General Comments

This study contributes to the fields of atmospheric and magnetospheric science by providing observational evidence that motivates improved proxies for EEP in global atmospheric models as well as obtaining more resolved spatial observations of energetic electron precipitation (EEP). This paper will likely be cited frequently to justify future observations and modeling studies involving energetic electron precipitation. This work compares radiometer measurements at Syowa Station from 2012-2017 with WACCM model simulations. The goal is to better understand the connections between NO concentrations in the upper atmosphere to geomagnetic activity and associated electron precipitation. The long-term, continuous, radiometer dataset and WACCM model (WACCM6 with meteorological reanalysis, ionospheric chemistry, ApEEP for medium-energy electrons, and Fang 2010 ionization) specifically highlights the role of medium energy electrons on both short and long-term NO variability in the mesosphere and upper thermosphere.

Strengths

This paper confirms that WACCM captures the observed year-to-year and seasonal variability of NO. The paper links day-to-day variability with geomagnetic indices and demonstrates the dominance of electron forcing over atmospheric dynamics in variability of NO during polar winter. (Section 3.4 on the polar vortex is useful for demonstrating that dynamical causes in day-to-day variability are likely not as significant as EEP.)

Another valuable conclusion is that WACCM underestimates NO column density in winter and does not adequately capture day-to-day variability. This most likely results from the statistically smoothed ApEEP proxy model for electron flux and demonstrates the need for proxies that include better representation of peaks in electron precipitation.  Consequently, this study also motivates the need for more spatially and temporally resolved observations of energetic electron flux.

This paper also presents simulations driven by a series of events based on Arase measurements, providing an example of the role of future observations in improving estimates of electron flux driving the modeled atmospheric ionization.

Major Recommendations

The paper would benefit from additional discussion of the following topics:

1. The spatial extent and duration of EEP events. Describe what is known (and what is not known) about the spatial extent and durations of EEP (MEE) events. How does this spatial scale compare with timescales of zonal mixing from localized EEP events at the Syowa latitude? How much of the WACCM underestimate in day-to-day variability is consistent with zonal and latitudinal mixing of sporadic precipitation events? Will improving the day-to-day variability also improve the discrepancies in 31-day averages (I assume this is the implication, but it is never explicitly addressed)? The paper alludes to atmospheric dynamic mixing in lines 347-350…but a more detailed analysis and more discussion in the context of this paper’s conclusions would be useful (referencing, for example, discussions of MLT dependence in Verronen et al., 2020).

2. Compare the ApEEP statistical proxy used in WACCM to other datasets of EEP. It would be useful to briefly discuss known weaknesses of ApEEP, for example as summarized in Nesse Tyssøy et al. (2021) "HEPPA III intercomparison experiment on electron precipitation impacts: 1. Estimated ionization rates during a geomagnetic active period in April 2010."  https://doi-org.unh.idm.oclc.org/10.1029/2021JA029128

The ApEEP model only takes into account the 0 degree MEPED telescope from the POES satellites and is known to underestimate electron flux. There have been efforts to include data from both 0 and 90 degree telescopes to produce electron precipitation maps that would be good to reference, such as Pettit et al. (2021),  "A new MEPED‐based precipitating electron data set", https://doi-org.unh.idm.oclc.org/10.1029/2021JA029667 

How might using other indices (Ap, Dst, AE) improve model results?  (It’s my understanding that there is also van de Kamp et al. Dst proxy similar to ApEEP.) What is the value of including higher energy electrons in WACCM, such as electron precipitation from EMIC waves that can reach lower altitudes?  See Capannolo et al. (2023), "Electron precipitation observed by ELFIN using proton precipitation as a proxy for electromagnetic ion cyclotron (EMIC) waves" https://doi-org.unh.idm.oclc.org/10.1029/2023GL103519. And Capannolo et al. (2019) "Direct observation of subrelativistic electron precipitation potentially driven by EMIC waves". https://doi-org.unh.idm.oclc.org/10.1029/2019GL084202.

3. Strengthen Discussion and Conclusions sections. Provide more detail and insights into what is needed for future studies as a result of this study. For example, how can results shown in Figure 7 be used to improve electron precipitation estimates used to drive WACCM (currently based on Ap)?  Are there examples of the “stochastic approach” recommended in lines 368-370? More discussion of how this study motivates next steps would be compelling, such as whether conclusions are consistent with recommendations in Sinnhuber et al. (2021) as well as articles such as Pettit et al. (2023), "Investigation of the drivers and atmospheric impacts of energetic electron precipitation. Frontiers in Astronomy and Space Sciences" https://doi.org/10.3389/fspas.2023.1162564. Adding a few additional sentences in the Conclusion to place the list of specific outcomes in context with other studies and promote future work would greatly enhance the impact of the paper.

Minor Recommendations

Line 40 “Ground-based radiometers provide a regional view on [of] NO variability.”  Recommend explaining how localized measurement from a radiometer can be viewed as regional. Line 76 states, “The horizontal size of the observe area is estimated to be ~2 km at an altitude of 100 km”.  I assume the regional aspect comes from the continuous measurements as winds transport enhanced NO over the site?   

Line 97-99. Simplify (or split) the sentence. For example: “This analysis uses WACCM data co-located with Syowa Station to compare daily-averaged NO column density. Global model data are also used to locate the polar vortex.”

Lines 107-110. Is there a way to be more quantitative about how the 0-10 pitch angle observations from Arase map to the bounce loss cone at the top of the atmosphere?  (I’m surprised it gives such good results and isn’t a huge overestimate with the mirrored particles).

Lines 111-112. How might the BERI (Boulder Electron Radiation to Ionization) model affect ionization rates?  (If it might be significant, recommend adding a reference to let readers know this ionization scheme is also available).

