Review of “Global perspectives on nitrate aerosol dynamics: a comprehensive sensitivity analysis” by Milousis et al.

Nitrate aerosols have recently emerged as a more dominant component of atmospheric composition than sulfate aerosols, yet accurately simulating them remains a significant challenge. This study employs the EMAC climate-chemistry model in combination with the ISORROPIA II thermodynamic module to investigate key factors influencing nitrate aerosol formation, particularly in highly polluted regions. The manuscript presents a comprehensive analysis of sensitivity simulations, which will be of strong interest to the atmospheric chemistry modeling community. However, before I can recommend this manuscript for publication, several issues need to be addressed to improve clarity and readability for the broader audience. These are outlined below:

L35 – “Sensitivity tests show ...”, Please clarify this sentence. What sensitivity tests do you mean? Please be more specific!

L230 – “even at low relative humidities”, Is it true for lower humidities than CRH?

L230 – “The first cast is used …”, I wonder why the authors chose this option. As far as I understand, many climate and air quality models employ a “metastable assumption” for the phase of aerosols in thermodynamic calculations.

Section 2.2 – A significant part is focused on comparing simulated PM1 and PM2.5 nitrate with the observations. Therefore, please elaborate on how the simulated PM1 and PM2.5 nitrate are calculated in the model for broader readers.

L304 – “PM2.5 accounts for more than …”, Please revise this sentence for clarity, such that “PM2.5 nitrate accounts for … 80% of the total nitrate concentrations”

L332 – A scatter plot is of particular interest for readers to understand the model performance, so it is worthwhile to be included in the main text.

L353-354 – “While the model …., the mean bias and .. be relatively unaffected”, What do you mean by this?

L431 – Section title is misleading. Please revise it to embrace the contents appropriately.

L465 – In summary, the lower resolution model simulates higher PM2.5 and PM1 nitrate than the base model. A brief summary would be highly appreciated, with a likely cause for this change, here and elsewhere in this section.

L495 – Using CMIP6 vs. HTAP in the model shows differences in simulated nitrate concentrations. I wonder what drives this change. It may be a bit difficult to examine year-by-year emissions, but authors can provide an insight into the apparent differences in simulated nitrate by looking at NOx and NH3 emissions from the two inventories.

Figures 5 and 6 – The model appears to have a smaller bias for PM1 nitrate than PM2.5 nitrate compared to the observations. Could you provide a determining factor for this?

L575 – Well, the difference could be minor in nitrate mass concentrations, but it could be very large and important in terms of aerosol optical depths because of hygroscopic growth. Could you comment on this?  

L637 – Can authors make a statement about a recommended uptake coefficient of N2O5 in the model from the analysis using HYDRO results? Or at least possible causes for the different performance of the model with different values depending on regions.

L653 – I think that “SCAV” results in the underprediction of winter observation by the model.  

L668 – Even in East Asia, the “SCAV” model appears to be better than the base for spring and autumn, as shown in Figure 5.

L762 – Capturing diurnal variation of nitrate aerosol can be associated with hourly emission of NOx and PBL variation, which is much more complicated issue. I would recommend that the authors omit this from the text. If they want to keep this, please include discussion of those two factors in the text.

L901 – I do not think that Table 5 is necessary because the text already discusses nitrate budgets in detail, along with Figure 9.  

Review of the manuscript of egusphere-2025-313 titled “Global Perspectives on Nitrate Aerosol Dynamics: A Comprehensive Sensitivity Analysis” written by Milousis et al.

This study provides valuable insights into nitrate formation influenced by factors such as grid resolution, emission inventories, chemical & physical processes in the CTM. A substantial amount of effort appears to have been devoted to this study, reflecting the complexity and comprehensiveness of the simulation performed. However, the authors have some issues that need to be clarified in the manuscript for publication in ACP. The following are my concerns.

