This manuscript describes a microwave radiometer that has been measuring ozone profiles from Seoul since 2016.  The authors do a nice job addressing details of the measurements that are extremely important in obtaining accurate retrievals (e.g. tropospheric opacity and pointing).  Understanding such details is important for any group attempting to make such measurements.  I thank the authors for addressing my concerns in the initial review.

Figure 8, line 287, and conclusion – Are the tropospheric opacities shown in Figure 8 for KLAPS and SORAS calculated at precisely the same time?  It is not clear whether the higher summer opacity of the KLAPS results come from periods when SORAS opacity measurements are not possible because of the high opacity.  In the conclusion the authors say that the O3 bias relative to MLS is different during the summer than during other seasons, but in order to establish this a clear plot showing the difference in tropospheric opacity biases is needed. 

Paragraph starting on line 382 and conclusion - While the authors are correct in attributing a portion of the observed negative bias to the use of an AC240 spectrometer, this is not thought to cause an altitude dependence in the bias.  Just as for the instrument presented here, the Sauvegeat study does show an altitude dependence in the bias between that ground-based instrument and MLS.  But, as for the instrument discussed in the Sauvegeat study, the altitude dependence of that bias is only weakly, if at all, caused by the AC240 (their Table II, last column).  This should be noted in the text.  

Lines 305 and 409 – What velocity does a 55 kHz doppler shift imply?  Is this physically realistic?  Is the 55 kHz smaller than the expected error of the local oscillators used in the instrument? 

Line 343 – ‘bacause’ should be ‘because’

The authors introduce the South Korean millimeter wave radiometer SORAS (Stratospheric Ozone RAdiometer in Seoul), designed to monitor the ozone emission line at 110.8 GHz. In its original version from 2008 SORAS has been a double sideband radiometer. In 2016 it has undergone a transition into a single sideband radiometer. The changes made to the original SORAS are unclear as the three references to earlier studies with SORAS are published in Korean language, indecipherable to me (Line 59). However, the manuscript presents measurements and results of the post-2016 period, which makes references to the older studies between 2009 and 2014 not necessarily needed.

Today SORAS is one of the two Asian radiometers currently in operation monitoring stratospheric ozone, which makes it a relevant instrument, even though neither technique, idea nor measurement processes are new. However, in SORAS there has been implemented the concept of direct amplification together with a high pass filter blocking the contribution from the lower (image) sideband, which avoids complications due to a quasi-optical single-sideband filter.

The description of the instrument, the calibration method and the data acquisition are described well, probably even too much in detail given the large number of references where these equations are presented. However, the methodology is sound and has proven successful with other radiometers as well. The authors address certain instrumental imperfections and their corrections.

The language is fluent and the authors give proper credit to related work.

The abstract provides a quite concise and complete summary and the overall presentation is well structured and clear.

In the end the authors present results (Fig. 13) which show a stable and reliable observation period over roughly six years which makes SORAS a valuable source of ozone data from a region that has rather small contribution to the overall knowledge of the state of the ozone layer.

In order to support and strengthen their results the authors validated their data with the ‘gold standard’ of atmospheric remote sensing, MLS. This comparison shows a nice agreement between data from SORAS and MLS, after MLS data altitude resolution has been adopted to SORAS with the help of SORAS’ averaging kernels. A slight bias is attributed to the AC240 spectrometer.

The number and quality of references are quite appropriate. I would suggest adding the Parrish et al. (1988) as a very basic reference to ground-based microwave radiometry.

I have not seen any supplementary material.

Specific questions and comments:

• Line 138: The authors write ‘As the tropospheric ozone is present in very small amounts, its contribution to 𝑇b can be negligible.’ Can the ozon below 20 km really be neglected when the AKs for the 20 km altitude in Fig. 12 shows substantial sensitivity even below 20 km and the SORAS altitude range is estimated to 16 to 70 km. Elaborate!
• From Fig. 2 and paragraph 3.2.1: I would suggest the authors use more common symbols for high pass and band pass filter in the block diagram and explain the symbol with the two arrows of which the lower one is crossed. I apologize if this is just my limited knowledge that lead to this comment but at least for the filters I am used to present the symbols as shown in the attached pdf document . (https://www.hobbyprojects.com/general_theory/filters.html)
  
  I would also avoid the mentioning of upper and lower sideband, USB and LSB, in the sketch as output from the mixer, since the mixer output inherently usually contains both sidebands, even if the RF signal INTO the mixer might have a suppressed unwanted sideband. I would suggest writing ‘IF’ for the intermediate frequency as the output of the mixers instead.
  
