This is an extensive and well written review of the measurements and analyses programs in place at the Izaña Observatory over the past twenty years. It is a review paper, differing from most in that the focus is on the FTIR measurements at this site rather than a broad overview of some subject matter. This is not a problem as the measurement and science record is extensive. This paper and the references therein provide a useful compendium of the research carried out at Izaña by the FTIR group and their colleagues, and it will provide a good starting point for researchers looking to understand the context of the work carried out there. It discusses some of the more significant results and provides the references to the supporting papers should a more in-depth look be of interest to the reader. The assessment of instrument performance over the 20 years is comprehensive and fairly well presented.  It is difficult to judge this paper in terms of  "substantial new concepts, ideas, methods, or data" as presented herein, but nevertheless I feel the paper does have merit in presenting in one place the extensive data record and the use of that data record to address many of the more prominent areas of atmospheric research.

Some general remarks:

The paper refers to the IFS120/5 HR. Is this to denote that the instrument is of the family of the IFS120HR and the later IFS125HR, or is it to denote that it entered service as an IFS120HR and was later modified to the specification of the newer IFS125HR?

With some of the figures that have many data points, the use of circles and filled circles (dots) of similar or the same colour can be problematic. It would be better to use clearly different symbols. Looking specifically at Figure 6, the use of gray circles and black circles is not a good idea.

Specific comments:

For Figure 2, could some indication of the mean number of measurements per day (binned per month) be given?

Line 120: Was the IFS 120/5HR operated in “vented mode” at all times, or just for the time period of the comparison with the IFS 120M?

Table 2: As formatted, this table is difficult to read. Is there a difference between the Target Gas and Gas columns? Perhaps the “Gas” should also be Target Gas. It would improve readability significantly if a few more spaces or a vertical bar were added to more clearly separate the left-hand grouping of gases from the right-hand grouping

Line 223: This would be more readable if it was written: This figure depicts the rows of the averaging kernel matrix A …”

Table 3: For readability rather than spacing the columns equally, I would suggest adding more space between the columns for the combined pairs, and removing a space between the M and σ, so the first line would look like:

C₂H₆  13625     1.48 0.15        1.81 0.39        5.44 0.17

Figure 3: As there are many similar averaging kernels, it would be of interest to see some explanation of why the particular kernel was selected for display in the figure.

Line 236: Are “great values” to mean those values that show large statistical uncertainty?

Line 241: maximum instead of maximal

Line 243: The systematic uncertainty budget is dominated by spectroscopic errors (instead of “led”)

Figure 4: The choice of lines and lines + circles in the same colour makes the plot difficult to evaluate for some constituents. Making the plot wider would help. Also, use just symbols for those constituents that have sufficient points and are fairly “straight” such that the lines are not necessary. H₂CO and HCl are both shown in cyan with the former being a thinner line. I can’t distinguish a difference in thickness in the lines in the plot.

Line 250: Do the authors mean to say that the predominant errors are located in one of the troposphere, upper troposphere/lower stratosphere or middle/upper stratosphere, or that the errors are in all three regions? If the latter, they do not appear to be equally concentrated in the regions and that should be discussed.

Line 257: Particularly, high error profiles … would read better as “large”error profiles unless the authors mean to suggest that there is also an altitude dependent component which does not seem to be the case as the lines are straight in Figure 4.

Figure 6: Caption refers to ‘grey-white dots” which is, I think, a grey circle that is not filled? See earlier comment about the use of circles and dots. Also the grey circle is easily confused with a black circle. Note also that the R_XCO2 quantity is used here but not yet defined in the text.

Line 312: I don’t think that ‘considering’ is the proper word here. I would suggest ‘using’. For example: “… using a scaling retrieval with a fixed WACCM a piori VMR profile, and PROFFIT software.”

Line 366: At the resolution the data are plotted with, it is not possible to discern a seasonal cycle in X_air N₂ . If the authors consider this to be a significant finding, then overlay a trace on the data that would display it. Otherwise, the statement could be left out without weakening the section.

Line 384: “… causing punctual downward and upward shift of the UTLS region,…” The use of punctual doesn’t make much sense. Punctual implies arriving on schedule. Is that what the authors mean to say?

