Review of Andrea Orfanoz-Cheuquelaf et al., Tropospheric ozone column dataset from OMPS-LP/OMPS-NM limb-nadir matching

The manuscript Tropospheric ozone column dataset from OMPS-LP/OMPS-NM limb-nadir matching by Andrea Orfanoz-Cheuquelaf et al. descripes the adaption of the LNM algorithm for SCIAMACHY (Ebojie et al., 2014) to the OMPS instrument. The errors are estimated and moreover the retrieved tropospheric ozone columns are compared to ozone soundings as well as comparable satellite data sets i.e. OMI-MLS and S5P CCD.

General Remarks / Questions

The total column is retrieved using a WFDOAS approach (or similar) here a view more details on the algorithm might be given e.g. is a stratospheric ozone profile necessary if so which one is used? As a reader we don't know all the details that might be obvious to the authors and have been presented in previous publications.

Concerning the tropospheric ozone retrieval, are the total and stratospheric columns retrieved independent from each other, or are the retrievals linked? For example one might think of using the stratospheric profile retrieved from the limb observations in the total column retrieval, or constrain the stratospheric column by the total column as upper limit.

Only cloud free pixels (cloud fraction less than 0.1) is used. Which cloud data are used here? If only the cloud fraction is used the VIIRS cloud fraction might be an option?

The authors discuss a very interesting issue in the OMPS profile retrieval over the Pacific Ocean around 10° N and attribute it to a possible cloud effect on the profile along the Line-of-Sight (p 16 l 335). I fully agree that clarifying this issue in detail might be worth a detailed study. On the other hand as a first check it might be worth to skip the profiles being affected by clouds some hundreds of kms north or south (along the Line-of-Sight) of the tangent point. This might give a first indication whether your hypothesis is correct, provided you have enough data. This effect is only observed over the oceans but not over the continents, even though or because the convection is stronger over the continents. Is this in agreement to the current hypothesis? Unfortunately there are no ozone soundings available in the most affected regions.

The tropospheric columns agree more or less with the data from OMI-MLS or S5P_CCD. All three retrieval approaches make use of the residual technique: TrOC=TOC - SOC. Can the observed difference to OMPS-LNM be attributes to the total column or the stratospheric column. For the drift relative to OMI-MLS (p 19 l 412) this might be of interest.

Data from 2012 to 2018 are presented, however no reason is given why the following 4 years (2018- 2022) are not included, yet.

Detailed remarks are listed in the supplement

The manuscript “Tropospheric ozone column dataset from OMPS-LP/OMPS-NM limb-nadir matching” by A. Orfanoz-Cheuquelaf et al. presents a new research tropospheric ozone product derived from Suomi NPP OMPS satellite measurements. The paper fits to the scope of the AMT journal. The paper is well structured and written. This dataset will be of interest to the atmospheric community especially if it will be combined with a similar dataset from SCIAMACHY as promised in this publication. I recommend this paper for publication after a revision. My comments are summarized before.

General Comment:

It is not clear why the dataset is limited to 2012-2018 time period. Considered that Suomi NPP OMPS is still an active instrument it would be desirable to extend this dataset and provided analysis at least untill the end of 2022. That would extend the overlap with TROPOMI substantially and provide community with the up-to-date information about the status of tropospheric ozone layer.

Specific comments:

p. 3 line 85. You need to rephrase this statement “The charge-coupled device performs instantaneous measurements of the entire atmosphere”. I assume you meant that radiances are collected simultaneously spectrally and spatially over the FOV.

p.4 lines 101: Why did you include only odd-numbered spectral points and how does that relate to the temperature dependence? I assume you meant the temperature dependence of O3 cross-sections, right?

p.4 line104: if your intention is to name the instrument first and then the algorithm type, then it should be “OMPS-NM TOMS” rather than “OMPS L2”.

Section 2.2. I was a bit confused with version numbers and references. First you refer Arosio et al., 2018, then you stated that in this study you use v3.3, but didn't provide a reference or any explanation of how that v3.3 differ from Arosio et al., 2018.  It is not clear what had happened in between. Are there v2.7, 3.0 etc.?  Please, clearly explain which version you use and provide the correct references. If the refernence doesn’t exist, then explain how this v3.3 differ from what is described in Arosio at al., 2018.

p. 5 line 124: why can’t you trust “OMPS-LP ozone profile time series based on level 1 V2.5 data” after 2018? Please provide reference or explanation. And again please clearly explain if v3.3 uses v2.5 Level 1 data or not.

