This manuscript investigates the interannual sensitivity of particulate nitrate formation to nitrogen oxide and ammonia emissions during wintertime over the Midwestern US during the period 2007-2023.

To do so, the authors utilize the chemistry transport model GEOS-Chem for simulating particulate nitrate formation in response to a fixed decrease of precursor emissions and deduce PN sensitivities. Satellite observations of NH₃ and NO₂ columns were also exploited to quantitatively define regime cutoffs. The investigation is also supported by ground-based measurements of gas concentrations, wet deposition and particle speciation.

Overall, I find that the manuscript highlights important findings regarding potential emission controls that can be implemented by policy makers in the agricultural intensive area of MWUS to efficiently mitigate winter PM pollution episodes. While the results are well communicated, I have major comments specifically in respect to the methodology.

I understand that the approach is similar to the one described in Dang et al., 2023. However, I am missing a basic explanation of how the species-sensitivities (from GEOS-Chem) are exploited to derive the regime cutoffs and if they are totally independent of the satellite observations. I am wondering if a schematic could help to clarify. In addition, I would suggest that the title also includes GEOS-chem or a reference to the simulated sensitivities.

In Fig. 3 I think the captions need to be revised and more precise regarding which data we are looking at (i.e., satellite columns/ simulated surface concentrations?).

I am not sure of the structure of the methodology and results, section 2.4 from Methods is called the same as 3.1 from results. Wouldn't it be more logical to also have the regime cutoffs definitions in the methodology?

While in Dang et al., 2023 we can clearly distinguish the 3 dominant regimes of PN sensitivity to precursors from the RMA, it is not obvious in the present work (Fig.3 a)) since NH₃-sensitive and NOx-sensitive grid-cells overlap across the regression line. Any comments on that? Is it then really the best approach to use these equations for defining the regimes?

More specific comments:

1. l.133: Did the vertical profiles of the simulated GEOS-Chem NH₃ columns also corrected with the IASI averaging kernels (newly given in the v4 version, Clarisse et al., 2023)? It has been demonstrated that this refinement can imply important changes in the simulated columns from LMDZINCA (Kumar et al., 2025 https://acp.copernicus.org/articles/25/12379/2025/).
2. 142: caption of the Table 1 is a bit confusing and could rather be “general model set up”. I think the chemistry row is not useful since it is indicated already in the version of the model.
3. 163: are these differences mainly due to agricultural sector? Any comments on the abrupt decrease happening in 2013 in the NEI2016 inventory?
4. 166: Which period are the soil Nox and BVOC emissions? Where do soil Nox come from?
5. 178: For how long were the sensitivity simulations running after spin-up?
6. 181: Table 2, instead of “normal”, I would replace by “Ref” and encourage to mention the total regional emission quantities for the baseline.
7. 195: How was the spatial and temporal matching between ground-based measurement and satellite done?
8. 201: Please define j.
9. 243: it is not clear to me what is shown in Fig. S3. We can clearly see a large decrease in NH₃ emissions from the NEI inventory in 2013. It does not seem to be revealed by observations. How can this be explained?
10. 249: Fig S6 and S9 need clarification. Is it averaged 2007-2023 or only year 2007 that is shown?
11. 265: It would be interesting to know more precisely what kind of agricultural emissions are dominant in winter in this region (livestock, fertilizer)?
12. 269: the caption of 3.2 section could be more precise. Implications for total particulate matter?
13. 276: I do not see where NH₄⁺ contribution is shown in the mentioned figures.
14. 314: I found the limitations very well described and I wonder if N emissions should not be shown on a seasonal basis to put in context the specific context of the study. I would guess that NH₃ emissions in winter are relatively low in the region compared to summer.

Review for “The Changing Sensitivity of Wintertime Particulate Nitrate to Precursor Emissions Diagnosed via Satellite Observations of Ammonia and Nitrogen Dioxide over the Midwestern United States”

General Comments

This manuscript investigates multiyear (2007-2023) changes in wintertime nitrate aerosol formation sensitivity over the Midwestern United States (MWUS), using satellite-derived NH3 and NO2 column observations together with GEOS-Chem sensitivity simulations. The authors conclude that winter nitrate formation has shifted toward increasingly NOx-sensitive conditions and that NOx emission reductions would be chemically effective for mitigating winter PM2.5.

The topic is important and highly relevant to ACP’s scope, particularly given the policy implications for secondary inorganic aerosol control in agricultural regions. However, the central conclusions rely on a column-based diagnostic framework whose physical and chemical representativeness for winter boundary layer nitrate formation is insufficiently demonstrated. In addition, mechanistic attribution is incomplete, and the statistical robustness of the inferred trends is not convincingly established.

