Overall – This paper provides a useful study of trade-offs for the inversion of carbon monoxide emissions for both the data and a priori considerations. I find that the claim that the method could be suitable for near-real-time inversions is not supported by the results. However, the main conclusions are sound and I think the paper could be published after addressing some concerns.

• Near real-time suitability: No timing trade-offs were presented, in fact the only discussion of this was that 5 real-world days were required for each inversion and more for satellite full res. Also, the grid resolution did not allow characterization of biomass burning events, a primary motivation for near-real-time. It was suggested regional analyses, with a finer zoomed resolution, would be possible, but these were not demonstrated here.
• Use of OH monthly climatological fields (L. 137) should have more discussion of why this choice is applicable for the TROPOMI time range.
• The only comparisons are for the surface CO flask observations. It would be of interest to see comparisons to other independent CO satellite observations.

Other suggestions:

L 66: should include the following reference:

Naus, S., L. G. Domingues, M. Krol, I. T. Luijkx, L. V. Gatti, J. B. Miller, E. Gloor, S. Basu, C. Correia, G. Koren, H. M. Worden, J. Flemming, G. Pétron, and W. Peters (2022), Sixteen years of MOPITT satellite data strongly constrain Amazon CO fire emissions, Atmospheric Chemistry and Physics, 22(22), 1473514750, doi:10.5194/acp-22-14735-2022.

Plots: Lines indicating color in the plot legend are difficult to distinguish  - maybe make these thicker.

Readability suggestions:

L1: “making knowledge” => improving knowledge

L8: “captures flask measurements south of 55◦ N almost as good as” => agrees with the flask measurements south of 55◦ N almost as well as

L11: “In part, the reduction over China can be attributed to policy changes.” => In part, the reduction over China can be attributed to policy and technology changes. (see Zheng et al., 2018 in your refs.)

L13: “Conversely, in most experiments,…” These last 2 sentences are confusing, it seems like both a model and data weighting consistency issue that could be stated more clearly.

L32: “Estimating regional CO emissions and dividing them up by source categories at a global scale is not trivial.” => Estimating regional CO emissions and partitioning them by source categories at a global scale is challenging.

L76: “whereas MOPITT and IASI are most sensitive to the middle and upper troposphere.” => “whereas IASI and MOPITT MWIR channels are most sensitive to the middle and upper troposphere.

L86: “This way, the everlasting trade-off, where increasing the model resolution implies not only rising precision but also rising computational cost, can be partly overcome.” => This allows us to partially mitigate the trade-off of improved precision from increasing model resolution for computational cost.

Summary: This study aims to provide global CO emission estimates for late 2018, particularly in the northern hemisphere, using a top-down inverse modeling approach. By incorporating TROPOMI satellite observations into the TM5-4DVAR model and further constraining emissions with NOAA surface flask measurements, six experiments were conducted to assess the impact of different emissions and observational datasets on inversion outcomes. The main findings of this paper are: 1) The satellite-only inversion closely matches flask measurements south of 55°N, suggesting suitability for real-time applications; 2) Up to 75% emission reductions in China and India, with reductions in China attributed to policy changes; 3) Outstanding issues were identified to include underestimation of OH causing lower emissions, and localized emission increments over Europe and the Sahara.

Main Comments:  While CO inversions have been conducted for over 20 years, there are still gaps in our understanding of CO sources and sinks despite availability of satellite data. This makes this paper potentially a relevant contribution to the scientific community and this journal given the importance of CO in understanding atmospheric composition. The authors have also described their inverse methodology in quite (and understandably) in detail which potentially can enable easier connection to relevant findings and issues identified in making the inverse framework suitable for near-real time applications. 

However, several concerns with this manuscript require revision:

1.The title “Efficacy of High-Resolution Satellite Observations” may be misleading, as approximations (e.g., model resolution and spatiotemporal scales of inversion) challenge this claim, and the findings don’t fully support it.

2. While the approach shows promise in reducing uncertainties compared to flask measurements, its suitability for near real-time inversions is not adequately demonstrated in this manuscript.

3. Verification is limited, as there’s no comparison with other satellite observations or top-down emission estimates for this period. Although TROPOMI offers extensive coverage, it has limitations due to its SWIR-only focus; integrating IASI, MOPITT, and CrIS would better constrain synoptic CO patterns and widespread anthropogenic CO, which flask data alone cannot capture. One way to alleviate this are: a) comparison with other top-down estimates, b) to make objectives to be more focused.

4. Methodology presentation could be clarified for readers less familiar with inverse modeling.

5. Enhance overall clarity, particularly by: a) clearly articulating key gaps in CO inversion methodology and current top-down estimates, and b) deepening the discussion of noteworthy findings that are better supported by experiments.

Specific Comments:

1) Recommend changing the title to reflect main findings supported by experiments.

2) While there is a good description of the methods, it may be more clear and easy to follow with an addition of a table listing all key approximations (resolution of observations and models, inflation parameters, super-observation parameters, inversion spatial and temporal windows, spin-up, specification of errors incl. length scales).

3. Abstract: While the length constraints are understood, several statements need clarification for clarity and impact: a) What is the significance of 55°N, and why would capturing measurements below this latitude suggest suitability for near real-time inversions? b) “Attributed to policy changes” – could you please specify which policies? c) “Appears to be underestimated” – is this based on comparisons with other OH fields? d) The last two sentences are unclear, especially “model’s limited capabilities to capture”; could you please provide more detail?

