The authors clearly demonstrate the significance and necessity of improved NH₃ flux estimates, and the results present interesting spatial and seasonal patterns that are valuable to the community. However, the methodological description is currently insufficient and relies too heavily on briefly reproducing elements from Ayazpour et al. (2024) without providing enough self-contained derivation or detailed explanation. To ensure clarity and reproducibility, I strongly recommend expanding Section 2 with a more thorough presentation of the directional derivative framework, explicit definitions of all key variables and terms, and clearer justification for the assumptions made. Addressing these issues will greatly strengthen the scientific rigor and standalone value of the manuscript. Furthermore, emissions are presented in quite some detail, but the deposition fluxes seem to draw the short straw. The lack of dry-deposition measurements of course does not help in being able to make a nice comparison, but some discussion or comparison with past modelled results would increase the value of the deposition results. Subsequently linking those results to for example critical load limits could greatly improve the overall value of the manuscript and enhance the impact of all the work that is already performed.

Main comments:

1. Section 2.2 — Please define “DD” clearly before using it throughout the paper. Also clarify the meanings of “∇” and “Ω” (column density) upon first mention.
2. Line 136 vs. Line 223 — There appears to be an inconsistency regarding the emission inventory year: Line 136 mentions using HEMCO for 2016, while Line 223 states the focus is on September 2019 to April 2021. Please clarify which year(s) of the emission inventory were used and how they relate to the period analyzed.
3. Line 143-144: Why are the wind data at 100-10m used when the observed air masses are clearly well mixed within the mixing layer, would a wind speed more representative of the mixing layer (500m-1000m) not make more sense? Wind speeds closer to the surface can also be expected to be much smaller, which seems essential to for the resulting fluxes. Either add a reference for this value or better explain the expected uncertainties.
4. Line 197–199 (Fitting criterion) — The statement “This fitting was limited to rough terrains…” conflicts with the description “This step was conducted in flat terrains…” and also appears to misrepresent the original criterion (Ayazpour et al., 2024, Sect. 3.2: “which eliminates open water and very rough terrain”). Please clarify whether the fitting excludes both open water and very rough terrain, or whether it is limited to rough or flat terrains, and explain whether the fitted parameters from flat terrain are appropriate for application in mountainous areas or for the entire CONUS domain.
5. Scale Height Assumption — The manuscript assumes a regionally constant scale height. Considering the substantial local variability in boundary layer depth and surface conditions, could the authors discuss how this assumption affects the flux estimation, particularly over complex terrain, and whether a spatially or seasonally varying scale height was tested?
6. Inverse scale height – As stated the scale height has a direct relation to the mixing layer, wouldn’t it make more sense to use a boundary layer height product to substitute into these functions instead of deriving them from the satellite observations? This will probably remove potential artifacts from spotty spatial/temporal measurement records and smooth out the resulting fields. (Based on L253-255 this does not seem to be a bad idea).
7. Chemical Loss Term — dropping the lifetime term because of a bad fit seems a bit easy and one that potentially has a large impact on the resulting emission and deposition fluxes, especially when moving away from the strongest emission gradients.

According to Ayazpour et al. (2024), a stricter maximum emission threshold is necessary for fitting DD_chem than for DD_topo. Could the poor fitting performance of the chemical loss term (line 247) be related to an insufficiently strict threshold? Please clarify and discuss whether further refinement of the X and k estimates is planned.

Additionally, the fit seems to have been performed on the whole of  the CONUS, whereas lifetime will vary strongly depending on the local pollution levels of other species (produced hno3/h2so4). A switch to locally varying fits would make sense from a chemistry point of view.

Alternatively, the authors could add an Alinea what the expected lifetime to chemistry is for typical hno3/h2so4 concentrations, and discuss from that point of view if chemistry is important or not. In its current form it’s not convincing.

1. Daily Data and Bias Correction — Using only daytime data may overestimate NH₃ due to satellite sampling biases. Have you considered applying a satellite bias correction, as recommended in Ayazpour et al. (2024, Sect. 2.2.4)?
2. Line 301-305: I completely disagree that the directional derivative minimizes the impacts of offsets and scaling differences between the products. As the authors know the bias in the satellite products are not spatially and temporally independent, which means that any offset/scaling difference will vary from point to point, especially around larger shifts in concentration levels. If anything the scaling between the satellite product biases will be enhanced by the derivative. Additionally, the detection limit of both satellites will of course also play an important role in the limitations to detectable gradients. For example the early spring peaks detected by CrIS but not by IASI could be a sign of detection limit (or a strong diurnal variability in emissions of course). Some validation studies are available for both satellite products. These can be used as a basis for an error/uncertainty estimate of the effects. Please show what is the expected uncertainty of the current product bias/scaling on the resulting fluxes. An update to the limit used later in the manuscript (i.e. +-2σ) might be needed.
3. Suggestion for further validation: The comparison with inventories is interesting, but more direct evidence of the value of satellite based emission and deposition estimates would be a comparison with in-situ data, which currently is missing. Feeding back the emissions into a CTM and comparing the resulting concentrations with in-situ data would strengthen the case that satellite based emissions are an improvement over current inventories. Additionally, a comparison of the simulated deposition data (based on the updated emissions) with the satellite derived estimates would further show the value of those deposition estimates. It is quite an effort though so I would understand if the authors state it’s beyond the scope of this manuscript.