Line 390/Figure 7: It would be a good idea to move the HF and N₂O frames to be next to each other to make the anti-correlation of the two easier to detect.

Line 559: There is a set of ellipses between CCMVal initiative and Schneider et al. Are there more missing from the list? There are 8 models listed and 10 papers referenced. It might make it clearer to list the CCM followed directly by the reference.

Lines 561-562: This is an awkward and confusing sentence and the details are lacking. There is mention of “attribution of sources/sinks”, but no comment on whether or not there is agreement. Similarly for the “representation of moist processes” evaluation. As the paper is already quite long, it might be best to make a general statement about the comparison to the CCMs and referring the reader to the above references for details rather than to try to call out these areas without further discussion.

Line 693 and below: Generally, when speaking of time the word used is coincident rather than collocation.

Line 690: It is unclear whether a actual single temporal criterion is applied here as multiple gases are being discussed. Sentence should read either “Similar temporal criteria …” or “A similar temporal criterion….” as the authors see fit.

Line 699: Over what time interval are the FTIR observations averaged? Are the total columns derived from a series of sequential spectra, averaged?

Line 703: Need a comma after “correction”

Line 706: If I understand what is being said, this might read better as “... each FTIR measurement is only paired once to the reference observation that minimizes the time difference within the temporal collocation window.” And I would use coincidence rather than collocation.

Figure 14: Presumably the “TRO” quantities refer to the tropospheric portions?

Line 785: In what manner are the PWV values over IZO “slowing down”?

Line 823: Do the authors mean to say that the NDACC FTIR product is able to capture “only” a part of its tropospheric variations?

Line 842: What is meant by “pure free conditions”?

Line 845: I suggest the wording be: Despite a considerable decrease in the number of coincidences, …

Line 849: I suggest the wording: ...which makes the comparison of the remote sensing and in-situ profiles difficult.

"Twenty years of ground-based NDACC FTIR spectrometry at Izana Observatory -  overview and long-term comparison to other techniques"  Omaria Garcia et al., 2021

The paper meticulously describes the methodology of FTIR retrievals and draws on previous work in the field. The writing style is clear and descriptive. It gives a very detailed description of the 20-years FTIR-related measurements from 120/5 HR at Izana. Such kind of overview paper is valuable for the global users (modelers, satellite developers, atmospheric scientists... ) to use their data. Izana is located in the subtropical region, which is crucial to understand the change of atmospheric compositions. The 20-years FTIR measurements have already been used in many scientific studies, leading to more than 100 peer-reviewed papers. Overall, I recommend this paper to publish in ACP, and I only have a few minor comments:

P8 line 182: WACCM model used in NDACC-IRWG is v4 instead of v6

P8 line 188: why only use the temperature and pressure at 12:00 UT? How about the H2O? The temperature and H2O variation can be very large even on one day. Would you like to address such uncertainty on your retrievals? 

P9 table 2. do you want to add N2 also here?

P14 line 261: "the total column-averaged amount of dry air (Xair) " is not appropriate. Xair is the ratio of o2 or n2 derived dry air to DPC, please use a better definition here.

P22 Figure 8. in the bottom panel, are you sure the colors are correct? because the CO is increasing, but you mention that in P23 line 464 that the CO is decreasing.

P22 Figure 8 in the middle panel and Figure 11, I see that the CH4 and N2O long-term trends are similar. However, other in-situ measurements show that N2O is increasing continuously while the annual growth of CH4 is variable: 1999-2007 stable, and re-increased after 2007 (https://gml.noaa.gov/ccgg/trends_ch4/). Any explanation here? Why we get a different CH4 tend from Izana FTIR CH4 measurements, especially between 1999 and 2007, compared to other surface measurements?

P39 line 811, the tropospheric XCH4 is compared to the surface measurements to found a bias of ~2.6%. I do not support such direct comparison, as they are sampling different vertical ranges still.

