This work by Pope et al. has a promising title and research goal, but fails to meet its expectations. Large parts of the methodology are flawed or missing (see critical points), and the overall presentation is too vague or even sloppy (see comments on presentation). This work does not properly answer its research questions as a result. I cannot recommend it for publication in its present format.

Critical points:

Why apply bias correction factors for trend studies? This is especially questionable if their construction involves ‘application’ of the averaging kernels that are under study! If this operation allows for more ‘like-for-like comparisons’ indeed, it should be rigorously and analytically explained in the main text, as a key methodology part of the paper.

Why involve a model and apply the averaging kernels thereto (hence hiding certain information), even without mentioning how, if the retrieval sensitivity can be thoroughly studied and discussed in itself (see Rodgers, 2000)?

The authors at several instances claim that the substantial IASI-FORLI trends are ‘believed’ to be due to changing meteorological input to the data processing in September 2010. The obvious check of doing independent trend studies before and after this change is missing.

It is agreed with the authors that “Ideally, this analysis would have utilised several more records (e.g. several UV-Vis and IR products) to quantify long-term trends in LTCO3 and investigate the potential reasons for any discrepancies, as shown by Gaudel et al., (2018) for TCO3.” At this point, none of these research goals is met. The authors could either focus on trend studies, or investigate whether the observed trends are (partially) due to spatio-temporal sensitivity changes. For the latter, it would suffice to compare ozone column trends with vertical sensitivity trends (without the need for models or bias correction).

Comments on presentation:

First key point: “trends […] show small scale trends” ?

Abstract: “year-to-year sampling is not an issue” is too vague. Do you mean changes in spatio-temporal sampling of satellite observations, or temporal changes in vertical smoothing (and hence apriori contributions) of observations? I see this briefly explained between brackets in the discussion only.

In the introduction and discussion, the authors fail to acknowledge that an important reason for the discrepancies observed by Gaudel et al. (2018) was the use of different tropospheric top level definitions in different satellite products.

All apriori data should be properly introduced and discussed, given its relevance for the results interpretation (e.g. missing seasonality in IASI-FORLI apriori).

What kind of averaging kernels are provided with the IASI-SOFRID L3 product?

C in equation 1 must represent an ozone value, not a month.

In sections 3.1 and 3.2, the essential information is lost in overdetailed number repetition that should be succinctly summarized. Correspondingly, Table 2 should go into the supplement.

Stating that “individual retrievals of LTCO3 are subject to multiple issues (…) which can result in noisy LTCO3 time-series” (lines 377-380) sounds unscientific, and diminishes the efforts done by satellite retrieval teams.

In the discussion, stating that “The IASI-SOFRID LTCO3 and apriori are very similar, with little inter-annual variability, which suggests that the IASI-SOFRID O3 retrieval in this height-range is more constrained by the apriori (i.e. less TO3 sensitivity than the other products).” is not necessarily true. The apriori can be close to the retrieved value, even for a perfect retrieval. Hence, again, the need for proper sensitivity studies, instead of derived quantities.

Why are ozonesondes considered in 30° latitude bins, and not matched with the three regions under study? Moreover, it is not mentioned how many stations / launches are eventually involved.

How are data interpolated between different vertical representations?

Equations 2 and 3 are essentially the same. One should not make a difference based on satellite data formatting, which is irrelevant.

How are IASI sub-columns totaled up the 450 hPa level? Does this require interpolation between levels?

Figure S1: Why are correction factors (multiplicative) expressed in DU?

SM-3 should be part of the discussion (if somehow the model comparison is maintained).

Not all authors are in the Author Contributions.

Overall:

This study presents a useful comparison of impacts from satellite sampling and vertical sensitivity on lower tropospheric ozone column regional averages, seasonal amplitudes and trends. Although the trend comparison from the IASI-FORLI data is impeded by known discontinuities in the record due to retrieval inputs, it is still valuable to see the effects of the sampling and application of averaging kernels to the reference model. I recommend publication after providing more information and references and addressing some structural and readability issues.

Specific comments:

L 47: “other factors that require further investigation” Please be more specific here since TOAR-II is trying to address these issues.

L 73: “... will have an impact on which part of the troposphere the O3 signal is weighted towards”  sounds somewhat confusing - maybe reword to:    “... will have an impact on the weighting of the tropospheric O3 signal in the reported satellite products”

L 79: “and around 11-12 km for IASI”  This is true for the maximum and in the global average, but neglects that there is sensitivity (i.e., a secondary peak)  in the lower troposphere around 5 km (e.g. Boynard et al. 2009)

L 116: The supplementary material should also show the bias correction factors for the IASI data versions. Supplementary material could also discuss issues with temperature, water vapor input to IASI retrievals and how that potentially affects bias correction.

L 116: This description of BCFs should mention that these are based on record monthly averages and would not change trends. This section should also include discussion of how this bias correction compares to the harmonization methods of Keppens et al., 2019.

L 292: This section would be easier to read if the average ± error information was in a table

L 304: If you keep this in the text, I think ±error (DU) would be easier to interpret than error ranges

L 379: should include water vapor

Minor comments:

References in the supplementary material were not included there.

L 58: presents -> presence 

L 187 titled subsections for each region (N. America, etc.) would help the reader here.

L 218 start new paragraph for IASI-SOFRID results

L 220 also new paragraph for ozonesondes

References needed:

Boynard, A., Clerbaux, C., Coheur, P.-F., Hurtmans, D., Turquety, S., George, M., Hadji-Lazaro, J., Keim, C., and Meyer-Arnek, J.: Measurements of total and tropospheric ozone from IASI: comparison with correlative satellite, ground-based and ozonesonde observations, Atmos. Chem. Phys., 9, 6255–6271, https://doi.org/10.5194/acp-9-6255-2009, 2009.

Keppens, A., Compernolle, S., Verhoelst, T., Hubert, D., and Lambert, J.-C.: Harmonization and comparison of vertically resolved atmospheric state observations: methods, effects, and uncertainty budget, Atmos. Meas. Tech., 12, 4379–4391, https://doi.org/10.5194/amt-12-4379-2019, 2019.