Lauster and colleagues present a study that investigates the accuracy of two O2-O2 CIA cross sections with long-path DOAS measurements. The measurements are carried out at an arctic site that experiences large temperature variations and is otherwise clean. Calculated oxygen abundances are compared to retrieved O2-O2 CIA, with varied fit settings. The study confirms that the cross sections are overall accurate and not the cause for scaling factors frequently needed in MAX-DOAS. The more recent Finkenzeller & Volkamer cross section seems to produce more accurate results for values of O2-O2 CIA, especially regarding the temperature sensitivity.

This scope of this study is nicely set up. The data set lends itself to this type of analysis. The logic of the analysis is appropriate. The figures are adequate. Overall, I am convinced by the work and the results outlined. I enjoyed reading the manuscript and I am looking forward to seeing this manuscript published. However, I have a couple points where I would ask the authors to refine the manuscript.

I do believe that referring to O2-O2 collision-induced absorption as “collision complex” or “collision pair” is wrong, and “O4”is misleading. It creates a wrong conception of the physical effect. I understand that “O4” is rooted in the MAX-DOAS community, and “O2O2 CIA” doesn’t come off the lips as easily. However, I do ask the authors, particularly in light that this is a manuscript investigating our understanding of this effect, to be accurate when referring to it. Please also consider referring to the excellent monograph on collision-induced absorption by L. Frommhold. I understand that getting away from “O4” is an uphill battle, but I invite the authors to pick up this fight, and I invite the editor and other reviewers to chime in.

17: what is “atmospheric” in “atmospheric absorption”

18: See main comment. Maybe something along the lines of “O2-O2 CIA occurs proportionally to the square of the oxygen concentration.” could work.

37: What art the differences and common aspects of the two cross sections? Please help the reader to understand the difference.

90: What is the resolution of the cross sections? What is the resolution of the spectrometer? (I think this is relevant information that should be within this manuscript.)

97-99: Is this novel? If not, please reference the method.

110-118: I was left a little unsatisfied about why 8000 iterations were identified as best setting. How would the further interpretation change if another value was chosen? How does the value affect the seemingly persistent offsets in the O2-O2 CIA? This should be discussed further. Maybe a duplicate of Figure 8, but with another iteration number, could be interesting.

Table 1: The number of iterations for the high-pass filter should be 8000, rather than 4000, shouldn’t it?

Figure 4: The highlighting frame does not work well. Could be left out or improved.

Figure 5: Which cross section underlies these data? Currently not clear. The link to the table is broken in the caption.

128: How does humidity play in here? In the arctic it won’t matter much, but it should be discussed as being generally important. Is 21% sufficiently accurate? I believe 20.95% might be more accurate (0.1% difference in O2-O2 CIA).

137: Do I understand correctly that the pressure changes were not considered? A pressure change of 2% (1000 vs 980 hPa) would lead to a O2-O2 CIA change of 4%, rather substantial. Accounting for the pressure could improve the closure – or make it worse… Please consider considering the pressure, or discuss the effects of not considering it a little more.

142: The temperature dependence pertains not only to the peak absorption cross section, but also shape and integral cross section. Better to eliminate the parenthesis “i.e., the strength of its peak value”.

145: The reference to Fig A2 is misplaced, it should only be a reference to A3?!

146,147: Temperature should be 223 K, not 233 K?!

153: I disagree with the statement that the integral cross sections are independent of temperature. Thalman & Volkamer were not on point in this regard, but presumably rather intended to argue that there is no bound state (“O4”). The more recent Finkenzeller & Volkamer study found a temperature sensitivity, which is also expected from physics: “With increasing temperature closer encounters occur, which leads to stronger induced dipole moments and thus greater intensities [1].”

168: The interpolation of temperatures has been used previously. Please add (a) reference(s).

