Tao et al. implemented hourly- or daily-varying emissions into the version 0 Multi‐Scale Infrastructure for Chemistry and Aerosols (MUSICA) model and evaluated the impact of 3 different time-varying emissions inventories. These included (1) hourly variation in all emissions, (2) hourly variation in NO only, (3) daily variation in NO only. These were compared against (1) a base case using a global emissions inventory, (2) a case using the US EPA NEI monthly average inventory over the continental US, (3) monthly NEI with NO emissions reduced by 30%, and (4) monthly NEI with VOC emissions reduced by 30%. The model runs are compared against TROPOMI NO2, HCHO, and CO data, as well as surface data. The latter has a heavy focus on 6 US cities (New York, Atlanta, Houston, Chicago, Denver, and Los Angeles). Implementing hourly emissions does have a clear impact on simulated NO2, O3, and CO; however, the authors show that doing so does not uniformly decrease bias compared to satellite or in situ observations. The authors explore how the impact of hourly emissions depends on time of day and day of week, as well as factors in the model driving regional differences in sensitivity to NO emissions. They conclude that changing from monthly to hourly emissions will have significant impacts on NO2 and O3 concentrations, even if the overall emitted mass is similar.

This work is technically sound. My main comments are around the context/motivation for the work and organization of the paper to improve clarity. I therefore recommend publication after those issues are addressed.

General issues

My first general comment is: what is the motivation or context for this work? Testing the impact of hourly-varying emissions seems like an intermediate step in a larger project. That's perfectly fine; it makes a lot of sense to publish the groundwork as its own paper to cite for later steps. However, it would be helpful to know whether there is a next step planned, and if so, what it is to put this work into context. Alternatively, if the case is that this was simply an effort to quantify the effect of hourly-varying emissions, then providing sufficient background to show that, while it may be expected that implementing these emissions would have a complicated effect on air quality simulations, the actual effect has not yet been quantified. Or, if it has been quantified but for other models, referencing that previous work and describing the motivation to test how MUSICA compares to those studies would be important context.

My second general comment is around the organization of the paper. The abstract focuses only on the impact of implementing hourly emissions. However, the case studies with 30% smaller NO and VOC emissions as part of the urban case studies are also a significant part of the results. It would help the reader if this part was introduced in the abstract, as well as if the last paragraph of the introduction was clearer that (a) the urban case studies included emission reductions scenarios and (b) why those are included or how they tie back into the hourly emissions test.

Minor comments

Lines 24-25: "We conclude that models used to infer NOx emissions from monthly mean concentrations may alias hourly emission variations into the inferred magnitude of emitted NO."  It's not clear to me what this means, particularly since the paper as written does not include any quantification of how inferred emissions would change. The results as given all focus on forward model results, rather than the inverse inference of emissions from concentrations (monthly mean or otherwise). Making the jump from forward effects to inverse effects is difficult to do mentally, as there's a lot of interaction between grid cells with transport and chemical nonlinearity that make this difficult to intuit. If you want to retain this conclusion, a new discussion section that walks the reader through your chain of reasoning would be necessary to support this statement. Since this comes back in the discussion & conclusions, see my final comment for a more specific suggestion there.

Lines 66-67: "We select July 2018 due to the availability of trace gas retrievals from the TROPOspheric Monitoring Instrument (TROPOMI) and concurrent field campaign observations." Given that the focus of the paper is on hourly emissions, why not select July 2024, when TEMPO was observing? What field campaign observations are available in 2018 that wouldn't be available during a TEMPO-observed time period? It's not clear that the SLAMS data was a short term set of observations, and there does not seem to be any other field data used in the paper. Also, why was a winter month not tested as well, given the strong impact of photolysis and meteorology on the results? (I presume the reason is that summer remains the season with the most O3 exceedences, but this should be stated explicitly.) This answer can likely tie back to my first general comment, i.e., how this study fits into a larger project, if it does.

