General Comments: 

The paper by Herman and Mao is a study comparing Total Column HCHO, NO2, and O3 from Pandora Spectrometers to OMI and DSCOVER-EPIC. They included multiple pandora stations located at various locations around the globe and during different seasons. They found that agreement is overall good, however OMI does not alawys capture the seasonal variation as seen in the pandoras and may not be sensitive to changes in surface concentrations. DSCOVER-EPIC agrees quite well with the diurnal pandora data. This is a much needed comparison study as there are few publications on the validaty of pandora spectrometers which are to be used in future satellite validation plans. The manuscript requires some minor changes as well as some additional discussion/figures before publication. 

Specific Comments: 

79-- Introduction mentions airborne data but does not include any in results. I would be interested to see the comparisons. Otherwise remove from introduction. 

83-- Why not use TROPOMI in this study instead of (or alongside) OMI? It is mentioned but not used. 

120-- The text files of Pandora do not need to be explained in such detail. However, I would like to know what data quality flags are being used to filter out bad quality data. 

Fig 01: 

-- Why not show a continuous time sereis for the July and September weeks? 

-- Change y axis for the NO2 day comparisons to be equal. 

-- Discuss why the Lowess line is important. 

Line 138-- No need to list the file name just state which station is being discussed. 

Fig 02:

-- What is the "0.003" listed on the figure? 

-- Fig 2 and Fig 1 both state that there is a seasonal dependence of HCHO but not NO2. Fig 1 is not necessary unless the daily panels are further discussed. 

Line 145-154-- What is the reasoning behind showing some pandora figures and not others? Why not include a a monthly average time series of all pandoras on one figure or at least group by certain locations. This would also help see the difference in magnitude of TCHCHO and TCNO2.

Line 154-- Please include a table of all Pandora stations included in this study. The wording is vague about which pandora stations in CT are included in this statement. There are several stations along the CT coastline. This will also prevent the lat/lon and PI from needing to be stated in every figure.  

Line 178-- Why is there a seasonal NO2 pattern if, like NYC, the pandora is near automobile sources? 

Line 191-- How were these days chosen? Is the OMI agreement dependant on the diurnal pattern of HCHO? 

Fig7:

--How were the pandoras used for the OMI comparison chosen out of a total of 147?

--Where is the monthly average TCHCHO comparison figure? 

Line 203-- Reword. It isn't that OMI and Pandora TCNO2 agree more at the overpass time, it is that the overpass time is the only availabe data for comparison. 

Line 205-- What method are you using to compare OMI and Pandora. Is it a single pixel that overlaps the pandora? A given radius in km? 

Line 220-225-- If the HCHO comparison results are due to the ozone retrieval influences then what is the TCO3 at these dates? Is NO2 better because it is not impacted by ozone spectral fingerprint? 

Line 220-- I would be interested in seeing a scatter plot comparing 13-14:00 UTC Pandora total column with OMI for all days of these Pandora stations. That way we can see if there is a constant bias and by how much. Otherwise explain why these days were chosen out of three years of data. 

Figure 10a and b should be separate figure numbers. 

Separate figure numbers for figure 13a and b

Fig 13-- why are these days and pandora sites chosen? Are others worse?  

Line 267--  No figure showing the OMI seasonal variation in TCHCHO.  

Line 274-279-- This paragraph needs reworked. I can't tell if you are trying to say if OMI and Pandora agree on total column amounts or not. Line 276 says agreement is only good between the hours of 13-14 UTC, but what other time period would you be comparing to OMI? 

Line 282-- Authors only show data for 6 pandora stations in comparison with DSCOVR-EPIC. All in Eastern US and Canada and in August. Yet this statement suggests that all pandoras have good agreement. I would like to see a figure with all pandoras (or grouped by either time of year or location) before I accept that pandoras as a whole agree with DSCOVR-EPIC. 