Minor edits/comments:

Abstract L29: “atmospheric” instead of “atmosphere”

L45-46, I’d rephrase this sentence. Spatially the emissions seem to align, but seasonally and in amplitude they do not.

L74: a few hours is on the low end of the model and measurement estimates, mostly derived from direct fits on satellite data, which are expected to bias low. Estimates of 8-12 or up to 24 hours seems more reasonable based on literature.

L84: Quite a recent reference, the relation between volatilization and environmental conditions was known much before this point.

L103-105: Quite the claim when later analysis mostly focuses on monthly or longer temporal resolutions.

L124 onward: add the observational periods of each satellite after each satellite name, this will make it easier for the reader to follow what satellite is in orbit when.

Line 182-185: Essentially you are gap-filling the record, but I fail to see the basis for just inflating the pixel size without any smart input of additional data. I can imagine this type of gap-filling introducing stronger or weakening gradients in regions with very localized sources and/or very common wind directions. Please add a few words on potential effects on gradients.

Figure 1: Whats going on with the few outlier months in the CrIS and IASI records, are these specific periods? And what does excluding these do for your results?

Line 313-315: what about the instrument detection limit?

Section 4.2: or in discussion: I miss a discussion on the potential effects of instrument/product bias changing over time, and the expected impact compared to the increasing trends you observed here.

References:

Ayazpour, Z., Sun, K., Zhang, R., & Shen, H. (2025). Evaluation of the directional derivative approach for timely and accurate satellite-based emission estimation using chemical transport model simulation of nitrogen oxides. Journal of Geophysical Research: Atmospheres, 130, e2024JD042817. https://doi.org/10.1029/2024JD042817

Li et al. presented an analysis using IASI and CrIS ammonia retrievals to study ammonia emissions and deposition distribution in the US and the associated seasonality and trends.  This is an interesting study. However, I have several major comments concerning (i) the validity of the presented results, (ii) organization and presentation of the analysis and related discussions. These comments need to be addressed before the paper should be considered for publication in ACP. I would like to encourage the authors carefully address the concerns raised in my major comments #1, #2, and #3, as they may impact the validity of the presented results for a robust flux analysis.

Major comments:

1. Section 2.2.1 (lines 154-171). The description of equation (1) is inconsistent with the labeling of the terms on line 169-171, or at least confusing to me.  The sum of all three terms on the right-hand is referred to as DD_chem. But if I understand this correctly, only the third term, k(omega), is the chemical loss.  The sum of all three equals  on the left hand.  Also I am not sure why you want to study DD, and DD_topo (which is the sum of the horizontal wind transport and vertical altitude related derivation), such as that shown in figures 4-9.  Why can’t you just show the separate impact of 1^st and 2^nd term, which gives a better sense of the relative importance of each factor? Looking at the figures, DD and DD_topo don’t look that different, which implies the 2^nd term is possibly not important at all?
2. In addition, on lines 246-248, the authors stated that chemical loss term, the third term, “was excluded from this study due to negligible contribution and poor fitting performance”. If the third term is negligible, and 2^nd term is very small (see above), that would imply inferred emissions is predominantly controlled by the horizontal transport term?   NH3 has an atmospheric chemical lifetime of a few hours to a few days, it is not convincing that atmospheric chemical loss does not play a role in the budget and inferred-emission analysis.  More robust analysis is needed to support the validity of these results.
3. Figure 11. The relative magnitude of source vs. sink in most regions don’t make sense to me.  As I mentioned in comment #2, if chemical term is negligible, the sink (which represents surface deposition) is also much smaller that the source term throughout the entire year and can be one magnitude smaller in some regions, shouldn’t that imply that the atmospheric budget of NH3 is not in balance and atmospheric NH3 abundance will grow rapidly?  The results presented in Figures 10 and 11 does not support the conclusion that chemical loss term being negligible.  Some processes, whether it is chemical loss or deposition, must be in play to create the observed the seasonal cycle shown in figure 10. Please clarify and provide adequate analysis result to support your conclusions.
4. The side-by-side comparison of IASI and CrIS related results. I strongly support the use of both datasets for this analysis as these datasets provide corroborative as well as complimentary information to study spatial and temporal variability of NH3 abundance and budget analysis.  As pointed out by the authors, the different overpass times by the two instruments provide unique information that can be exacted using the right analysis. With that said, I would like to encourage the authors put a bit more thought in when it would be necessary to show panels (or lines) from both instruments and when one dataset would be adequate to convey the scientific message.  For example, figures 4-9, I personally don’t see the value of showing IASI & CrIS panels side by side.  First, they look similar on a broader spatial scale perspective. Second, you don’t spend much effort discussing the differences between the two different datasets in the text.  Third, repeating IASI and CrIS panels in 6 figures (Figures #4-#9) took up too much space. Therefore, I would recommend showing just one instrument dataset and discuss the relative scientific points. If you prefer, you can include the other instrument result in supplementary material for completeness.
5. Figures 4-9. Here is my recommendation for authors to consider improving these figures for a more informative presentation: a) only show one dataset, either CrIS or IASI, b) show DD, the 2^nd term (which is DD_topo – DD) intead of DD_topo, c) add another panel for NH3 VCD.  It would be helpful to have the column abundance distribution information on the same figure to better relate with the emission, transport, source and sink information.