SE sector, then it further adds confidence in the analysis.
Figure 3 is the key figure in this manuscript, in my opinion. This clearly shows the different chemical composition of the air measured during each of the important drilling/hydraulic fracturing/production steps. Why doesn’t PMF pick up these differences and lump all into a single generic factor? How do these results compare to other studies?
Lines 288-292: Be sure to specify the units as you are comparing the ethane (ppb) to methane (ppm) ratioor else add 10^-3 to the ratios. How does the ethane to methane ratio compare to other ONG emissions (i.e., Yakovitch et al. and may other papers)?
Lines 328: Source Profiles. Another useful reference regarding PMF analysis in an oil and gas field to add to the discussion could be:
Source characterization of volatile organic compounds in the Colorado Northern Front Range Metropolitan Area during spring and summer 2015. A. Abeleira, I. B. Pollack, B. Sive, Y. Zhou, E. V. Fischer and D. K. Farmer. 122(6), 3595-3613, doi:https://doi.org/10.1002/2016JD026227, 2017
Figure 6: I’m still trying to figure out why CO2 and methane have virtually the same PMF factor fingerprints. Clearly, the natural gas factor isn’t just natural gas as raw natural gas does NOT contain NOx and I would not expect it to be composed primarily of CO2. Also, the engine emissions factor doesn’t contain an appreciable amount of CO2. Why? It seems to me that the PMF factors aren’t fully resolving in a meaningful way. Also, why do hexane and benzene not have any attribution to “regional transport” as these two species are sufficiently long-lived in the atmosphere to have a significant background, much more so than toluene that has ~25% attributed to transport?
The last, “big picture”, piece of this analysis that is missing is how the natural gas drilling activities actually affected air quality, which is stated as being the motivation for this paper. Since you have the PMF factors for each species, then you should be able to answer the question of how the air quality would be different if the “natural gas” factor was removed or how it compared to the other factors by calculating an Air Quality Index or OH-reactivity, or some other metric.