The authors present an approach to use ionization rates derived from electron fluxes calculated by the eVlasiator magnetospheric model in the WACCM atmospheric model for to calculate the response of atmospheric nitrogen oxides and ozone to auroral electron forcing. Traditionally, ionization rates for such atmospheric modelling are derived in proxy-based parameterization so I think that this new approach would allow for a more physical input. As a disclaimer, I don’t feel competent to judge the magnetospheric modeling part of the study, but feel more at home with the atmospheric modelling part. However, I see two major issues that I’d like the authors to consider before a publication can be considered regarding a) the motivation for this development, and b) the evaluation of the approach.

a) In the Introduction the authors state that “accurate modelling of the MLTI is required to […] further our knowledge of the region” and that the complex dynamics lead to “great uncertainties in our understanding of the region”. Furthermore it is stated that the lack of knowledge about “auroral electron forcing in climate models could […] obstruct accurate evaluation of polar climate variability”. While this may all be true, it is also very vague, and therefore it is difficult to judge whether the proposed new approach could help to generate new knowledge. What specifically is not known and why do the authors think that their approach can help to fill the knowledge gap? The abstract emphasizes the variability of the electron forcing on the hourly time scale. This point is not addressed in the manuscript anymore, but if it is the ability of eVlasiator to cover such time scales I think it would be necessary to argue why this is important for the stated purposes, e.g., the representation of polar climate variability. The WACCM results show that the eVlasiator electron fluxes lead to quite similar responses of atmospheric NOx and O3 as provided by the traditional Kp-based parameterization. What is the added value of the new approach? If a proper hypothesis about the benefits of the new approach is provided in the introduction, it can be tested and discussed in the manuscript. Instead, in the discussion, the authors put forward the “enhanced information on energy and spatial distribution” provided by the new approach. But if the expected benefits of this are not discussed it is difficult to judge whether the new approach is useful or not. The Discussion mentions that the approach would enable future “near-real-time predictions of the atmospheric response”. Isn’t this already possible with the existing parameterizations and the available Kp predictions?

b) Somewhat related to point a), I’m missing an evaluation of the electron fluxes (or ionization rates) calculated by eVlasiator. This point is related to a) because if the goal is unclear, it is also unclear how to assess whether the presented new approach is useful to reach this goal. Regarding the atmospheric response, I think that the comparison of NOx responses produced by the new and an old approach, as provided in Section 3, is sufficient for the current study. However, the manuscript leaves me uncertain as to whether eVlasiator in its current form provides any useful information. Again, I’m happy to admit that this may be due to my limited knowledge of magnetospheric processes and modeling, but other potential readers from the atmospheric community may share this knowledge gap. In principle I don’t see a problem in scaling the eVlasiator output with satellite observations. However, from Figures B1 to B3 in the Appendix, one might get the impression, that the eVlasiator output doesn’t provide any similarity to the observations. Often, the observations and the model output differ by factors of several orders of magnitude. Is there any correlation between model and data? If I understand correctly, the magnetospheric simulation is run for fixed boundary conditions that represent a situation similar to that on the two days of the satellite overpasses. Why not check the quality of the scaling by comparing the scaled data with independent observations from a third day with similar conditions. Or if that is not possible, consider using only one of the days for scaling and checkif the observations on the 2^nd day are realistically reproduced. There may be many other ways to evaluate the approach, but without any evaluation I find it difficult to assess the usefulness of the approach.

In the following I will list further minor comments:

L26: “Hence, … complex dynamics between the neutral atmosphere and the electromagnetic ionosphere”. Maybe it is just the wording, but what is meant, here?  “Complex interactions? And why “hence”?

L39: “in long-term atmospheric and climate simulations although uncertainties are still present in the latter”. Do you want to say that only climate and not atmospheric simulations are uncertain?

L41: “its impact on the stratospheric ozone balance is to a larger extent affected by polar atmosphere dynamics and is not fully understood”. Almost nothing in our field of research is “fully understood”. I think this is generally a poor motivation for a study. Please try to be specific on the knowledge gaps.

Fig. 1 caption: “simulation of the 3D-3V magnetospheric simulation”?