Sect. 2.1: It would be easier to follow if this section began with a short paragraph introducing the reader to the experimental design, that CAMS will be used as the baseline global emissions, NEI for the more detailed CONUS emissions, and that different variations of NEI emissions will be used to test the model response to different emissions scenarios. The rest of the section is fine, adding just that extra context up front would help understand why the two different inventories are used.

Lines 143-144: "...as in NEI_monthly, but applying daily NO emissions with weekday-weekend differences (NEI_daily_NO)." So do NEI_hourly and NEI_hourly_NO not include weekend/weekday differences? Please be clear about which emissions scenarios include what temporal variation - it might be clearer to add columns for "temporal resolution" (e.g., monthly, daily, hourly) and "weekend/weekday differences" (yes or no) to Table 1.

Line 163: "We use measurements collected from State and Local Air Monitoring Stations (SLAMS)..." A map showing how many stations are available for each grid cell would help clarify how representative the SLAMS data is of the regions shown in Fig. 1. This is important for interpreting Fig. 2.

Lines 190-191: "...and apply the TROPOMI AKs, linearly interpolated vertically to the MUSICAv0 vertical resolution..." Unless the MUSICA profiles have significantly better vertical resolution than the TROPOMI AKs or the MUSICA profiles aren't reasonably smooth, it's usually better to interpolate to the AK levels, as this avoids interpolation issues around the sorts of sharp jumps in the AKs that occur when a pixel is partly cloudy. This is also something to use care with when averaging the AKs together to the grid cell scale. Since it isn't specified whether you filtered the TROPOMI data according to any QC variables, it's impossible to know whether clouds will impact the results. (So please also clarify what TROPOMI filtering criteria were used.) A supplemental figure comparing the averaged and interpolated AKs to the original AKs for a few different grid cells (perhaps one clear and one partly cloudy) would help give the reader confidence that the averaging of multiple pixels' AKs together to the grid cell level and interpolation to the MUSICA levels is not introducing unphysical smoothing of the AKs.

Lines 198-199: "Across the six CONUS regions, spatial correlations between modeled and observed surface concentrations are stronger for NO2 and O3..." For which model configuration(s) are these correlations calculated?

Lines 215-216: "CO concentrations increase relative to the BASE simulation..." Could point to Fig. 3 here - in general, if text is discussing multiple figures within the same paragraph, it is helpful to include liberal cross references to figures or tables to be clear which results are being discussed.

Line 223-224: "Compared to the BASE case, the NEI_monthly simulation worsens the model underestimates of NO2 VCDTrop by 16-21% (~1×1014 molecules/cm2)..." I've also seen in the past that models have lower background NO2 tropospheric columns that satellite observations, going back to OMI. This might be due to incorrect NO2:NO ratios in the free troposphere (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018GL077728) or the stratospheric/tropospheric separation in the satellite products. If you think your modeled stratosphere is accurate, you could see if the total NO2 columns have less bias - that would remove the strat/trop separation algorithm as a source of uncertainty (though it then adds in the stratospheric column as a new source). Otherwise this could be part of the discussion.

Lines 228-230: "These discrepancies reflect vertical distribution heterogeneity and diurnal variation, as column comparisons are for 1:30 p.m. local time, while surface comparisons include all times of the day." Not necessarily, see my previous comment.

Lines 244-245: "...changes in NO2 VCDTrop and CO VCDTotal largely mirror the surface patterns (Fig. 3b)." Fig. 3b shows NO2 and HCHO columns, not CO. Were CO columns supposed to be in that figure? Even if not, it would make sense to include them.

Fig. 1 caption: "July mean nitric oxide (NO) emissions from the adjusted 2017 National Emissions Inventory..." What does "adjusted" mean here? Is that just the weekday shift and regridding? If so, please call that "adjusted NEI" in sect. 2.1 so that readers know this is how you will refer to it elsewhere.

Fig. 2: If most of the SLAMS data are in the six urban areas shown in Fig. 1, is it really fair to label Fig. 2a with regions, or should panel a be labeled with the city names instead? This is where the map of SLAMS stations per grid cell would help.

