General comments:

The manuscript "Total Column Ozone Retrieval from Novel Array Spectroradiometer" published by Egli et al., presents a study on the use of a relatively new array spectroradiometer for ground-based measurements of total ozone column in the atmosphere. The new array spectroradiometer has the potential to provide more accurate and precise measurements compared to traditional methods such as those using Dobson instruments, grating spectrophotometers, etc. The study also provides a comparison of total column ozone retrievals between the proposed method and established methods. This helps to demonstrate the potential of the new system and the associated retrieval technique.

One of the strengths of this study is that it presents an approach for measuring total column ozone that is new, fast and automated, while utilizing available, easy to acquire software packages and hardware. The study also leverages on the established instrumentation and expertise at PMOD for standard calibrations.

However, I believe that this manuscript would still benefit from a chapter on error analysis containing a detailed error budget. I understand that some of the aspects of the methodology have already been done elsewhere, nevertheless it would be useful and important to include such a chapter. For example, I would be interested in knowing the signal to noise ratios of the spectra, typical wavelength shifts, how the uncertainty in the LSQ retrieval is calculated, and so on, without much digging through literature.

In conclusion, this manuscript fits well within the scope of AMT. Therefore, I recommend its publication after addressing the general comments and some of the comments and corrections below.

Specific comments:

1. The authors refer to the “low cost” of the Koherent system but do not provide any estimates of the costs involved and how they compare to available systems like the Brewer, BTS-Solar, etc.
2. Can the authors please comment on any effects of UV radiation on the degradation of the optical fiber and if this would have noticeable effects over time?
3. Why use an optical fiber instead of adopting a similar design to the BTS-Solar?
4. BTS2048-UV-S-F array spectroradiometer: According to specifications, this spectroradiometer has a calibrated measurement range of 200 nm to 430 nm. Why do the authors truncate the upper wavelength range to 345 nm? Would it not be useful to include the maximum range covering the Fraunhofer lines at around 393 nm? I think this will make it easier to determine any wavelength shifts, will it not?
5. Two-point calibration: I don’t quite understand the rationale of changing the absorption coefficients in addition to adjusting the ETC. It seems to me that the absorption coefficient is simply used as a “tuning parameter” in this case. Aren’t the slit functions well determined, as well as the ozone cross sections? How would the authors explain the need to change the absorption coefficient?
6. The authors refer to minimal least squares, what do they mean by “minimal”. A sentence or two explaining this would be sufficient.

Technical Corrections and Suggestions:

P.1, Line 24: “within less than 0.7%” --> within 0.7%

P.2, Line 45: “In the Dobson instruments, prisms are selecting ...” --> In the Dobson instruments, prisms are used to select ...

P2, Line 46: “Most of the Dobsons are manually operated and require therefore …” --> Most of the Dobsons are manually operated and therefore require …

P2, Line 52: “contrary to single ...” --> in contrast to single …

P2, Line 52 and 69: “suffer from stray light …” --> suffer from the effects of stray light …

P2, Line 53: “The Brewers were formed to a network of automatic stations, which required few ...” --> The Brewers were used to form a network of automated stations, which required less …

P2, Line 53: “best consistency …” --> greatest consistency

P2, Line 60: “irradiance ratio at the top of the atmosphere” : irradiance ratio of what?

P2, Line 76: “Similarly as Pandora …” --> Similar to the Pandora …

P2, Line 83: “Contrary to the …” --> In contrast to the …

P2, Line 83: “fiber coupled” --> fiber-coupled

P9, Line 363: “clears sky” --> clear sky

P9, Line 369: ”The two-years ...” --> The two-year …

Fig. 1. Caption: “… spectra on morning of 15 September …” --> spectra on the morning of 15 September

Review of Egli et al., " Total Column Ozone Retrieval from Novel Array Spectroradiometer"

General Comments

Egli et al. present the results of a two year comparison of total ozone measurements from a novel CCD array based instrument (named "Koherent") to a well-calibrated double monochromator Brewer (#156) at Davos.

Koherent is based on a modern array spectroradiometer manufactured by Gigahertz Optik, as used in the BTS-Solar instrument, but differing in that it is coupled by fibre to a telescope mounted on a solar tracker.

The instrument was previously described in Zuber at al. (2021) where a similar comparison to a Brewer in Davos was also reported, but here the retrieval technique has been refined and the comparison period extended.  

Such work is extremely useful for the global ozone observing system, where there has long been a desire for a relatively cheap, automated and easy-to-maintain instrument to supplement or even replace (if the quality of measurements can be demonstrated to be high enough) the existing Dobson and Brewer networks. The subject of the manuscript is thus very suitable for AMT.

While the analysis presented is very sound, I believe the current version of the manuscript is deficient in terms of motivation and discussion of results, and needs enhancement before it is suitable for publication. Several options for the retrieval are presented and the results compared, but there is minimal discussion of motivation such as the theoretical  advantages and disadvantages of the different methods (in particular, least squares fitting versus the ratio method) and how these are manifested in the results. In some places, the text states parameters have been optimized but there is no explanation given of how these optimizations were performed.

Rather than identifying or developing a preferred retrieval method, the authors made a decision to simply average the results of three different retrieval methods. This seems an odd choice to me and definitely needs to be properly explained.

Also lacking from my perspective is the motivation of the design of the Koherent – what are its anticipated advantages compared to the BTS-Solar (particularly in the light of Zuber et al 2021 which overall seemed to favor the BTS-Solar configuration) or to the other CCD instruments mentioned in the introduction.

I would also like to see at least a small comment about the anticipated long-term stability provided by the design of Koherent, in particular where degradation might be expected over time. This of course is a crucial aspect for an operational instrument.