P39 line 814, the reference "Zhou et al., 2019" is wrong. It should be "https://amt.copernicus.org/articles/12/5979/2019/"

Manuscript by Omaira E. García et al. titled “Twenty years of ground-based NDACC FTIR spectrometry at Izaña Observatory - overview and long-term comparison to other techniques” presents a comprehensive analysis of the long-term FTIR-monitoring which is being carried out at Izaña Observatory. Izaña Observatory whose history dates back to 1916, has a strategic location for the investigation of atmospheric processes and contributes to numerous international programmes and observational networks (GAW-WMO, WDCGG, WOUDC, NDACC, TCCON, AERONET, BSRN, MPLNET, E-GVAP, NOAA/ESRL/GMD CCGG, etc.).

Authors provided a thorough description of the unique FTIR-experiment which was started in 1999. The abstract clearly presents the subject matter and findings of the paper. The scientific basis of the results reported in the paper is the reliable and recognized technique of atmospheric FTIR-spectrometry, and widely used inverse methods for atmospheric sounding (formalism by Rodgers(2000)). Both, the acqusition of MIR spectra of direct solar radiation using FTIR-system installed at Izaña Observatory and the following FTIR-spectra processing are described by authors in detail.

The investigated time series of C2H6, CH4, ClONO2, CO, HCl, HCN, H2CO, HF, HNO3, N2O, NO2, NO, O3, OCS, and three isotopologues of water vapour (H216O, H218O, and HDO) are of fundamental importance to the atmospheric studies including the interactions of atmospheric composition and climate, the investigation of trace gases temporal variations and processes driven these variations, the verification of modern CTMs (chemical transport models) and the validation of satellite obsevations. The manuscript is well-written and structured, contains new results that can be of interest to scientific community. Bibliography, in general, provides a relevant list of references, nevertheless, according to referee’s opinion, the number of references could be reduced because the bibliography section occupies about one six of the whole paper volume.

1) Lines 131-134: “By evaluating spectral signatures of vibrational-rotational transitions contained in the solar absorption spectra measured, the FTIR technique allows total column amounts and low-resolution vertical profiles of different atmospheric trace gases to be retrieved with a high degree of precision.”  The “degree of precision” is expected to be different for different trace gases and not necessarily “high” for those retrieved species which have weak absorption signatures in the analyzed FTIR-spectra. If statistical errors/uncertainties can be considered as a measure of “degree of precision”, we can see in Table 3 that these errors can reach ~50% for H2CO and ~100% for ClONO2.

2) Section 3.1: Whether FTIR-instruments (an IFS 120M and an IFS 120/5HR) at Izaña Observatory have been operated remotely or by an operator/technician? Could you please specify?

3) Lines 209-211: “The only quality filter applied on public FTIR products is that observations taken at high solar zenith angles (≥85°) have been excluded to avoid imprecise retrievals (mainly caused by misalignments of the solar tracker or spectroscopic issues). These data represent less than 1% of the total data set.” It is expected that clouds are one of the most important factor leading to the outliers in retrieval results. Are the FTIR-observations at Izaña Observatory free of this effect?

4) Lines 191-193: “Most relevant changes are those related to CH4, for which the spectral micro-windows are adopted from Sepúlveda et al. (2014), and the spectroscopy parameters correspond to the improved linelist provided by Dubravica et al. (2013).”  What are the principle differences between CH4 retrieval strategies reported in Sepúlveda et al. (2014) and Sussmann et al. (2011)? Does the modified CH4 retrieval strategy by Sepúlveda et al. (2014) provide the homogeneous results with other IRWG-NDACC sites which make retrievals according to Sussmann et al. (2011)? Please, clarify this. Reference: Sussmann, R., Forster, F., Rettinger, M., and Jones, N.: Strategy for high-accuracy-and-precision retrieval of atmospheric methane from the mid-infrared FTIR network, Atmos. Meas. Tech., 4, 1943–1964, https://doi.org/10.5194/amt-4-1943-2011, 2011.

5) Fig.14 (page 37) and Fig.15 (page 41): This is not easy to distinguish between the sizes of dots which correspond to R²=0.5 and R²=0.3.

6) Maybe, it is worth adding to manuscript a table summarizing all the long-term trends reported and discussed in the text in Sections 5, 6, and 8. Such a table will simplify reading and navigation through the manuscript.

7) Section 5, Fig.7, and Appendix B: Which methods and/or criteria were implemented for the selection of an optimal set of frequencies used for the construction of multi-regression fit presented in Fig.7? Evaluation of statistical significance, cross-validation, etc.?