Fig1: It’s a beautiful figure, but it is hard to extract information from it. What is the color scale of panel a)? Also the grey grids are hard to identify.

Fig. 4: I'd prefer using the same size for all panels. With panel a) larger than the others I find it hard to identify differences, e.g., in "sharpness".

Section 3.1.2, comparison of ionization rates from the parameterization and from VLAS: While I understand that there are differences I’d like to understand which of them are fundamental to the approaches. If, e.g., the missing “sharpness” of the oval is considered a weakness of the parameterization, is this fundamental or couldn’t the parameterization be modified accordingly?

L294: “realistic auroral electron precipitation fluxes from eVlasiator”. This relates to my main point b). “Realistic” by which metric?

L303 “Our results also indicate that the current parameterization of aurora may be overestimating the auroral forcing”. Yes, may be, but later it is stated that “eVlasiator […] underestimated the total precipitating fluxes. Is there a reason to believe that the first effect dominates?

L337: “for the detailed study of solar wind – atmosphere interaction”. Besides the main issue a) which applies also here, because I find this very vague, I’d also suggest to reconsider the use of the word “interaction”, which I usually understand to work in two ways. However the coupling of the magnetospheric and atmospheric models suggested here is clearly one-way, so it would only allow to study effects of auroral electron forcing on the atmosphere. I know that publications in the larger field often use the term “solar-terrestrial interactions” or similar, but I’d prefer an unambiguous use of the language.

Review of “Atmospheric odd nitrogen response to electron forcing from a 6D magnetospheric hybrid-kinetic simulation” by Hakkila et al.

This study applies the WACCM model to simulate the effects of energetic electron precipitation in the Earth's thermosphere, mesosphere, and upper stratosphere, focusing on the impact on NOx and O3 levels. The study analyzes model runs with electron precipitation derived from magnetospheric hybrid-kinetic simulations using the combined Vlasiator and eVlasiator model framework. Additionally, comparisons are made with WACCM results incorporating simplified, nominal auroral precipitation maps parameterized by the geomagnetic Kp index (ranging from 0 to 5, with 0 indicating no auroral inputs). The findings indicate that auroral electron precipitation significantly enhances NOx concentrations during the polar winter, while its impact on upper stratospheric O3 is negligible. This study is very interesting, however, the following concerns need to be addressed.

My biggest concern is that the electron precipitation estimated using Vlasiator-eVlasiator to drive the WACCM model seems to be fixed, not only spatially (precisely, in the geomagnetic coordinate system) but also temporally. This simplification significantly diminishes the potential benefits of using the sophisticated hybrid-kinetic simulation of particle precipitation, as compared to the simpler parameterization using the Kp index. Apart from a brief comparison of ion production rates in Figure 5, the study lacks detailed discussions that would adequately justify the preference for hybrid-kinetic simulations. To convincingly demonstrate the necessity and value of these complex and computationally intensive simulations, more comparisons and in-depth analyses should be conducted.

Clarifications regarding the Vlasiator-eVlasiator simulation are needed.

• The proton-electron mass ratio is inconsistent between line 97 and line 115.
• Lines 110-111: The upstream solar wind speed and electron temperature are significantly higher than typical values, while the solar wind density is considerably lower. Note that the obtained simulation results are used to drive the WACCM model over an extended period to represent typical precipitation conditions near Kp=1-2. Therefore, these settings are highly inappropriate.
• Does eqn (1) specify the precipitating flux along the local magnetic field direction? Are magnetic mirroring effects considered?
• In Vlasiator, are electrons treated as a massless fluid?
• What are the boundary conditions near 4.8 RE? Do the models account for varying ionospheric conductances?
• The dates used for eVlasiator-DMSP calibration in line 135 are not consistent with the date used for magnetic field line mapping in line 356.
• Under the assumption of the non-tilted dipole magnetic field, the north-south asymmetry (in MLT-MLAT) is generally neglected, as seen in Figure 3. Explicitly clarify this in the text.
• I don’t understand how the calibration after the DMSP data along specific orbits can be useful, in physics. In particular, note that the calibration is temporally independent, while the model-data disagreement varies with time. Therefore, I suspect that even after applying the proposed calibration as in Figure B5, the ratio plots (similar to panels e-f in Figure B1 and Figure B2) will still show discrepancies of orders of magnitude.