  Has the effect of Styrofoam on the measurements been investigated under different weather conditions and over a longer period? Does it take into consideration that water vapor could penetrate into the material? If so, looking through the Styrofoam at different angles might lead to a varying contribution to the measured brightness temperature.
• Line 94: Frequencies lower than 100 GHz are cut off …
• Line 98: The baseband converter with a LO frequency of 2.0 GHz produces an output of 0.609 GHz +/- 415 MHz I assume. What are the numbers in the sketch (1070~1900 MHz) telling the reader?
• Line 101 What does the sentence mean that starts with ‘As a result …’? Does it state the NOMINAL frequency range of the baseband converter should be 110.327 to 111.157 GHz, BUT it turned out that the ACTUAL frequency range of the baseband converter is somewhat shifted? And thus, the spectrum started at higher channels within the FFTS? Even a shift of 100 MHz should keep the entire spectrum well inside the FFTS bandwidth of 1 GHz. For reasons of clarity I would suggest presenting one example spectrum as measured by the 1 GHz bandwidth FFTS.
• 6 (Right) I would suggest ‘Sun azimuth angle’ for clarity reasons. How was the zenith of the sun calculated? As the sun was rather low above the horizon, was the light refracting property of the atmosphere included into the calculations? The sun might have appeared higher than it actually was and the angle correction needed would be even larger.
• Line 130 I suggest to mention the condition for which the Rayleigh-Jeans approximation can be used: h ν ≪ k B T
• Line 140, Eqs. (5-7): It seems unfortunate to me that the authors use the variables ‘a’ (italic a) and ‘a’(italic alpha) for air mass and absorption coefficient respectively in the same paragraph 3.1. The italic ‘a’ is hardly distinguishable from the Greek ‘alpha’. I suggest using ‘A’ as variable for the air mass, as for instance Parrish et al. (1988).
• Line 143 As a matter of taste, I leave it to the authors to decide whether the issue of the air mass factor has to be explained in such detail or rather referred to any of the references.
• Line 184 and Eq. (12): How confident are the air temperature measurements on the roof of the building? Solar radiation at low latitudes can lead to increased air temperature due to convection on the roof, with rather strong diurnal variation. Other sources, such as Parrish et al., (1988) assume 7K as a reasonable temperature correction, so 14.9 K seems a rather large number, even if Ingold et al. suggest 10 - 20 K in Ingold.
• Line 256: which observation platform?
• Line 295 and paragraph 4.1 and Fig. 10 : This paragraph is of larger concern and the problem of frequency fluctuations needs more attention. The average 55 kHz frequency difference between the measured center frequency and the one in the JPL catalogue is a large deviation from the catalogue value. Moreover the spread in frequency offset (± 200 kHz) as shown in Fig. 10 is surprisingly large and cannot by any means be explained by Doppler effect. Already a 50 kHz offset would require wind speed of roughly 500 km/h in the stratosphere. In Fig 10 there are deviations of hundreds of kHz from the catalogue value, leading to 3000 km/h for an offset of 300 kHz.
  
  For me this rather looks like a sign of an instable local oscillator somewhere in the three down conversion processes. There seems to be a lot of discrete ‘levels’ of the offsets, which might point to an oscillator that switches between discrete frequencies which then might have multiplied by a certain factor. I would probably try to produce a histogram of all the offsets and check whether this observation can be correct. If my suspicion is correct this would have consequences for adding up spectra at a certain zenith angle over a longer period as is implemented in the measurement scheme.
  
  Anyway, I would like the authors to elaborate more on the large spread of the frequency offsets.

References

1. Ingold et al., 1998. Your reference.
2. Parrish, R. L. deZafra, P.M. Solomon, and J. W. Barrett , A ground-based technique for millimeter wave spectroscopic observations of stratospheric trace constituents, Radio Science, Volume 23, Number 2, Pages 106-118, March-April 1988