P5., line 127: I didn’t find any mentioning of IUP-OMPS v2.6 in Arosio et al. 2018. Are you including the right reference?

P5., line 131: Why do you see the improvements in v3.3? How does v3.3 differ from v2.6?

Figure 1. what is the time period for the comparisons? 2012-2018? Please, specify.

P.7 lines 162-163 I assume you meant that retrieved ozone is very noisy below 12.5 km. Have you tried to use retrieved values instead of climatological? Does that change calculated SOC values in any significant way?

P.7 Lines 167. Are you deriving the cloud height from OMPS-LP? Then you should have mentioned how it’s done in Sec. 2.2.

P.7 line 184 Since the cloud fraction is one of the key factors in producing TrOC, it would be good if you describe how the cloud fraction is calculated in Sec. 2.1 when you talk about TOC.

Figure 2. All labels should be explained. What does “Pixel n^o 110 and FOV n^o 14” or “State n^o 85” mean? Also, it might be better to say: “The red points mark the footprints of tangent points (TPs) of the limb observations”.

Table 2. I am not sure I understand the source of error titled “Tropospheric ozone increase”. First I thought it represents the error in apriori, but then I found you have another entry for “O3 and T apriori”. Please, explain how the changes in tropospheric ozone affect the retrieved TOC.

Table 2. What is the threshold for “enhanced aerosol”?

p.9 line 218. I assume you mean to include a reference after “…are quantified using synthetic retrievals and extensively discussed”.

Figure 3. What did you mean by “syst. bias”? Biases in measured radiances or biases in retrieved ozone?

Section 4.2. and Fig. 3. In my view several types of errors (T, P, albedo and TH) are highly correlated and it’s hard to isolate contributions from them. Limb scattered radiances are directly proportional to atmospheric air density. T and P are used to calculate the density, and therefore these errors are correlated. I feel that TH error is perhaps the leading source of uncertainties in limb scattering measurements. I am not sure what uncertainties in TH you are assuming to get the TH error. I believe that the TH error has two components systematic and random. Same is with Pressure. If sea/surface pressure is off in GEOS assimilation that would produce a systematic bias in P profiles. In some conditions the two errors will cancel each other, in other cases they will amplify the resulting error. Wouldn't error in albedo be a systematic error? Albedo depends on absolute calibrations and TH accuracy. I am confused with how the errors are sorted by systematic and random.

Section 4.3. Again you are ignoring the systematic part of the error in TPH which might be much larger than the error in vertical resolution you quoted here.

P.12, lines 279-281. Since you are considering only nadir pixels and illuminating cloudy pixels, you end up with the stripes of data along the orbital tracks. The statement that “you are binning data” in my view misrepresents the reality. Perhaps, it would be better to say that you map your sparce measurement on to the regular grid.

Figure 5. Please, specify how these anomalies were calculated. Did you subtract the long-term annual averages or seasonal averages? Also please, add a) and b) signs to each panel.

Figure 6. Be more specific. I assume SOC is from OMPS-LP as well as could top height, right? Is the surface reflectivity from OMPS LP or OMPS NM?

p.15 lines 316-318: In the text you need to add references to the figures you meant “The SOC anomalies (Fig.5b) show lower values over the Pacific and Atlantic, matching the band of high TrOC (Fig 4). This feature is not evident in the TOC anomalies (Fig. 5a)”

P.15 lines 322-324. How did you calculate the surface reflectivity? Is it from OMPS LP or OMPS NM?

16, lines 340-346. Are these weighted averages? How many adjacent pixels were used/considered? What is the range of distances between the station and co-located OMPS? How was the temporal averaging applied?

P.16, line 360. How was the bias calculated? Did you calculate the mean for each instrument and then estimated the bias? Please, explain.

Figure 7. Are you showing individual measurements or weekly/monthly averages. Please, specify that in the figure caption.

P.20, line 420. From figure 7 at Broadmeadows it seems that OMPS LNM overestimates the seasonal cycle compared to sonde as well.

P.20, conclusions, lines 433-435: I disagree with the conclusion that “no seasonality in the differences”. There is a clear seasonal bias in SH between LNM and OMI/MLS shown in Fig. 9.

Minor comments:

3 line 70. Should it be “OMPS comprises of three instruments…”

P6., line 143. You defined PV above, but never defined PVU.

Figure 6. It should be “over the Pacific Ocean from OMPS”