There are a few issues in atmospheric ionization calculations.

• Line 176, the parameterization method of electron impact ionization is not applicable to incident energies below 100 eV. Nevertheless, it is adequate to consider only the incident energy range of 100 eV to 50 keV in this study, as <100 eV electrons do not penetrate below 140 km. See Fig 6.10 in Fang [2022], “Chapter 6.2—Fast calculation of particle impact ionization from precipitating energetic electrons and protons in the earth’s Atmosphere” (https://doi.org/10.1016/B978-0-12-821366-7.00005-6)
• Line 177, the use of the NRLMSIS model for the ionization calculation introduces inconsistency with the WACCM-specified neutral profiles, especially for energetic electrons penetrating below 140 km. Are the resulting ionization rates above 140 km, as shown in Figure 2b, disregarded in the WACCM runs? If so, why not use the WACCM atmosphere directly to calculate the ionization?

Other minor comments:

Line 13, and throughout the paper, delete the space between numerical values and the percentage symbol (%).

Line 19, delete “unique”, or change to “particular”

Line 30, change “solar energetic radiation” to “solar radiation”

Line 36, “There are three primary EPP sources of NOx”

Lines 36-37, change “solar protons” to “solar energetic protons”

In addition, what about “solar energetic electrons”?

Lines 42-43, rephrase the sentence. Auroral precipitation is a continuous phenomenon that occurs not only during substorms and has sources beyond just the magnetotail.

Line 70, briefly explain “6D” (space and velocity space)

Line 74, what does “Cartesian 2D” refer to?

Line 81, what does “fields” refer to? It can easily be confused with “magnetic field” and “electric field”.

Line 86, the use of “at full strength” is not appropriate.

Figure 1a, what does the color represent?

Figures 1c-1e, what are the relative locations of the three points? Are they along one specific open magnetic field line? Why are the velocity space plots not organized in terms of parallel and perpendicular velocities? What findings are drawn from the comparison among these three plots?

Line 128, swap the order of the two processes to align with the two items listed earlier in line 127.

Line 167, the sentence is confusing. What is the relationship between the NOEM-specified NO and the precipitating electron induced NOx  at 140 km altitude? Is there any inconsistency here?

Lines 171-172, what is the relationship between “particle impact ionization” and “dissociative ionization”? What is “secondary electron dissociation”?

Line 175, delete “temporal”

Line 185, briefly specify the recommended forcing conditions so that readers do not need to refer to the reference to understand the driving conditions.

Line 195, I don’t understand this sentence.

Line 197, delete “and no Kp driven parameterized aurora”, which is redundant

Line 197, it is my understanding that the REF run excludes auroral precipitation, but still includes SEP impact, according to line 263?

Line 206, how is the Kp index used to drive the VLAS run?

Line 220, I understand that these ionization rates are used to drive the WACCM model, not “from” the model.

Figure 4, I don’t see how altitude-integrated ionization rates can be useful. As dissociative recombination rates are altitude dependent, the efficiency of ionization in converting into ion/electron density increase also varies with altitude. This makes the integration of the ionization rate over altitude not meaningful.  

Line 233, what does it mean by “the lower boundary of the parameterization”?

Line 237, the polar cap excludes the auroral oval. You may want to change “polar cap averaged” to “polar averaged”

Line 269, where are the “troughs”? I cannot find them in Fig 7d.

Line 283, change “showing” to “due to”

Line 293, change to “our results demonstrate the coupling between the magnetosphere and the atmosphere through electron precipitation”, or something similar.

Line 303, change “at least” to “likely”

Line 307, insert “be” prior to “considered”

Line 351, change “seed points” to “start points”? The word “seed” implies sources.

Figure A1 caption, what are the thoughts behind the use of “7.5 RE”? In addition, change “Cartesian in MLAT-MLT” to “regularly spaced in MLAT-MLT”.

Figure B1 caption, “DMSP/SSJ (contour lines)”

Line 379, briefly explain why the high-energy component is missing

Line 399, delete “el cm^-2 s-1 sr^-1 eV^-1”? I think this is the ratio, not flux, that you are talking about.