One unrelated general point is that the authors frequently use quite recent works (Gröbner et al. 2021 and to a lesser extent Redondas et al. 2014) as references for some longstanding points about Dobson and Brewers. I think it would be better scientific practice to cite the original literature in these cases.

Specific comments

Line 9 – I find this statement misleading because total ozone retrievals from Koherent have been presented before and compared to a co-located Brewer at Davos (ie in Zuber et al. 2021)

Line 37 Please re-word this sentence for better clarity as it is not clear what exactly you are referring to – do you mean the date when ozone depletion was first identified in observations?

Line 42 I suggest a reference to Figure 2 here would help the reader.

Line 56-57 As I mentioned in the general comments, this is an example where I don't think it's appropriate to only mention the papers of Gröbner and Redondas.

Line 94 You should explain what the motivation is for this configuration.

Lines 95-112 I find the discussion of the different LSF algorithms confusing – you mention a "first version", then an "improved, standardized and optimized" version a "standard version … which is "considered here as the standard retrieval". However, you then go on to modify this algorithm further, eg applying a stray light correction. Saying "the standard LSF retrieval algorithm" might also be interpreted as meaning this algorithm is in wide use outside of your institution which I don't think you mean to imply.

Line 123 You say here "selecting the best retrieval" but I don't think you ever did that?

Lines 130-132 Just as a question, would the temperature stabilization be sufficient for extreme environments, eg desert, tropical or polar?

Lines 155-156 There also appears to be a small dip in the middle of the morning for 305 and 310 nm?

Line 172 You should explain why you are using a different solar spectrum now.

Line 173 What does "non-linear" mean here?

Line 190 The use of the Payerne data needs more detail - I take this to mean the total ozone value can only be calculated after the next ozonesonde flight so that you can use the interpolated  'daily input'? How are you treating the portion of the ozone profile that is above the burst height of the balloons?

Line 194 Why wouldn't you use the value of pressure at the actual time of the measurement?

Lines 185-197 You don't explain what values you are using for the ozone airmass and the aerosol airmass. Are you using the latest ozone profile from Payerne for the ozone airmass?

Line 200 At this point, I would have expected to see some motivation for the double-ratio technique. What are its advantages and disadvantages compared to least-squares?

Line 204 "According to Gröbner " – I would prefer you to say "Following the notation of Gröbner "

Line 211 You should explain the reason why you "selected" these particular wavelengths.

Lines 221-223 Yes, but you should say why

Line 232 This makes it sound as if Gröbner et al. invented the double ratio technique!

Line 241 Explain "consistency with units" more clearly.

Line 243 But Section 2.2 also doesn't give any detail about how you are using the balloon soundings to calculate a daily effective ozone temperature

Line 248 As for the LSF, I am surprised you use the climatological pressure?

Line 260 You say they are all "customized for Koherent" but you don't give any information about how this customization was undertaken.

Line 303 According to figure 3, there is only a small improvement from restricting the wavelength range, but in figure 1, the stray-light effect is quite large for 305 nm compared to the longer wavelengths – could you please comment on whether this is consistent?

Lines 313-315 What is the physical basis for the stray light correction, considering the previous discussion?

Lines 319-323 The stray light performance is certainly greatly improved but the offset seems to have slightly increased?

Lines 324 Does this mean that, in general, Koherent would not be recommended for use for ozone slant paths greater than 900 DU? How does this compare to a Dobson?

Line 351 The point is though that the retrieval assumes the slits have been correctly adjusted (eg Köhler et al. 2018 which is in your references)

Line 358 This confuses me – isn't this just calculated for Koherent based on your definitions?

Line 379 Have you actually shown that it is stable for two years? Are you saying this because of Figure 5?

Lines 383-386 This seems very odd to me. What is the rationale for merging the results of three different retrievals, which would all presumably have their own strengths and weaknesses? There is only one true value of total ozone at any one time. This point needs more explanation.

Lines 392-393 Do you mean 0.88% is the uncertainty corresponding to the difference between the three methods?  

Line 403 I don't think Redondas et al. 2014 was the first to report this fact

Line 426 – As mentioned previously, how are you accounting for the portion of the profile above the burst height of the balloon? This might be contributing to the difference with the reanalysis.

Lines 329-429 From what I understand of your approach, the retrieval of temperature is almost accidental, in that you have applied a Dobson-style retrieval and a Brewer-style retrieval, and knowing the Dobson-Brewer difference is sensitive to effective ozone temperature, then gone on and calculated your regression. Couldn't this method be improved though, if you went back a step and selected from the start, a combinations of wavelengths chosen specifically for the retrieval of effective ozone temperature? Could you comment on this please.

Line 343 I agree this is a very useful advantage for operational instruments.

Line 438 I don't this statement is quite true because this was already shown in Zuber et al. 2021.

Line 459-461 I hope you will continue to monitor and assess the performance of Koherent over a longer time period and report on it.

Line 640  I find this plot slightly misleading, because the blue dots, the light blue circles and the black line all relate to the difference between Koherent and the Brewer, but the grey band relates to a different quantity, namely the difference between the three retrieval methods. I know this fact is stated in the caption, but I wonder if there is some way you could make this more intuitive visually for the reader?  

Minor comments

Line 1 "From a novel array …"

Line 52 Dobson-> Dobsons

53-54 Please re-word this sentence for better grammar

55 "are selecting" -> "prisms are used to select"

Line 209 Were -> Where

Line 254 "Analagous" -> "Analagously,"

Line 273 "attributed to" -> "due to"

Line 349 "good" -> "well" or "the performance is at least as good"