General comments:

Ding et al. presented a comparison of NH₃ emissions in Europe derived from the CrIS satellite instruments and existing emission inventories. The authors included NO₂ observations from the TROPOMI instrument in the emission calculation as NH₃ emissions were found to be sensitive to NO_X. They also compared satellite and inventory derived NH₃ surface concentrations to the Dutch monitoring networks and found those using satellites had least bias against in-situ observations. This work makes a good example of why we need accurate emission data for policy interventions as bottom-up inventories often fall behind real-world changes. In this regard, satellites that offer daily global coverage are an excellent tool. Overall, I am curious as to why the authors only studied 2020 despite that CrIS and TROPOMI have data records that can be jointly studied beginning in 2018 (first year of TROPOMI operational product) until current day. We know 2020 was a special year in recent decades as anthropogenic activities and emissions were largely altered by the pandemic. As the authors mentioned, NO_x emissions decreased in 2020. Meanwhile, NH₃ was found to have an “anomalous increase” during the lockdowns (Kuttippurath et al., 2024). It is not surprising that inventories did not capture these sudden changes as observations (satellite and in-situ) did. In fact, the authors clearly stated one of the inventories was the 2018 data, which does not make a very fair comparison. Since DECSO is not computationally demanding to run, I would be curious to see at least one more year of results, otherwise it is difficult to comment on the general applicability of DECSO.

Specific comments:

Line 28: Is the 0.2% precision the uncertainty of derived emissions or the agreement between the two satellites? Also please include the uncertainty of other emission estimates, such as Line 305-306.

Line 83: You should clarify that JPSS-1 was renamed as NOAA-20, which is the name being used later in the text. Also clarify JPSS-2 (NOAA-21) was launched much later and therefore not used in the study.

Line 174: Can you expand on what you mean by superobservations and how is it different from the original observations?

Line 187: How large is the uncertainty in the in-situ data? This is an important part that should be discussed since you are using in-situ data as the reference for comparing satellites and inventories.

Line 298: Is this DECSO-parallel version using the setting of multi-species DECSO or DECSO-NH3 only?

Line 315: Do you know if the underestimation of emissions in East Europe by inventories is due to inaccurate information collected in those countries or other reasons? A little more insight on this would be helpful.

Line 373: Can you explain more specifically what caused the displacements of emissions? Is it intrinsic to the data, or does it have something to do with the way you aggregate the emissions (Line 213)?

Line 387: Please explain how you define and compare the biases here, on Line 392, and in Table 2 (assuming they mean the same concept). I can see from Figure 7 that HTAP is underestimating and CAMS is overestimating. However DECSO also has many extreme values indicating it can overshoot either way, so I’m not sure if it really is the better option, especially if you look at Table 2 the spatial correlations and biases of DECSO and CAMS aren’t that different (with CAMS having a much better temporal correlation). And please include the uncertainties of these numbers too.

Line 417: I would include the correlation coefficients and biases here or show them on Figure S9.

Line 480: Given that your derived emissions (8 Tg/year) are almost twice as high as inventories (4.0-5.9 Tg/year), how much of this difference do you think may be attributed to the fact that the inversion is solely based on satellite overpasses in the afternoon when NH₃ concentration is usually higher than rest of the day (in other words, not incorporating diurnal changes)?

Technical corrections:

Line 18: “2.5” in PM2.5 should be subscripted throughout the text.

Line 54: Studies show

Line 122: to validate emissions

Line 151: The NH₃ profiles

Line 268: there are almost no

Line 331: Figure 3 caption says LRTAP 2018 but figure legend says LRTAP 2020.

Line 336: is closer to

Line 345: Spring does not need to be capitalized here.

Line 462: redundant “of”

References:

Kuttippurath, J., Patel, V. K., Kashyap, R., Singh, A., & Clerbaux, C. (2024). Anomalous increase in global atmospheric ammonia during COVID-19 lockdown: Need policies to curb agricultural emissions. Journal of Cleaner Production, 434, 140424. https://doi.org/10.1016/j.jclepro.2023.140424

This MS deals with the ammonia emission estimates for Europe using satellite observations. As there are not many studies like this, the MS has new information and thus, can be considered for publication. However, there are certain things to be clarified before the MS can be accepted.

Minor:

1. Neither the Abstract nor Conclusions provides a clear message about this study. What is the main conclusion from this study? Just the comparison of different estimates? How we can improve the emission estimates? What are current uncertainties? What uncertainty is addressed in this study? This has to be clearly mentioned in the Abstract and Conclusion, and the problem must be properly discussed in the Introduction too. Now this is just a comparison paper, as you stated in the last sentence of the Abstract.
2. Please state why the year 2020 is selected for this particular study.
3. When you add NOx, how much the emission estimates improved in percent? Please state this.
4. There is a clear seasonal difference in emission estimates. This suggests that the meteorology plays a big part in these calculations. So if you change meteorological input, how much will be the difference? If you consider the uncertainties in meteorological data, what would be the uncertainty or bias in the estimated emissions?
5. Conclusion is almost two pages, but I do not see that much discussion in the paper. So please make it short. A 350-word Conclusion would be much more effective than this.
6. I find some language issues in the MS, please pay attention to that.

technical:

L19: affect

L26: limited daily observation

L28: be specific about the region you mentioned here

L52: citation is not correct

L66: too many “type” here. Please rephrase

L270: this aspect is discussed in this paper https://doi.org/10.1016/j.jclepro.2023.140424

L301: How small is this “small”? Please specify the value

L385, L388: Please state the reasons for the differences

L410: If you look at the table, the comparison shows better for DESCO, but Figure 7 shows the least bias for HTAP?

L420: Figure 7 shows the smallest bias for HTAP

L465: “For Europe”

---