4) Line 32. Can you please rephrase? What do you mean by 'dividing them up by source categories. 

5) Line 34. 'they carry insufficient information' . Please elaborate

6) Line 35. What do you mean by 'incorporating some additional information?

7) Line 37-38.'process that caused the emission is measured';  'emissions can be extrapolated'. Please rephrase to make it more accurate.

8) Line 42-45. 'direct observations of the source event', 'loose observational requirements'. Please rephrase to make it more accurate. also, 'potentially more elaborate' - why potentially?

9) Line 48. 'top-down approach in the form of inverse modeling'. are there other forms?

10) Line 58. 'including information from additional observations'. what do you mean by 'additional'?

11) Line 83-96. The zooming capability is a very important point (strength) for this paper which should be highlighted more and taken advantage in extracting full information content of high resolution datasets to address a science objective. Why would you reduce the observation resolution then? especially that this paper is considering: a) efficacy of high resolution obs, and b) near-real time application. 

12) Line 97. 'as a proof of concept'. has there been no inversions using tropomi yet?

13)  Line 117. Can you please elaborate why CMIP6 emissions are used? 

14) Line 129-135.  For inversions using full TROPOMI resolution, was the model resolution also change appropriately?

15)  Line 137. Please elaborate on the rationale for the use of monthly OH from  TransCom-CH4. Several studies have pointed out issues with using prescribed OH climatology especially for 'regional' inversions.

16)  Line 173. ' please elaborate on ' assimilating multiple datasets with different spatial and temporal resolutions at once and co-sampling of observations across datasets is neither necessary nor detrimental'. what do you mean by 'at once', 'co-sampling' and detrimental to?

17) Line 191. 'no daily cycles'. while it is consistent with OH, it is not suitable for inversions using 7km data. there's a mismatch in scale.

18) Line 230-270. While the use of error correlation lengths is commendable, please elaborate on the scales use in the inversion? How are these calculated/estimated? This could be very useful to the community.

19) Line 296-318. Please rephrase these paragraphs to make it clearer. It is unclear for example what is the exactly spinup period (and pinup inversion) and how was this conducted. For CO, this spinup period can be quite important, especially that there are differences in the model configuration of the initial conditions (incl. more importantly a different OH).

20) Figure 2. Nice figure to better understand the elaborate gridding strategy. Still not sure though that this is appropriate for near real-time applications.

21) Line 517-524. Nice discussion on the limitation. It still begs the question (same as above comment) if this elaborate weighting strategy is appropriate.

22) Table 3. Very helpful to have. 

23)  The manuscript’s honest discussion of inversion increments and identified issues is commendable, yet it risks obscuring key findings and blurring the clarity of its objectives. Reorganizing the structure and sharpening the focus would enhance clarity and impact.

Review of Efficacy of high-resolution satellite observations in inverse modeling of carbon monoxide emissions using TM5-4dvar (r1258) by Johann Rasmus Nüß et al.

The article is quite comprehensive with many tests and evaluations. It is somehow frustrating that a lot of efforts has been put on looking at fire emissions with different priors and a setup attributing more errors to fire emissions, but in the end, the results not being presented (sect. 4.3.3). The paper would read better either with the inclusion of the fire emissions results, eventually with some explanation about why it does not work as intended to guide future studies, or by removing the sensitivity to prior fire emissions to make the paper more concise.

The CO budget is not well presented with a lack of references to other CO inversions studies and other publications that compare inversion results, for instance Elguindi et al (2020). 

Reference: Elguindi, N., Granier, C., Stavrakou, T., Darras, S., Bauwens, M., Cao, H., et al. (2020). Intercomparison of magnitudes and trends in anthropogenic surface emissions from bottom‐up inventories, top‐down estimates, and emission scenarios. Earth's Future, 8, e2020EF001520. https://doi.org/ 10.1029/2020EF001520



Minor comments:

The title is misleading as most of the work is done with large scale setup for monthly emission inversions, and is focus on the comparison of TROPOMI inversions with global in-situ network. 

Abstract: “Compared to the bottom-up estimates, all experiments result in strong (by up to 75%) broad-scale emission reductions in China and India. In part, the reduction over China can be attributed to policy changes.”

Explain the time period, the differences can be in absolute sense, or because of the change in emissions over time, please clarify.

L81: “spatial sampling of IASI (up to about 25×25km2; Clerbaux et al., 2009).”

The citation indicates that IASI has footprints with diameters of 12 km diameter footprint.

It is confusing because it looks like the MOPITT pixels are of similar size compared to IASI

L142: 2.2 4DVAR approach: I am sorry if I missed it, but what is the assimilation window ?

L227: “the a prior error is set to zero over the ocean”

Chose between a priori or prior

L248:”Therefore, we use an exponentially decreasing correlation time of 9.5 months for the secondary CO production at different times from the same cell.”

This seems to be a long time as CO itself has an average lifetime of 2 to 3 months.

 

L265: “and the fairly up to date inventory (with historical data up to 2014 and projected data from 2015 onwards),”

It is not really up to date. But what matters in the end is how the scenario matches the observations. 

L728: “Regardless, our extrapolated annual a posteriori budget terms are much closer to the ones found in literature (e.g. Zheng et al., 2019) than the a priori terms, implying that the a posteriori terms are more realistic.”

Please explain and clarify, maybe cite more paper about CO budgets, it is an interesting part of the paper that is not really complete at the moment.