Line 404 and throughout the paper, change “quantile” to “percentile”

Review of “Atmospheric odd nitrogen response to electron forcing from a 6D magnetospheric hybrid-kinetic simulation” by Häkkilä, T. et al.

The manuscript presents a novel approach to combine magnetospheric modelling with whole-atmosphere modelling to investigate the auroral production of odd nitrogen. The magnetospheric model drives the precipitating electrons, and the whole-atmosphere model WACCM is used for the production of the odd nitrogen species, NO in this case.

The study is well thought out and an interesting approach that could shed more light on current discrepancies between measurements and current climate model simulations of the lower thermosphere. Thus it would be a worthy contribution for Ann. Geophys. However, I could recommend publication after minor revision of the manuscript. I feel some of the arguments need to be improved and made more clear, some of the issues might be caused by the language used and might be resolved with a slight rewrite or rephrasing.

General comments

1.  I understand that for numerical reasons the electron mass is different from the real value. However, this changes also the (kinetic) energy for the same speed, or the speed for the same kinetic energy. The energy and energy flux affect the precipitation characteristics, and the velocity is part of the Vlasov equation.  How are these corrected back to realistic values corresponding to real electrons? Apparently the mass used is roughly 46 times the real electron mass, which would results in almost a factor of 7 for the velocities. Although [1] briefly discusses the impact of the different mass, the main arguments and justification could be stated here to justify the choice.

In addition, the changed electron mass is given relative to the proton mass, I suggest the authors also state it relative to the real electron mass, maybe denoted by m_e0 or similar, like it is done in [1]. I think it would make it clearer that it is the electron mass that is scaled and not the proton mass.

Also, currently there are two values listed, 5.11 MeV, citing [1], and what amounts to roughly 23.46 MeV a bit further down in the text. Which one was used? I suggest to remove the other to avoid confusion.

2.  I see that the authors use classical quantities for the kinetic energy (after Eq. (1)). [1] lists the range of velocities up to 42000 km/s per dimension, which leads to a maximum speed of about 73000 km/s, or roughly 25% light speed. How large is the error introduced here by using the classical approximation instead of the relativistic description? Why not use relativistic quantities throughout? At least some of the impacts and reasons should be discussed briefly (best with a reference).

A related question is why Vlasiator uses velocity and not the momentum to model the phase space more directly? One could then use for example  the Liouville theorem to constrain the solutions further. I am not suggesting to rewrite Vlasiator for the current study, but the authors and model developers might want to think about that for future developments.

3.  Please list the absolute changes in addition to the relative ones, e.g. as 200% (from x to y). That would make it easier to judge if the change is from 1 to 2 (or 3?) molecules, or from 10⁸ to 3 × 10⁸.  If not in the abstract, then at least in the results and conclusions sections.

4.  I understand that maybe due to limited resources, only single WACCM runs have been performed and not ensemble runs. However, I did not find that mentioned anywhere in the manuscript. Please add some words of discussion, and maybe add it to the “next steps”, as only this would allow to assess the quality of the resulting NOx values.

Specific comments

L. 13: See general comment #3, absolute changes would help.

L. 22–23: “The polar MLTI depends on solar radiation” is a bit vague.  Please revise to make it clearer to what aspect of the MLTI the authors refer to.

L. 48: “… simplistic realization …” only applies to the CMIP6 setup.  There are other parametrizations available (e.g. [2]), which have been compared to each other and to measurements (e.g. [3]; [4]). However, so far not explicitly focussing on the auroral part.

L. 75–76: In other words, the “6D” space is what is also known as the “velocity phase space”.

L. 81: What are the “fields” here? Is this solution obtained iteratively? Please be more specific.

L. 84: What is “the ion inertial length”, can the authors give specific numbers here?

L. 86: Which magnetic field model was used?

L. 95: What are “connected field lines”? Aren’t field lines “connected” by definition? Maybe just “along field lines” is enough.

L. 96: See general comment #1, please state the actual value used in the study and explain why it was chosen.

L. 100: Please add ℱ_e to the text: “… number flux ℱ_e at position r and energy E is given by …”

L. 104: Which quantities are averaged?r From reading Eq. (1), this would be θ and ϕ. If this is correct, please add it to the description.

