The manuscript entitled: “Observation of horizontal temperature variations by a spatial heterodyne interferometer using single-sided interferograms” by Ntokas et al. detailed and discussed a new method to derive two independent temperature profiles from 2-D interferograms obtained by a limb sounding spatial heterodyne interferometer. The method makes use of the horizontal information of the 2-D interferogram and is feasible. In general, the manuscript is well written and structured. However, some revisions are needed to be done to improve the work before it can be published.

General and specific points:

To split the 2-D interferogram into two single-side interferograms around the zero optical path difference is the key process to successfully derive two temperature profiles.  The temperature information comes mostly from the center of the interferogram, so it would be expected that the temperature retrieval is sensitive to the determination of ZOPD. The authors may consider to add some retrieval results when ZOPD cannot be determined precisely, which could often happen during actual observations.

The text has tendency to omit the definite article "the" in some of its sentences. It is recommended that the authors review and add "the" where needed.

The colored lines in some figures are sometimes difficult to distinguish one from another, e.g., Fig.2(b), Fig. 10(b), Fig. A1(a), and I would suggest to either change the colorbar or use markers if possible.

Line 125-157: For O₂ A band, self-absorption cannot be omitted below 90 km. When no self-absorption is assumed for above 80 km, it will affect the temperature retrieval to some extent between 80 km and 90 km. Authors may add some discussions on this.

Line 228-230: "values ..." is not clear in this sentence, please consider to rephrase/complement the sentence.

Technical correction:

Line 18: "...(Vincent (2015))..." would be "(Vincent, 2015)" and also in the other indirect citations in the text, e.g., L32.

Line 20: "...summarize..." would be "summarized"

Line 22: "...point out..." would be "pointed out"

Line 22: "...underline...outlines... " would be "underlined...outlined"

Line 45： “if it it possible...” --> "if it is possible..."

Line 88: "...a electronic transition..." would be "an electronic transition..."

Line 89: "..., which" would be "...., and" /"..., where"

Line 132: "...which..." would be "whose"

Line 138: "...show..." would be "showed"

Line 139: "...1.4 and 1.6 refers..." to "1.4 and 1.6 refer"

Line 325: "...decreases..." to "reduces"

Line 326:  duplicated "that" in the sentence

Line 333: "...affects..." to "affect"

This study discusses how a satellite-borne limb viewing spatial heterodyne interferometer could be used to retrieve MLT temperature profiles from O2 A-band observations. Specifically, how across horizontal LOS info can be used to produce 2 profiles per image by using the two halves of the interferogram separately. The paper is generally well written, and well thought out, and what’s presented is clear. However, a few issues (listed in “general issues”) need to be addressed before publication, as well as some more minor issues detailed below.

General issues:

You absolutely cannot ignore self-absorption in the 80-85 km region. Even in the 85-90 km region it is not negligible. If you want to include this region (80-90 km), you must account for self absorption.

My main concerns are in Section 4:

 It is not enough to simply say you’re “using HITRAN” to forward model the line intensities. At line 147, you say that you convolve the line strengths with the ILS. You’ve skipped a few steps here. How are you accounting for broadening? What types of broadening are you accounting for? Are you actually just convolving the line strengths? Because you need to convolve the emission spectrum (which you calculate from the line strengths), see Babcock and Herzberg, 1948 (doi:10.1086/145062).

The discussion of temperature precision is good. However, a discussion on accuracy is also needed, especially for the daytime retrievals. Specifically, on how you’re going to deal with background solar radiation and stray light, and how those will affect the accuracy of the temperature retrievals. It’s only at altitudes very close to 90 km where the background solar signal is somewhat negligible compared to the airglow signal. And, if this is intended to be on a nanosat, you’re likely going to have limitations on the size of baffle you can use, which means stray light will certainly be an issue. The source of that stray light will be from the bright Earth below, which will have a complicated self-absorption A-band signal, ie, it’s not a simple linear function across the spectrum that you need to subtract. These background signals need to be accounted for and discussed.

Specific issues:

Introduction: there have been two instruments launched recently that also use the A-band to measure MLT temperatures, MIGHTI on ICON, and the Swedish MATS satellite instrument. Please mention/reference these as well.

Line 26: This sentence is quite vague, please elaborate on why/where/how the instrument was developed.

Lines 26-28: This section is somewhat misleading. It sounds like you’re saying that the first instrument (described in Kauffman et al. 2018) was successful in measuring temperature profiles. In that paper, it says that the instrument worked nominally on a rocket launch, however, wasn’t able to produce temperature profiles. And the second part of this section makes it sound like a second instrument has been built and is ready to be tested. Is this the case? It should be made clear that Chen et al. 2022 is a simulation study.

Line 55 (and throughout text): “asses” should be “assess”

Figure 2: It would be helpful to split these into solar max and solar min in different plots. Also, maybe separate daytime and nighttime

Fig. 10b: the legend should also include the grey interferogram with no gradient

Line 316: I don’t recall any special attention being given to results above 120 km. Is this the intended altitude?