Figure A2: I was misled by the caption. It was not clear to me that the temperatures are the interesting parameter, rather than the origin of the O2-O2 CIA cross section. I suggest starting the caption with something like "Sensitivity of O2-O2 CIA to temperature. …”

This is an important study that aims to investigate the inconsistencies in MAX-DOAS measurements of O2-O2 collision induced absorption (CIA) and radiative transfer models. To this end, the authors analyzed two years of observations from a (LP-) DOAS instrument deployed at the Antarctica station.  In particular, they investigated the relation between measured and modeled O2-O2 (CIA) over a wide range of temperatures. Especially interesting is their access to the lower temperatures as a point of controversy or lack of data from experimental measurements. They have concluded that the set of laboratory data recently published by Finkenzeller and Volkammer provides the best agreement, however the discrepancies in MAX-DOAS are not compeletely  understood. It is an important paper using field data to evaluate reference spectroscopic information and should be published after following comments are addressed:

Major comment:

One should not expect constant or even linear dependence of CIA on temperature, whether looking at integrated band absorption or at individual spectral points. In particular for integrated absorption one should expect that the CIA should be more intense at low T because of bound states, and more intense at high T because it's easier for colliding molecules to get close together, and in between there must be some minimum. See, for instance, a paper by Vigasin on the temperature variation of intensity of the CIA underlying oxygen fundamental (https://doi.org/10.1016/j.jms.2004.02.003). In the Figure 1 of that paper you will see that the temperature dependence is not expected to be constant or linear.

Also, The distribution of the intensity within a band also should vary with temperature because of the Boltzmann population of energy levels. 

Could the interpolation potentially be done as a quadratic function combining the data from 2013 and 2022 papers from Volkammer’s group? The earlier work can be given enhanced uncertainties but still could be useful. In any case the 2013 data should not be fit to linear function. It would be good to include the figures associated with interpolation

A couple of minor suggestions

1. You can mention that Thalman and Volkammer CIA is currently recommended by HITRAN
2. In line 166 change “...a couple..” to “..a few…”  

Review of “Assessment of laboratory O4 absorption cross-sections at 360 nm using atmospheric long-path DOAS observations”

The paper by Lauster et al. presents long-path DOAS measurements in the remote location of Neumayer Station 3 in Antarctica to retrieve atmospheric O4 concentrations and to assess the accuracy of the laboratory O4 absorption cross-section, commonly used in DOAS applications. In the past, some studies reported deviations between MAX-DOAS O4 measurements and results from radiative transfer simulations, while others showed none. The authors questioned if the used O4 cross section in the DOAS fit can explain these deviations. Different O4 cross-sections at different temperatures are investigated in the UV spectral range. The study provides more insights into this debate and shows that the retrieved O4 concentrations align with the expected concentrations calculated from meteorological measurements on-site. However, better agreement is found for the newer version of the commonly used O4 cross-section.

The study covers an ongoing debate in the DOAS community, is well-structured, and provides further insights from another point using long-path DOAS observation.

I recommend publication after including the following points.

Specific comments:

Line 35: I am a bit confused about the wavelength range of 352 to 387 nm given here and the grey bar given in Figure 1, which are not the same. What are these wavelength ranges, the spectral range of the spectrometer, and the fitting window, maybe you can clarify this in the text shortly.

Figure 1: The unit of the cross-section on the y-axis is missing.

Figure 1: You mention that the 477 nm band is not covered, I think you can additionally mention that the 577 nm band is at the edge of the LP-DOAS visible wavelength range.

Line 55: Why have direct sun measurements further difficulties due to “small atmospheric absorption”? What is causing this problem, is it only the light path, which is shorter in the lower troposphere compared to MAX-DOAS, is this relevant for O4 absorption?

Line 58: MAX-DOAS and direct sun measurements don’t “measure vertical column densities”. I think both “measure” and “vertical” are technically not correct here, I suggest changing it to “retrieve” or/and changing it to slant column densities along the light path.

Line 75: I think it is not clear for the reader what kind of conditions are needed for your study. Large temperature range is good for analyzing the cross-sections at different temperatures. What is the low aerosol optical depth good for in your study – less attenuation? Why is this helpful?

