This study compares seven collocated infrared spectrometers over a month‑long deployment to evaluate consistency in radiance measurements and retrieved thermodynamic profiles.

The instruments agree within 1% in radiance, display minimal spectral calibration differences, and produce temperature, humidity, and derived quantities (such as PWV and boundary‑layer height) that match well within expected uncertainties.





General Comments

This is a clear, rigorous, and well‑presented intercomparison study that convincingly demonstrates the reliability of modern IRS instruments for network applications.

The scientific approach is sound, the analysis thorough, and the presentation is clear and well structured. I find the work robust and valuable, and would recommend publication subject to minor technical corrections.

Technical corrections

r160 has -> have

r162 provides -> provide

r182 missing Fig. number

r218 TROPoe acronym is already defined in r71

General Comments

This study provides an intercomparison between seven collocated ground-based infrared spectrometers (ASSIST systems) during a one-month period in the fall of 2023 in Boulder, CO. The purpose of this study is to show that identical instruments (i.e., ASSISTs) produce the same observations (within an instruments’ uncertainty) and thus can be used/interpreted interchangeably when deployed as a network system. This is an important assumption to confirm in order to discern whether differences within an observational network are associated with atmospheric changes and not instrument characteristic differences. This work also provides further confidence in the use of infrared spectrometers for planetary boundary layer observational work, with real world potential applications particularly with estimating advection. 

The instruments’ measured infrared radiances agreed to within 1% of ambient radiance across both opaque and transparent spectral channels, and their spectral calibration differences were quite small (≤7.1 ppm). Using the TROPoe optimal estimation retrieval, temperature and water vapor profiles derived from each instrument showed near-zero mean biases and excellent agreement well within retrieval uncertainties under both clear sky and cloudy conditions. Derived quantities such as precipitable water vapor and planetary boundary layer height also closely matched across all systems. These results demonstrate high precision and repeatability of both the ASSIST instruments and the TROPoe retrieval framework, supporting their suitability for deployment in profiling networks where observed spatial differences can be confidently attributed to real atmospheric variability rather than instrumental or retrieval artifacts.

The presentation of this work, along with the methods used, are sound and well presented. I would recommend publication after minor revisions, which are mostly technical in nature. 

Specific Comments

1. Line 72: Suggest rephrase to ‘TROPoe uses the line by line radiative transfer model (LBLTRM; Clough et al. 2004)...’
2. Lines 181-193: You point out the ‘spikiness’ in Fig. 6, associated with the sides of absorption lines and different residual instrument response functions. However, it isn’t clear how this would effect the end user or how you address this (and the other instrument differences in this paragraph) in the following sections. You show clear agreement among retrievals, but don’t close the loop of how the differences highlighted here correspond to the final product. 
3. This intercomparison study was completed in a relatively dry environment. Would you expect the instruments and/or retrievals to behave similarly in a humid region, such as in the deep south? I think this would be a beneficial piece of discussion for the conclusions section. 

Technical Comments

1. The following references are missing from the reference list:
1. National Research Council 2009
2. Illingworth et al. 2018
3. Michaud-Belleau et al. 2025
4. Gero and Turner 2011
5. Shupe et al. 2013
6. Turner and Mlawer 2010
7. Adler et al. 2013 
8. Mlawer et al. 2024
9. Minnett et al. 2001
10. Clough et al. 2004
11. Hu et al. 2019
2. The Sisterson et al. 2016 reference in the reference list is not referenced in the paper.
3. Line 182: Figure number is missing. 
4. Line 218: TROPoe is already defined, no need to redefine
5. Figures: Consider using colorblind-friendly choices, particularly in Fig. 3 where red/green are used together in a way that would make it hard for an impaired individual to tell the difference between the HBB temperature and HBB emissivity apart.
6. Figs. 4 and 5: Update with higher resolution, it appears fuzzy.

This manuscript presents an intercomparison of seven co-located infrared spectrometer observations. The work is very valuable as the relative calibration and random uncertainties between different instruments proved to be very small, thus allowing a network use of these instruments. 

General comments:

The manuscript is well written and structured. Also, concise and clear language is used.

Minor comments:

Abstract, line 27: you mean "quantities", not "qualities", I guess? 

Line 195: check phrase ("observed" is repeated)

Line 253, and Fig. 10 (right): Do you have an idea why ASSIST-20 has a larger disagreement in the lower layers? If so, could you give a short explanation in the text?

Line 266: PWV uncertainties are 0.1-0.3 mm (not 1-3 mm) according to Fig. 12

Figure 1 (and 3): ASSIST measures up to 3300 cm-1. Why do you show only the range until 1400 (1500) cm-1. If you have a good reason for this, please mention it in the paper why you restrict these plots to this selection of wavelengths. Otherwise, I would suggest to show the whole spectrum. 

Figure 4: Why is the upper axes range set to 150? It seems that there are also data above this value. Do you have a reason for that? I would suggest showing the whole range. 

Technical comments:

Please check the list of references. It seems that many references in the introduction are not listed.