This work significantly contributes to understanding the influence of EPP on D-region ionization. The considerations of the seasons and the high latitudes in both hemispheres deliver new insights and raise questions for further investigations.

While the idea and approach of this study are very promising, the text is often imprecise and incomprehensible (e.g. the description of the two step smoothing is to general, lines 169-176).

The equations do not have a consistent notation (e.g. in equation (4) it is z instead of h').

The structure of some figures and switching between kev and MeV (1500 keV and 1 MeV) is confusing (e.g. order of the panels in Figure 2 and Figure 3), and the text information in the figures is not always readable due to a too-small font (Figure 6).

Sometimes, there are just text segments (instead of complete sentences) or text duplications, giving an impression of inaccuracy. 

Overview:

This paper presents observations from the CSES satellite that combine ELF wave, x-ray, and particle observations to infer the change in the D-region reflection height during solar minimum. Results show a change in the D-region reflection height during energetic electron precipitation events, as expected. The paper shows that the change in height is greater during winter, also as expected. Notably, the results are limited to nighttime, although all latitudes are covered. 

Assessment:

The combined observations of ELF (to infer the reflection height directly), along with x-ray and particle measurements, is a unique dataset and therefore a unique analysis. The results in this paper are not surprising, but they do provide a valuable dataset and analysis of EPP and its impact on the lower ionosphere. 

Compared to a previous version of this paper submitted to GRL, this paper provides further analysis, explanation of methods, and figures that explain the data and model outputs. The new figures provide insight into the measurement of the D-region reflection height, fitting the modeled ionospheric profiles to determine h’, and insight into modeled electron density profiles. Overall, the paper presents valuable contributions to precipitation science and is worthy of publication after revisions. There are two significant issues that need to be addressed (the same issues that I identified in the GRL review), and a few minor issues.

Major Comments:

1. The assessment of reflection height is improved from the previous version of this paper, but it still appears to be incorrect. The measurement of the first cutoff frequency altitude, as shown in Figure 1, is sufficient. But the model-derived reflection height (section 2.4) is problematic. The model provides an electron density (as in Figure 2); the authors then fit a Wait and Spies profile (equation 2) to the model D-region profile. They extract h’, the Wait parameter, and use that as the reflection height. But, as I noted in the GRL review, h’ is not the reflection height; it is merely a reference height for the electron density profile. The reflection height is correctly described in the paper as the height where X = Z in the Ratcliffe formulation, or wp^2 = w*nu (equation 3, except a factor of 2*pi is missing). 

The next steps in the paper are confusing. The authors use the first waveguide cutoff frequency from the ELF data, plug that into equation 3, and then solve for h’... But they already assumed a reflection height from the cutoff frequency through f = c/h’ for the cutoff frequency. This circular logic is confusing, and it is not clear (and not stated) which h’ is used in the rest of the paper. 

The assessment of the reflection height from the model can be made much simpler. Use the electron density from the model to calculate wp^2. Use equation 4 to calculate nu, the collision frequency profile. Now, you can plot X/Z as a function of altitude (on the y-axis) and frequency (on the x-axis), just as Ratcliffe did in his book. For the frequency of interest (say, the f1 determined from the ELF data), you can find the altitude where that frequency reflects. 

The authors should still point out that for ELF/VLF frequencies, this is not a hard reflection; it occurs over a range of 5-10 km in altitude. But it is true that a lower effective reflection height is consistent with precipitation, as shown in Marshall and Cully [2020]. 

2. X-ray observations: This comment is repeated from my GRL review, since I do not see that it was addressed. X-ray observations are shown, which give a spatial distribution similar to the particle fluxes. But apart from a spatial distribution, and a general amplitude correlation (not shown; a figure would be nice). Furthermore, the energy range of the X-ray observations is concerning. These energies (0.9--35 keV) are going to be dominated by bremsstrahlung from auroral electrons (<50 keV), NOT the relativistic electrons usually associated with EPP. Yet, it is intriguing that the X-ray fluxes seem to peak around L=5, clearly sub-auroral and in the radiation belts. Did the authors filter the data to remove times of auroral precipitation somehow? If so, this should be explained. Then, what were the fluxes of lower-energy precipitation electrons at this time? If lower-energy precipitation (<50 keV) is present, at high fluxes as is typical, then the x-rays may have been produced by those electrons, are we would not expect them to be strongly correlated with the electron fluxes or the reflection height (since those electrons deposit energy above the ELF/VLF reflection heights). But all of this information is missing, so cannot be properly assessed.

Minor comments:

Figure 2: These plots should be extended down to 40 km or so to show where the new ionization profile blends back into the background profile. 

Also, it’s not clear why the authors show both 1500 keV and 1 MeV, and in that order. Those energies seem too close to show any meaningful difference. Why not 100 keV, 300 keV, and 1 MeV? 

The results in Table 1 are a nice addition to the paper, but the authors should explain what “Moran’s I” means to those who are unfamiliar with that method.