What kind of measurements are used for your study/are you interested in? Do you use all available measurements from January 2016 to August 2019, do you filter for specific measurements, do you filter out bad weather? What is the temporal resolution of the measurements?

I think the section lacks information about the LP-DOAS instrument. The spectrometer is not mentioned at all; what kind of spectrometer, which wavelength range, which spectral resolution is used?

Figure 3: Monthly averages over which period, 2 years (2016-2017)?

Line 86: To clarify for people having no/little knowledge about LP-DOAS, add something like “creating a short-cut for the light, which stays inside the telescope and goes directly into the spectrometer”

Line 95: The high-pass filter is only needed in the UV but not in the visible (Table A1), is this right?

Line 101-104: That the O4 bands in the visible are not covered is not a general problem but only of this specific instrument, right? If you would have a different spectrometer you could cover other wavelengths.

Line 115: You mention that the HP 8000 is the standard, however, in Table 1, you write 4000, please clarify.

Line 131: What is the temporal resolution of the LP-DOAS and the meteorological data, how is the matching done?

Line 139: How is this grey bar defined?

Figure 6: How is the grey bar defined? Can you add it to the middle and right plots as well? Change “values” to “O4 concentrations” in the caption.

Figure 7 and 8: Change “values” to “O4 concentrations” in the caption.

Line 180-186: I think you should be here more precise – provide some numbers. What is a good agreement, what is needed/good enough? How good is your agreement for the 2022/2013 cross-section retrieval with the calculated concentrations?

Which wavelength ranges were used in the studies that are in need or don’t need the scaling factor, are there deviations, are they also in the UV – like the focus of this study? How well can you draw conclusions for the visible, maybe the problem is in the visible only?

You mention that the previous studies used the 2013 cross-section, for which you also see discrepancies with your measurements. Can you comment on this, are these discrepancies too small to explain the issues seen in these studies?

Technical corrections:

Line 40: This sentence is hard to read; maybe something like “At first, differences between MAX-DOAS and LP-DOAS measurements are introduced to answer the question of why LP-DOAS observations are well suited to further investigate the reported inconsistencies from MAX-DOAS studies.”

Line 58: “consider vertical profiles of temperature and pressure” instead of “consider temperature and pressure vertical profiles”

Line 66: I was confused by the “(Met retro)” and “(Atka retro)” when reading this, without retro reflectors mentioned before at all, and haven’t looked yet in Figure 2. I suggest adding something like “across a distance of 1.55 km to the closer retro reflector (Met retro), or to another retro reflector at 2.95 km (Atka retro).”

Line 66: The sentence is hard to read, I suggest something like “Depending on the prevailing weather conditions, the amount of reflected light varies, and the light path can be chosen depending on the atmospheric conditions to optimize the amount of received light and covered light path length. “

Line 75: Add a link to the AERONET webpage.

Line 76: Change “measurement days” to “measurements”

Line 87: To improve readability, change "in addition by shutting off the light source” to “without the light source”.

Line 93: Delete one “the”

Line 95: To clarify, change to: “Different analyses will be shown in the following, including one of the O4 absorption cross-sections listed in the table.”

Table 1: Change “various” to “various temperatures”. Add something like this to the caption to clarify that only one cross-section is used at one temperature at a time: “Different analyses will be shown in the following, including one of the O4 absorption cross-sections listed in the table.”

Caption Figure 5: The link to the table is not working.

Line 128: Use a point instead of a cross for the multiplication sign.

Line 134: This is more than two years, maybe just write “covering data from January 2016 to August 2018”.

Line 139: I would argue here on the retrieved O4 concentrations instead of the calculated, which you also do in the next sentence and the following section. I think it would improve readability.

Line 153: Brackets missing around reference.

Line 154: Shown where?

Line 186: shows instead of show

Line 186: Change “expected” to “calculated”.