Lines 135-136.   Does the 27% of “observed magnetic variability” refer to the slope? If yes, why is the slope used instead of the coefficient of determination (R2 = 0.42) to compare the variability between model and observations?

Lines 137-138.  Why is there a lower bound to NO column densities in WACCM?  (Also suggest finding a better word than “saturate”)

Lines 194-196. Simplify sentence. For example, “However, here we use geomagnetic indices to relate EPP events with geomagnetic disturbances.”

Lines 194-197. Recommend briefly describing the difference in Ap, Dst, and AE indices and why one might expect EEP to behave differently with each.  This is done for Dst, but more explanation would be useful for what each index means with respect to magnetospheric disturbances that lead to electron precipitation.

Lines 265 – 274. Are the Arase events used for the electron forcing of WACCM all at similar L-shells as Syowa Station (as in Figure 8?) How many hours MLT does Arase travel through during the 12-hour averaging period? Could the radiometer be detecting peaks of local MEE events that both Arase and ApEEP smooth out because of zonal mixing?

Lines 350-351. What does this sentence mean?  Doesn’t the choice of proxy for EEP determine the magnitude of forcing?

Lines 366-368  Clarify this sentence.

Figures and Tables

Figures 3 and 4. Recommend adding legends to the contour colors (the annotated labels are too small to read).

Table 1. Recommend re-labeling “Peak NO” to “Peak NO difference” or “Peak ∆NO” in column labels. Even though this is explained in the caption, it would be easy for the reader to mistake the NO column as representing densities during the peak instead of differences.

Figure 9. Recommend changing the vertical axis scale to make the daily variability easier to see. Is there a need to include values less than zero? Or could those be omitted (and noted in caption) to help with the visual comparison?

General comments:

This study examines the variability of nitric oxide (NO) in the polar mesosphere–lower thermosphere region (MLT) using long-term ground-based observations from Syowa Station from 2012 to 2017. It combines high-latitude NO column density observations with WACCM output to evaluate the model’s ability to capture both long- and short-term variability. The topic is interesting and highlights the role of considering energetic electron precipitation (EEP), the polar vortex, and medium-energy electron forcing in NO variability. Furthermore, the use of electron flux data from the Arase satellite adds further strength to the study. It demonstrates how future observations could improve the representation of atmospheric impacts in models.

The manuscript is well structured and the text is well written. However, some clarifications would enhance its readiness for publication.

Specific comments:

1. Lines 231–234: "Comparing the distribution... between Ap and WACCM NO." The authors show clear differences in the correlation coefficient (r) distribution between NO and geomagnetic indices, suggesting that daily NO variability is more strongly linked to Dst and AE than to Ap. However, at Line 350, they state that "the choice of proxy for EEP seems to be of lesser importance than having better accuracy in the magnitude of forcing." Please clarify.

2. Line 372: 'Overall agreement on .....27\% of the observed variability.' There is an agreement in trends but not magnitude. Please, write more carefully.

3. In discussion section, authors mentioned that L338'Comparing these observations with WACCM results,...... differences between model data and observations. However, I think it is important the authors to expand further this section, providing a more comprehensive discussion of their findings related to the key discrepancies, their causes, implications for model improvement and future work.

Technical corrections:

Define key terms at the beginning of the paper and use them consistently throughout, such as WACCM, CO. Define MSISE.

Line 42: "Antartic" to 'Antarctic'

Figure 4: 'WACCM time series of 2012–2017' to 'WACCM time series from 2012 to 2017' and 'NO colum density' to 'NO column density'

Add color bars (legends) to each plot in Figures 3 and 4

Line 15: 'the model captures only 27\% of the measured magnitude in the day-to-day variability.' -> 'the model captures only 27\% of the observed day-to-day variability.'

Line 400: 'Nagyoa' to 'Nagoya'

This study looks at the changes in nitric oxide density in the MLT region over a years-long time duration and compares WACCM simulation results with the observations to understand the model’s accuracy for capturing the NO variability. The results show that EPP forcing is driving much of the day-to-day variability, during which time the NO changes correlate well with geomagnetic indices. Using Arase data, the study also looks at the usefulness of providing inputs from satellite observations. This is a very timely study in that the research on EPP-produced NOx is gaining traction and the use of new data sets like Arase should be explored more.  

Moderate comments: 

Line 133: Is the scatter plot showing data for all times, summer and winter? If not - if you break it up seasonally is there a different dependence?

Lines 136-139: Is it true that the lower values “saturate”? Perhaps this is a physical lower limit on the densities. Comparing these to the simulations, which show negative lower values, is not reasonable since the negative values are certainly unphysical. Unless there is a reason you think the measurement saturates at this lower value, perhaps explain that it’s unclear what causes this lower value of 0.25e15 - perhaps it is measurement capability but perhaps it is due to physical processes. 

Lines 145-146: In Figure 3b,d, I don’t see where the 50% contour ever gets down to 94 km altitude. Is it perhaps a mistake in the lower altitude boundary range? There is no 50 contour exactly to see for sure, but extrapolating between 40% and 60%, I don’t see the “50%” ever getting to 94 km. It seems more like 100 km to me. Can you check the data please? Or explain better, since I might be missing something. 

Lines 149-150: It’s not obvious to me where the SEPs are in the plot. Can you describe them more or somehow point them out? 

Line 225: Can you explain why the correlation might be stronger with AE (in the Discussion)? Also please here explain the two different indices (how they are calculated and what they are a proxy for). 

Minor comments: 

Line 103: “Van” Allen should be capitalized

Line 151: “ration” —> “ratio”

Line 161: Do you mean figure 3b? 

Line 289: “exist” —> “exists”