Lines 310-313: "These results highlight the need for improved representation of diurnal cycles, particularly boundary-layer dynamics that strongly shape near-surface concentrations (Adams et al., 2023), as biases in daily range and peak timing increase uncertainty and limit the application of model simulations, particularly in the era of high-temporal-resolution observations from satellites and ground-based networks." I agree that PBL dynamics are a likely explanation, but they aren't the only one. Has there been any evaluation of the PBL height in the MUSICA dynamical core? (It's not clear to me if the Adams paper evaluated this.) Another direct source of error would be an incorrect diurnal profile in the NEI emissions - how is the diurnal cycle of emissions calculated, and how well has it been validated? What about representativeness error - Fig. 4 seems to be comparing a whole grid cell to a single SLAMS station in each case, do you know whether there are highly local effects that could be in the observation but not in the model grid cell?

Lines 338-339: "West-East Contrasts in Surface Pollutant Responses" - this sentence fragment looks like it was supposed to be a subsection title?

Lines 376-378: "Collectively, these processes shorten the NO2 lifetime against deposition in the eastern CONUS, contributing to greater NO2 accumulation in the western CONUS." Need to be careful about over-generalizing here, and differentiate between drivers that are consistent month to month or year to year (e.g., average VOC concentrations) and ones that vary with meteorology. Was there potentially some kind of blocking high or stagnation event during this time period that might be a factor in this west/east dipole?

Fig. 5 caption: please confirm that the PBLH is given in meters above ground level, not meters above sea level. I assume so from the magnitude, but it follows the topography well enough that this would be good to clarify.

Lines 433-434: "These MUSICAv0 simulations indicate that urban areas in both the western and eastern CONUS are generally NOx-saturated, with surface O3 increasing under reduced NO emissions (Fig. 3)." It's difficult to see urban/rural differences in Fig. 3; the O3 panels show decreases everywhere to my eye. Either zoomed-in maps on urban and surrounding areas or box plots that show the change in O3 for urban and rural grid cells separately would make this point clearer.

Lines 437-438: "...suggesting that these urban centers likely remain within a NOx-saturated regime but are approaching the transition toward NOx sensitivity (Fig. 7)." It's not clear to me how you come to this conclusion from Fig. 7. Are you making this argument because in LA the 30% reduction moves the high NOx points over toward the peak O3 production, so if further reductions occur, they should move into the NOx-limited regime? If so, please so that explicitly and explain how that applies to NYC, as the high-NOx points for NYC largely seem to stay at the same P(O3) level.

Fig 7: the green and blue are hard to distinguish; using different markers as well as colors may help.

Lines 454-455: "Using hourly NO emissions shifts the distribution and peak of P(O3) toward higher NOx concentrations in Los Angeles and toward slightly lower NOx in New York City, with more pronounced changes in Los Angeles (Fig. 7)." It's not clear from Fig. 7 that peak P(O3) does shift to higher NOx in LA - to me, it looks like the difference between the top left and top middle panels is that the 15-20 ppb NOx box actually has its P(O3) decrease slightly, with peak P(O3) still around the 5-10 ppb NOx range.

Lines 456-457: "...the NEI_hourly_NO simulation shifts the P(O3) distribution in Los Angeles toward higher NOx levels, indicating more strongly NOx-saturated conditions." This statement needs a bit more clarity: it looks like you're specifically talking about the long tail at the right end of the distribution, i.e., the hourly emission run has more points in the highest NOx concentration bins with correspondingly low P(O3). Since the immediately previous sentence was discussing peak P(O3), not the whole P(O3) distribution, it took me a few reads to catch the shift from discussing peak P(O3) to its general distribution.

Lines 459-460: "...the morning reduction dominates the monthly mean O3 change (Fig. 6b)." Please clarify - do you mean that the decrease between about 9 and 14 local for LA is big enough that it makes the overall change in the monthly O3 negative?