Technical Corrections: 

Line 148-- typo

Line 239-- typo on figure number

Summary: This manuscript compares OMI columns to Pandora columns for formaldehyde, nitrogen dioxide, and ozone. Section 2.0 references 145 Pandoras, but figures show data fewer. The authors highlight that TCNO2 has a diurnal profile that requires time-pairing when comparing Pandora to satellite. They also highlight that OMI THCHO lacks seasonality that is expected and seen in Pandora.

Response overview:

This manuscript is not ready for publication. The goal appears to be evaluation the OMI dataset using Pandora and EPIC. It needs to be better organized, with more methodological details, and improved quantification. There are a combination of inconsistent methods (Loess vs monthly avg), statements that seem to imply methods that are neither discussed nor have results reported. One of the current conclusions seem like they would be removed if the methods had been more appropriate. With additional methods and quantification, this will be a nice contribution to assessing the satellite assets currently monitoring our atmosphere.

- The introduction does not currently describe the motivation for the study and needs  reorganization. The introduction discusses sources of pollutants and closes with the idea that it will compare OMI and Pandora. It does not describe why a comparison of OMI and Pandora would be useful. It appears to be a validation paper, which is good but is not clear. If this is a validation paper, the paper should include descriptive statistics (and more stations).

- The manuscript needs a methods section.

  - The authors need to describe the Pandora data selection and filtering.

    - The authors describe example file names, but do not discuss their meaning. As a Pandora user, I am aware that rnvs means direct sun NO2 and rfus means sky-scan HCHO, but the average reader may not.

    - This raises the question why are you using sky-scan measurements for HCHO? And, how were stations selected as having sufficiently high quality sky-scan HCHO for comparison?

    - Which stations are used and where are they? Section 2.0 references 145, the acknowledgements references 63, and I did not count the number referenced in the figures. Given that conclusion statements like "most sites" are made, it should be clear which sites were part of the analysis.

    - How OMI and Pandora were paired. Are OMI pixels within a certain distance used? Or only when the Pandora site is within the pixel geometry defined by the corners?

  - The OMI product is insufficiently described in the document. There are no version numbers or citations. The authors have provided the URL of where to get OMI, but nothing about the data product or when they acquired the data. Websites change and the Aura website will likely be lost after Aura is decomissioned. More details about the data product need to be included in the publication for posterity. According to https://ozoneaq.gsfc.nasa.gov/products/ozone/, "Overpass (OVP) products are a weighted average of data within a defined range to a set of ground station locations." What the distance of the defined range? There is a 50km version that is very clear about its distance, but the README for the standard OVP products is not clear.

  - If you're going to use a Loess fit, it would be good to introduce it somewhere. The Loess (not Lowess right?) fit seems unnecessary given that all your other plots use simple running means. Perhaps you could explain why it is appropriate for Pan 180 NO2 in Figure 1, but not Figure 7.

- Figures are often scatter plots where the markers are so dense that often only a cluster is visible. The distribution of values is not decipherable. I recommend creating some sort of synthesis plots.

- The longer time-series would benefit from some sort of statistical analysis that quantified "agreement" and "disagreement" in the abstract. Right now, there is little more than visual analysis of datasets that were processed by others and downloaded.

- Similarly, it would be nice to quantify seasonality. XX% higher in JJA than DJF or similar. The conclusions starts with a paragraph about seasonality, but right now the manuscript simply says it is seen in one dataset and not the other.

- The idea that Pandora would agree best if paired in time seems like an obvious conclusion. Figure 7 and analysis could be simplified by highlighting (or citing) the diurnal variation in the methods sections as the reason for time-pairing.

- The conclusion that OMI "underestimates" the degree of atmospheric pollution does not seem novel or quite accurate. OMI only "underestimates" pollution if we assume that overpass (13:30LST) is representative of the whole day. We know that vehicular emissions clearly peak at rush hour, so we would expect columns not to peak at Aura overpass (13:30LST). There is much evidence of this understanding in the literature. For example Anenberg et al. (doi:10.1016/S2542-5196(21)00255-2) use a series of ratios to translate overpass-time data to daily averages (see Figure S1 and discussion).  The submitted manuscript should cite existing works highlighting the coincidence of the local minimum as a need for temporal co-sampling rather than highlighting this as a finding.