Major comments

1. Despite conducting extensive simulations, the study lacks in-depth scientific discussions. The manuscript currently provides limited explanation of which factors are most influential and which are less so, and it lacks sufficient interpretation of the findings in the context of other previous studies. The results appear as a list of outcomes without a scientific interpretation, making it difficult to understand the key message. I recommend that the authors consider re-analyzing the results, revising the manuscript accordingly, and potentially resubmitting to the journal after further development.
2. Evaluating the emissions of precursors: Although the manuscript emphasizes the importance of evaluating precursor emissions (lines 66-68), which I agree with, this aspect is not addressed in the main analysis.
3. Regarding the second comment, without evaluating the emission fluxes of the precursors (e.g., NOx and NH3), the subsequent sensitivity tests may lack meaningful context or scientific validity. Also, the results from each sensitivity case vary considerably, making it difficult to determine which factors are truly important and how much they contribute to nitrate concentrations. Although the conclusion section presents possible region- and species-specific optimal conditions, the criteria used to reach these final conditions are not clearly explained, making them appear somewhat arbitrary.
4. Nevertheless, since the stated goal is to identify the most relevant parameters for each region, if multiple key parameters are selected, it would be necessary to test and validate an integrated set of conditions. Without such a verification process, it is difficult to assess the robustness and practical relevance of the proposed conclusions.
5. Annual variations of emissions (lines 203 & 265): It is unclear how the annual variation in emissions was considered in the simulations spanning 2008 to 2018. The methodology section should clearly describe how year-to-year changes in emissions were handled, as this is critical for interpreting long-term trends and sensitivities.
6. Emissions of precursors (lines 278 & 495): The manuscript should explicitly discuss the spatiotemporal variations of NOx and NH3 emissions, at least for these key precursors that strongly influence nitrate concentrations, based on CMIP and HTAP datasets. Relevant figures illustrating these variations should be included in the Supplementary Materials.
7. Lack of validation in model performance: The manuscript lacks an overall evaluation of the model’s performance. A proper assessment of key species such as PM2.5, O3, and major precursors (particularly NOx and NH3) should be conducted as a prerequisite for interpreting the simulation results (Notably, observational data for NH3 are available from established monitoring networks such as AMoN in the US and the AMoN-China in China). Currently, the performance evaluation is limited to nitrate, which is insufficient to establish the reliability of the model output.
8. The GEIA NH3 emission inventory used in the study is outdated (Line 206). Even if it is not feasible to incorporate a more recent inventory into the model simulation, it is important to compare GEIA emissions with those from other available inventories to assess potential biases. Such a comparison would provide valuable context for interpreting the model results and understanding the limitations of the emission inputs.
9. Lines 149-151 & 293. The hydrolysis of N2O5 is known to be highly dependent on environmental factors such as temperature and relative humidity. However, in the current sensitivity test, a fixed uptake coefficient (γ_N2O5 = 0.002) is applied without any justification.
10. Lines 436-437. Please explain why the coarse grid resolution results in higher nitrate concentrations compared to the base case.
11. Lines 661-664. I don’t agree with this statement. If the base case incorporates CMIP6 or HTAP emission inventories with simplified scavenging considerations, the model’s performance is likely to degrade. Additionally, the assumption of a pH of 5 in the model should be justified, as it may not be appropriate without any reason.
12. Lines 719-712. If the increase in evaporation is contributing to the observed data, this factor should be reflected in the analyses presented in Figure 2, Figure 5, and Table 2. It is not just an issue limited to the observations, and therefore, failing to account for this in the analysis creates a logical inconsistency.

Minor comments

1. Line 55: The manuscript refers to "recent decades," but the supporting references cited are outdated.
2. Change ‘Europe’ to ‘Europe (EMEP)’, ‘East Asia’ to ‘East Asia (EANET)’ in Figs 5, 6, and 7.
3. Figures 5-8. It is recommended to use a consistent y-axis scale for each site. This would be allowed for a more balanced and accurate comparison, as the current use of varying axis ranges may lead to over- or underestimation of the magnitude and variability in some cases.
4. Line 296 or Table 1. Consistent alignment in Table 1.
5. Lines 259-260. Please specify the period for which the modeling was conducted.
6. Line 299. size fraction --> fraction
7. Line 395. the observed underprediction --> the underprediction
8. Line 398-399. urban downwind values --> "urban or downwind values" or if both, "urban and downwind values"
9. Line 449. The statement is incorrect. The lower-resolution model run actually outperforms the base case, as demonstrated for winter and autumn in East Asia.
10. Line 504. The statement of "For EPA observations, this case underestimates in all seasons except winter" is wrong. For EPA observations, this case shows underestimates in all seasons. There is no exception.
11. Lines 506-507, 611-613, 652-653, 666-668, & 686-687. These statements are not clear or correct. Those need to be checked.
12. Line 705. What is the reason for the largest discrepancy occurring in March in the EMEP data?
13. Please combine Fig. 9 and Table 5, as they are redundant.