Major Comments

1. Statistical robustness and trend attribution

The reported NO2 trend is statistically indistinguishable from zero. Given the magnitude of retrieval uncertainties over the 2007-2023 period, and strong meteorological control of winter boundary layer height and removal of nitrate and its gaseous precursors through snowfall, the manuscript does not convincingly separate emission-driven trends from meteorologically induced variability. Attribution of long-term chemical shifts to precursor emission trends requires the application of deweathering techniques or model-based methods that explicitly separate emission effects from meteorological variability. Such analysis is not presented. Consequently, the claim that increasing NOx sensitivity is driven by relatively flat NO2 and increasing NH3 columns is not statistically robust.

2. Lack of explicit thermodynamic regime analysis

Wintertime nitrate formation is strongly governed by thermodynamic partitioning. In agricultural regions of the MWUS, conditions are frequently ammonia-rich, implying that nitrate mass is limited primarily by HNO3 production rather than NH3 availability. Additional NH3 increases may not significantly change nitrate mass until stoichiometric thresholds are crossed. The manuscript does not include explicit thermodynamic diagnostics to determine whether the system is ammonia-limited, nitric acid-limited, or in transition. Instead, regime classification is inferred from column abundance ratios. Without demonstrating shifts across stoichiometric boundaries, the conclusion that sensitivity has transitioned toward NOx control lacks mechanistic grounding. 

3. Incomplete treatment of winter heterogeneous chemistry

Winter nitrate production in the MWUS is dominated by nocturnal chemistry involving NO3 radical formation, N2O5 hydrolysis, and strong dependence on aerosol liquid water, particle surface area, and chloride content. The manuscript attributes the diagnosed sensitivity shift primarily to flat NO2 column trends combined with increasing NH3 columns. This interpretation overlooks that NO2 column trends do not directly reflect HNO3 production trends. Changes in aerosol liquid water and heterogeneous uptake coefficients can strongly modulate nitrate production efficiency. Increasing NH3 may enhance aerosol water content, indirectly accelerating heterogeneous pathways. The GEOS-Chem simulations are not evaluated with respect to trends in N2O5 uptake, aerosol liquid water, or nighttime chemistry diagnostics. As a result, the mechanistic explanation for the inferred sensitivity shift remains incomplete.

4. Representativeness of satellite NH3/NO2 columns for surface nitrate sensitivity

The study’s central diagnostic is the tropospheric column NH3/NO2 ratio derived from satellites. This ratio is used to infer surface nitrate sensitivity regimes. However, winter nitrate formation in the MWUS occurs predominantly within shallow boundary layers characterized by strong vertical gradients in NO2, rapid NH3 deposition and near-surface concentration maxima, and frequent boundary layer decoupling. Tropospheric columns integrate free-tropospheric contributions that are chemically decoupled from surface thermodynamic partitioning of ammonium nitrate. The manuscript does not provide quantitative analysis demonstrating that column NH3/NO2 ratios correlate with surface HNO3 availability, nitrate mass, or model-diagnosed sensitivity regimes in winter. A column-based proxy must be explicitly validated under the meteorological and chemical conditions of interest. Validations, such as comparison with surface observations, vertical profile diagnostics, or model-based representativeness metrics, are currently lacking.

Minor Comments

Model evaluation against long-term surface nitrate measurements should be expanded and clearly quantified, including bias, trend agreement, and seasonal cycle fidelity. Uncertainty propagation from satellite retrievals into regime classification should be explicitly quantified.

The manuscript reports a sensitivity analysis of particulate nitrate, a major PM2.5 component in the target area, in the agricultural region of the Midwest US. The authors use a state-of-art chemistry-transport model to assess what are the sources most effective to be reduced for the abatament of PM2.5 concetrations. The task is pursued alternatively altering in the simulation the local emissions of the most relevant precursors. The method may be suitable to make a first assessment, but I have some points to be addressed before considering the manuscript for publication.

The main point is that the works lacks an adequate validation on purpose of the model used to make the conclusions. These are my comments/suggestions:

- l. 32: Actually, Figure S1 shows that for all the period shown, starting in 2007, PN has always been higher than PS. Suggest to rephrase accordingly.

- section 2.1.1: Would be useful to add the link to the web sites from where the OMI and IASI data were downloaded.

- section 2.3: Also here the reference to the web sites from where the data were downloaded would be useful.

- section 3: Since the basic tool used to diagnose the relative sensitivity of PN to NOx, NH3 and VOC is the GEOS-Chem model, a section displaying the validation of the model against available measurements is needed. Please add this section before section 3.1, concluding summarizing skills and limitations of the model. This would be useful to make sure that a realistic representation of main processes simulating nitrate formation are represented, in terms of sources multiphase chemistry, mixing and deposition. As for example illustrated in a paper also on ACP some years ago, the production of nitrate in the boundary layer may be the result of the interplay of these different processes (https://acp.copernicus.org/articles/15/2629/2015/)

- l. 260: I believe the authors intended to reference Figure S3, not S10. Please check.

- l. 317: The fact that the region became more NOx-sensitive is expected, because local NOx emissions significantly decreased thus making NOx the limiting factor for PN formation. I would add that further abatement of NOx emissions in the region is however expected to become gradually less effective, because the control of the regional NOx concentration would further shift to sources outside the boundaries.