L. 110: How typical are these conditions? Was the driving force applied constantly over the whole simulation period, or intermittently to simulate reality? Please explain the rationale for choosing those numbers.

L. 115: Here, suddenly m_p/m_e is different from what was described before, please make these numbers consistent.

L. 129: “… fluxes … are lower than …” is difficult to assess without knowing the true flux. I suggest to use a more careful wording: “… fluxes … might differ from reality.”

L. 131–132: In my opinion, the authors are stretching the notion of “calibration” here. Typically instruments are calibrated by comparing to a measurement standard to adjust the scale over a wide range of values.  Here, only two days with the same conditions are compared. A few questions arise, first, are these typical conditions, in the sense that they occur very often? Second, how would this comparison hold up against a third day with similar conditions, and third, how would it hold up in a comparison with different driving conditions? Ideally, a “calibration” would have to be carried out based on the average fluxes (from SSJ or any other source) during different driving conditions. Unless this is fixed, I suggest to use a better terminology, maybe “adjustment” instead of “calibration”. At least another comparison on a different day should be checked if the author’s adjustment holds up.

L. 145–149: At the first reading, this part was hard to understand, it only became clear after reading the appendix. It is not fully correct either, the “distributions” are nowhere shown, only the median and the percentiles are shown in the appendix. Please revise for readability, and include all the necessary information, the median/percentile of what exactly? “2nd and 3rd order polynomial” could be stated explicitly instead of vaguely referring to “a polynomial function”. Does the “integrated energy flux” take into account the different electron masses between SSJ and eVlasiator? See also general comment #1. And what is finally done with the adjusted ratio, are the eVlasiator fluxes simply multiplied/divided by that factor? Please be more specific about how the final values are obtained.

L. 167: What altitude is “above its altitude range”? How can WACCM account for any effect above the altitude range? Although NOEM is derived for up to 200 km, the main NO distribution is centred around 105–110 km. How is this combined with the auroral forcing, both from the “normal” 2 keV “Kp-model” and the eVlasiator input to avoid double counting? Please be more specific about how the individual sources are combined.

L. 175: See general comment #1, does this electron energy take into account the changed electron mass? How is it converted to the real electron mass used and energy used as input for WACCM?

L. 176: The parametrization in [5] was derived for an electron energy range from 100 eV to 1 MeV, How did the authors extend that range to lower energies of 50 eV? Is the energy grid in log-space, i.e. in log(energy)?

L. 177: Why was NRLMSIS-00 used for the ionization rates, and not the WACCM atmosphere? NRLMSIS-00 provides a good climatology, but here it might be a good idea to use WACCM itself, especially in terms of consistency. This choice needs to be motivated better, especially since CMIP6 is not part of the presented paper, and the method described in [6] and [7] is used for medium-energy electrons.

L. 186: “WACCM-D runs from … 2005 to … 2006” is confusing and needs more explanation. The study compares the impact of events in 2011 and 2015, yet the WACCM runs are for an entirely different time period. What is the purpose then, and how do the authors reconcile consistency?

L. 190: “turns off Kp …”, what about the NOEM input which is also aurora related? Is it switched off as well for the “no-aurora” reference run?

L. 197: “… no Kp-driven … aurora” depends on if NOEM has been switched off or not, otherwise there will be remnants of Kp-driven aurora via NOEM.

L. 203: For completely “no aurora”, NOEM should be switched off too. It is not clear from the text.

L. 206: Why is Kp 2 needed for eVlasiator? Shouldn’t it drive all the aurora itself? What Kp-level would correspond to the solar wind conditions used in the Vlasiator simulations? Table 1: The difference between REF and Kp0 could be made a little clearer.

L. 214: Does the “precipitating electron integrated energy flux” account for the heavier electron mass?

L. 222: Please state the altitude range of the vertical integration.

L. 223: The described features are very difficult to see with the chosen colourmap. See also below my comment and suggestion for this figure.

L. 227: The limited magnetospheric domain only explains the hard cutoff at the lower latitudes. The VLAS ionization rates also do not reach as high latitudes as the KP2 run, between 10⁻⁴ and 10⁻⁵ hPa. What would be the reason for this?