Lines 485-486: "Incorporating hourly NO emissions amplifies the spatial gradients in simulated surface NO2 concentrations between urban and rural areas..." As mentioned previously (in the comment for lines 433-434), the current set of figures don't provide a clear visualization of urban/rural differences.

Lines 505-507: "Our results demonstrate that hourly variations in emissions can produce nontrivial impacts on NO2 and O3 concentrations even without changes in the total emission magnitude. Neglecting such variability may introduce biases when interpreting monthly mean concentrations from once-daily satellite overpasses..." There is some missing model description and analysis that weakens this claim. First, I did not see where it was shown or started that the hourly NEI emissions emitted the same total mass of pollutants as the monthly average one, which is important for the reader to understand that the differences in the model outputs are solely due to emission timing. Second, a worked example of how the timing of emissions affects the relationship between total emissions and early afternoon columns would help make the point: something simple like the analysis in Sect. 3.2.4 of Lamsal et al. (2014, https://acp.copernicus.org/articles/14/11587/2014/), but using the NEI_hourly model simulation as the "observed" columns and the NEI_monthly as the "modeled" values to show the error in retrieved emissions would help solidify that argument.

General Comments

In “Contrasting Air Pollution Responses to Hourly Varying Anthropogenic NOx Emissions in the Contiguous United States”, Tao et al. address a timely scientific question regarding how the potential lack of temporal variability of emissions in global chemical transport models shapes the resulting concentrations or related metrics, such as vertical column densities or ozone production rates. The study is carefully designed to test the hypothesis that monthly, daily, or, especially, hourly variation in emissions over the U. S. alter the diurnal cycle and spatial distributions of the emitted species and other species formed from them. This work is timely especially on account of TEMPO, which will provide higher temporal resolution in remotely-sensed column densities that are frequently used for revising estimates of emissions. Doing so without accounting for hourly variation in emissions could embed errors in the emissions refinements that could have been addressed with existing estimates.

This manuscript describes reasonable differences for daily and monthly variation in emissions with reasonable explanations. The authors support their analysis throughout the work with excellent contextualization with other literature. The amount of explanation in the supplemental material is suitable. The changes in concentrations observed when incorporating hourly emissions have unexpected differences from the western to the eastern portion of the U.S., and the authors introduce regional and local data analysis to demonstrate and explain the potential causes.

My primary concern regards the implementation of hourly emissions. The supplemental information describes that a CMAQ-ready version of the NEI 2017 inventory was transformed for integration into MUSICAv0, which seems like an efficient approach. Yet, the unexpected patterns of increase in the west and decrease in the east made me curious. Figure 4a shows the change of concentration of CO (the species shown that is most like an emissions tracer) and, to a lesser extent NO2, when changing from monthly to hourly. An abrupt shift from negative to positive changes is evident with the demarcation of time zone from Eastern to Central. The shifts from Central to Mountain and Mountain to Pacific are not as prominent, but the deepening of the increase is suggestive of the offset in time contributing to the differences. Further investigation of other figures to understand whether a time shift in hourly emissions associated with time zone may be inducing these results was challenging to reconcile. For instance, if Figure 6a is at local time for LA and for NYC, it suggests that at midnight the hourly emissions of NO are typically above the monthly average in LA whereas in NYC the emissions are below the monthly average, which would be more reasonable to expect. In Figure 6b, the diel plots, which are defined as being in local time, do not have this offset in time.  Additionally, the shift in time of the productivity of ozone for LA but not NYC in Figure 7 contributes to the idea that an offset of three hours of emissions changes may be shaping these results. These patterns are indicative of the time zone not being handled correctly when hourly emissions were implemented.

I have hoped that this explanation for the temporal shifts is not the best one since the authors carefully show and explain the potential physical and chemical reasons for differences in these outcomes across the country. If the authors demonstrate the accuracy of the hourly emissions implementation, I would be excited to revisit this substantial contribution that demonstrates unexpectedly different chemical responses between the western and eastern U.S. to the incorporation of hourly emissions.