- The PGN website requests that "The PGN is a bilateral project supported with funding from NASA and ESA." be added to the acknowledgements. https://www.pandonia-global-network.org/home/documents/pgn-data-use-guidelines/

Line-by-line notes:

 - 35: the abstract discusses seasonal dependences, but isn't clear one what would be "big" or "little".

 - 36-39: the abstract and conclusions assert that OMI is not observing near the surface, but the authors only show that it fails to capture seasonality. Could the failure have to do with reference sector correction? Or some other failure? Could you explain why you specifically think it is a failure to sense the lowest levels.

 - 39-41: It seems obvious that excluding rush-hour from the comparison with 13:30 would be good. Why is this a noteworthy finding? (see discussion above)

 - 43-44: Agreement and disagreement should be put into some sort of context. Is there a pattern (under-estimating high values, over-estimating at low latitude) or is it random? What does agreement mean (bias within X? correlation above Y?)?

 - 45: Does EPIC provide any particular meaningful result?

- 54-55: Is the point that most methane that later forms HCHO comes from these sources? Or is this arguing that the majority of HCHO comes from this specific pathway (more so than isoprene + other methane sources)?

- 70: Are these citations only for the first half of the sentence? If so, what are the citations for the rest and their ranking?

- 79-104: This discussion does not mention surface monitors that sample in situ air or airborne in-situ measurements. The apparent focus is column integrals, but the sentences that introduces it simply says "typically measured by." Given that surface monitors and in-situ air sampling are more common, I think this needs clarification.

- 83: Given the timing of this submission, it is worth noting the TEMPO and GEMS satellites if this is a list. If this is really the methods section, then I don't think you use several of the data sources in this list.

- 117: seems weird to note that the website is Austrian.

- 119-122: Rather than providing file names, perhaps it would be better to discuss the meaning of the codes. For example, my read of the names is that you're using direct sun for the NO2 and skyscan for the HCHO. However, you do not discuss that. You also make no mention of data filtering of any kind.

- 126-127: Did you use their other measurements?

- 129-136: Using a single week a representative of day-of-week distributions is not a good idea. Friday July 8 might have been an outlier with winds blowing from a specific source that was active but downwind on Thursday. Why is this specific week a good case study?

- 133: "summer seasonal dependence" should probably be "summer peak"?

- 197: To me, it looks like NO2 peaks in the DJF period in the lower left and lower right plots for OMI (green line). Perhaps the seasonality in OMI is larger than the seasonality in Pandora.

- 206-207: "not statistically different" -- this stands out to me because I do not see any statistics anywhere. Nor do I see a discussion of how differences will be tested for significance. t-test? Welch's? Mann-Whitney? To see a statement like this, I would expect to see some data characterizations (mean+-std) for both datasets and/or the differences.

- 211-212: Can you at least state which Pandora sites you looked at?

- 217: How did you measure the "cases of agreement"? For example, did you consider the uncertainty in either measurement?

- 226-233: Is this the only site where the OVP file is not aligned with the Pandora site? If so, why is it a good site to show?

- 234: Why are 10a and 10b not 10 and 11?

- 234: Figure 10a -- what is the AVG for Pandora? Is this 13:50 to 14:50? Or some other window?

- 235-241: The discussion of highway vs heating seems speculative.

- 250: There is currently no discussion of Figure 11 or 12.

- 253: "good agreement ... at most sites" - is this comment based on the 4 sites shown? or was more analysis done? It is good that here there is a statement about what the difference is, but are we really talking about just one day?

- 267: "for most sites" should be quantified. Of the N sites, M do not... I say this because the paper uses a small set of Pandora as representative. The conclusion could be interpreted as "most" Pandora sites. This may be true, but the paper does not show that. Instead it relies on a few (3?) case study sites.