L. 233: I would assume that the “lower boundary” is because of the fixed 2 keV energy for the Kp-WACCM, and probably the limited altitude range used to derive it.

L. 237: Has the “polar cap average” been calculated with area weighting (cos (lat) weighting)? Please state it if so, otherwise not weighting correctly will over-emphasize the low values at high latitudes.

L. 240: See general comment #3, please add the absolute changes for reference.

L. 249: See comment above, why is Kp needed for the Vlasiator run?

L. 254: As the authors point out, there was a NH SSW happening in 2006, which makes it even more confusing why the authors have chosen 2005 and 2006 for the WACCM simulations, and not 2011 or 2015 as for the comparison with SSJ.

L. 258: The authors could show the polar-cap average time-altitude distribution in both hemispheres showing the downward transport for comparison with earlier studies.

L. 261: I could not find any profiles presented as line plots. From the maps it is difficult to figure out what the authors are discussing here.  Please add at least one figure actually showing profiles for a visual comparison.

L. 268: How would “solar irradiation” work in polar winter?

L. 271: “… adds to the effect.” What effect? It is unclear what the authors are trying to point out here.

L. 277–278: “higher REF levels of NOx”, higher than what? Please clarify.

L. 280: “difference in the REF background levels”, difference between what? Please clarify.

L. 293–298: I suggest to move these sentences to the conclusions since they provide more of a summary than a discussion.

L. 303: Add “in WACCM”: “current parametrization of aurora in WACCM …”.  How do the authors arrive at the conclusion that the “standard” auroral forcing overestimates it? Overestimation compared to what? So far, no comparison to NOx measurements is presented.

L. 310: What makes the authors so sure that they are “underestimated”?

L. 323–325: See general comment #3, please provide absolute changes as well.

L. 333: I suggest to replace “made” by “carried out“.

L. 336: I suggest to replace “makes way” by “paves the way”.

L. 350: I might be nitpicking here, but since a field line is an integral curve of the vector field (axial vector field in the case of B⃗), it is already “traced”. Thus “tracing a field line” seems tautological, and “tracing the field” or “field lines are calculated” would already be unambiguous enough.

L. 351: I suggest to use the term “initial point” instead of “seed point”, as I assume that is what the authors are describing.

L. 359: “follow the geomagnetic field” is unambiguous enough. But again, what magnetic field model was employed here?

L. 365: “Better self-consistency” feels a bit strange, better in what sense? In my opinion, things either are consistent or they are not.

L. 365: I suggest to remove “lines” and “file” here, it is not clear what this “file” is and how it can be obtained.

L. 368: It might just be a poor choice of words, but to my understanding a field line can only end on a surface, not in free space except when the (magnetic) field becomes exactly zero. Please revise for clarity.

L. 369: I suggest to remove “therefore”.

L. 385: The distinction could also refer to NH and SH instead of dayside and nightside, since the dayside passes are in the NH, and the nightside passes are in the SH. Only an additional comparison during NH nighttime and SH daytime would allow to use one or the other.

L. 388–391: A figure marking the selections for Oct 2015 is missing, in addition to Figs. B1–B3.

L. 392–393: This sentence essentially repeats the statement above it and can be removed.

L. 395–408 (Appendix B3): This appendix describes a method that is not used in the end and should be removed to reduce confusion, including Fig B4. A (more extensive) comment about the issue with mean/median can be included in appendix B4. However, a better description is needed about over which dimension (of Fig. B3e,f) the quantiles are calculated.  I can only assume that the authors mean “along the DMSP orbit”, or equally “time”. If that is correct, please state it in the text to avoid too much guessing there. Since the percentile is presented on a log-log plot, how do the mean and median of the log-log behave? Are the distributions still skewed as in the linear case? Are the correction factors applied to the eVlasiator fluxes or to their logarithm?

L. 410–414: This description is very hard to understand, please revise.  The 0.9 quantile of what exactly, and why should its ratio be 1?

L. 411: “Insufficient” for what?

L. 412: Referring to my earlier remarks, does this comparison of (integrated) energy fluxes account for the different electron masses used in eVlasiator and DMSP/SSJ?

L. 410–419 and main text, either replace “quantile” with “percentile” or adapt the numbers to real quantiles between 0 and 1 (divide by 100).

L. 418: What is the final ratio of the integrated energy fluxes?

L. 422, Figs. B4, B5: The authors fit a third-order polynomial in log(energy), based on an energy range from 50 eV to about 5 keV (hard to identify in the figures). However, the “final” energy range is from 50 eV to 50 keV, and the 3rd order polynomial reaches a correction factor of about 10⁵ already at 10 keV. Is there any “safety” mechanism that keeps the correction factor in a sensible range? Using an empirical model (here the polynomial) outside the parameter range used for fitting is dangerous, unless the authors can ensure that their model is valid outside that range.

Figures

Fig. 4: I suggest to find a better colourmap, the current one makes it very difficult to distinguish VLAS from KP0–KP2. Maybe overlay contour lines from VLAS over all panels for comparison.

Fig. 5: A suggestion would be to show a geomagnetically averaged zonal mean instead of a fixed longitude. For direct comparison, one could add the contours from one panel to the other.

Fig. 6: Please replace the bright yellow line by a different colour, it is very difficult to see on white paper. Please add the altitude ranges in to the panels or their titles so that they are easier to identify.  The authors might consider removing the KP5 line so that the y-axis scale can be reduced, making the other lines easier to distinguish.  Again, showing only relative differences is only part of the story, an additional comparison of absolute values would be helpful. And if the authors can find, real NO data would be very interesting to compare to as well.

Fig. 7: Please revise the colourmap of the panels (a)-(f), the features described in the text are difficult to see in the figure. In particular the black contour lines and labels are unreadable.

Fig. 8: See comment about Fig. 6 and replace the bright yellow line by a different colour. Similar, reducing the Kp range to 0–2, the y-axis scale can be adapted to better distinguish the other lines. Also, please add the absolute changes for reference. Please label the panels clearly with “SH” for (a) and “NH” for (b) to make that easier to find.

Fig. B5: Instead of only showing a line for the selected percentile, the authors could add contours for the underlying distributions, including additional lines for the mean and the median, with an emphasis on the percentile line that was used in the end. This would make it clear how skewed these distributions are, since the authors claim that using the mean or median are not good choices.

References

[1] Alho, M. et al. (2022), Electron Signatures of Reconnection in a Global eVlasiator Simulation, Geophys. Res. Lett., 49(14).  doi:10.1029/2022gl098329.

[2] Wissing, J.M. and Kallenrode, M.-B. (2009), Atmospheric Ionization Module Osnabrück (AIMOS): A 3-D model to determine atmospheric ionization by energetic charged particles from different populations, J.  Geophys. Res. Space Phys., 114(A6), A06104. doi:10.1029/2008ja013884.

[3] Tyssøy, H.N. et al. (2021), HEPPA III Intercomparison Experiment on Electron Precipitation Impacts: 1. Estimated Ionization Rates During a Geomagnetic Active Period in April 2010, J. Geophys. Res. Space Phys., 126, e2021JA029128. doi:10.1029/2021ja029128.

[4] Sinnhuber, M. et al. (2021), Heppa III Intercomparison Experiment on Electron Precipitation Impacts: 2. Model-Measurement Intercomparison of Nitric Oxide (NO) During a Geomagnetic Storm in April 2010, J. Geophys.  Res. Space Phys., 126, e2021JA029466. doi:10.1029/2021ja029466.

[5] Fang, X. et al. (2010), Parameterization of monoenergetic electron impact ionization, Geophys. Res. Lett., 37(22), L22106.  doi:10.1029/2010gl045406.

[6] Kamp, M. van de et al. (2016), A model providing long-term data sets of energetic electron precipitation during geomagnetic storms, J.  Geophys. Res. Atmos., 121(20), 12, 520–12, 540.  doi:10.1002/2015jd024212.

[7] Matthes, K. et al. (2017), Solar forcing for CMIP6 (v3.2), Geosci.  Model Dev., 10, 2247–2302. doi:10.5194/gmd-10-2247-2017.