The paper describes the design and performance of Shortwave Array Spectroradiometer-Hemispheric (SAS-He) instrument, with the shadowband front optics, which significantly increases spectral coverage and complements the multifiler rotating Shadowband radiometer (MFRSR) instruments employed by ARM network. The  SAS-He uses  two grating spectrometers: one for UV-VIS-NIR (350-1040nm, Si, FWHM~2.5nm) and another for SWIR (990-1700nm, InGaAS, FWHM~6nm) spectral ranges. Light is collected by a hemispheric diffuser  connected by a fiber optical cable to the spectrometers and solid-state linear detectors in a chilled housing.  Rotating shadowband allows measuring sun-blocked and unblocked irradiance spectra every 30 seconds from which direct -normal, diffuse -horizontal and global-horizontal irradiances are calculated similarly to MFRSR.  The design provides moderate spectral resolution with 2-4 oversampling rate , which is not optimal for measuring atmospheric trace gases, but allows measuring continuous spectra of the total optical depth (TOD), aerosol optical depth (AOD), diffuse and direct irradiances, and direct to diffuse ratio (DDR) outside of strong gaseous absorption lines.

Some of the limitations are internal stray light in UV-blue wavelengths which limits UV coverage to wavelengths longer than 380nm (Fig 2b) and non-linearity, which is empirically corrected taking advantage of the co-located MFRSR measurements (Fig. 3).

I found the description of SAS-He calibrations and data corrections too brief. Some important corrections are not mentioned. The general reader would benefit from more detailed calibration descriptions and more complete references.I provided some suggestions in the technical comments below.

The SAS-He instrument was in operation for 10+ years and significant volume of measurements have been collected and archived over three extended field deployments as part of ARM mobile Facility (TCAP, TRACER;EPCAPE). The results are presented as scatterplots (Figs. 7-9) and regression tables. On average SAS-He measurements agree well with multispectral MFRSR and Cimel Sun-photometer (CSPHOT). However, there are outliers and even biases (e.g., AOD at 1020nm: Fig 7, Tables 3-4), which deserve more detailed description. Comparisons with MFRSR of DDR (Fig. 8) and diffuse irradiance ( Fig. 9) show larger scatter and problems with Diffuse irradiance at 1625nm, (Fig 9).

Since MFRSR and CSPHOT measurements are considered as a reference, the paper would benefit from examples of the new information uniquely provided by the SAS-He instrument.

One main advantage compared to direct-sun photometers (CSPHOT) is the ability to measure spectra of diffuse irradiance under overcast conditions-retrieval spectral cloud properties.  However, cloud data are not specifically discussed.

The paper is suitable for AMT, and I recommend publishing after correcting web links and references as suggested in my technical recommendations below.

Technical suggestions:

44: infrequently (typically several times a day) and do not capture the diurnal cycle

There are now Geostationary aerosol measurements from NASA:

https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/applications/geoleo/

72  add MFRSR SSA references:

Mok, et al., Comparisons of spectral aerosol absorption in Seoul, South Korea, Atmos. Meas. Tech., https://doi.org/10.5194/amt-11-2295-2018 , 2018

Mok, et al., Impacts of atmospheric brown carbon on surface UV and ozone in the Amazon Basin, Sci. Rep., https://doi.org/10.1038/srep36940 , 2016

Corr, et al., Retrieval of aerosol single scattering albedo at ultraviolet wavelengths at the T1 site during MILAGRO (2009), Atmos. Chem. Phys., 9, 5813–5827, https://doi.org/10.5194/acp-9-5813-2009 , 2009

Krotkov, et al., Aerosol ultraviolet absorption experiment (2002 to 2004), part 2: absorption optical thickness, refractive index, and single scattering albedo, Opt. Eng., 44, 041005, https://doi.org/doi:10.1117/1.1886819  , 2005

143:. Better to move to Figure 2 caption. Explain color bar units.

Section 2.2:

Describe if tilt and misalignment corrections if applied, e.g.,

Mikhail D. Alexandrov, Peter Kiedron, Joseph J. Michalsky, Gary Hodges, Connor J. Flynn, and Andrew A. Lacis, "Optical depth measurements by shadow-band radiometers and their uncertainties," Appl. Opt. 46, 8027-8038 (2007)

In 2.2.5 give more details of the angular calibration and figure with the Lab optical setup and example of the measured cosine response at different wavelengths.

Describe corrections to the diffuse irradiance due to blockage of the forward scattered light by shadow-band (aureole correction), e.g.,

Min, Q., E. Joseph, and M. Duan (2004), Retrievals of thin cloud optical depth from a multifilter rotating shadowband radiometer, J. Geophys. Res., 109, D02201, doi:10.1029/2003JD003964.

In 2.2.6 give a more detailed description of the non-linearity correction if this has not been published before or give a reference. Acknowledge that MFRSR DDR is biased for coarse aerosols (e.g., dust) and cirrus clouds, due to the blockage of the forward scattered aureole light as discussed in Min et al., JGR 2004

170-175: Show lab setup and add figure with the lab measured angular response at different wavelengths.

179:  Add figure which shows results of non-linearity testing or add reference.

180-181: clarify this sentence:  To first order, the non-linearity of the direct irradiance measurement becomes incorporated in the cosine correction described above

Figure 4: Give units for Y-axis

231:  Gaseous NO2 absorption becomes important for single scattering albedo retrievals at small AODs, e.g.,  Krotkov, et al  (2005), Partitioning between aerosol and NO2 absorption in the UVA (https://doi.org/10.1117/12.615285)

235-236: Give proper references and correct URLs:For TOMS total ozone product:     https://acd-ext.gsfc.nasa.gov/anonftp/toms/

TOMS Science Team (Unreleased), TOMS Nimbus-7 Total Column Ozone Daily L3 Global 1 deg x 1.25 deg Lat/Lon Grid V008, Greenbelt, MD, Goddard Earth Sciences Data and Information Services Center (GES DISC), Accessed:

, https://disc.gsfc.nasa.gov/datacollection/TOMSN7L3dtoz_008.html

For OMI total ozone product:

Pawan K. Bhartia (2012), OMI/Aura TOMS-Like Ozone and Radiative Cloud Fraction L3 1 day 0.25 degree x 0.25 degree V3, NASA Goddard Space Flight Center, Goddard Earth Sciences Data and Information Services Center (GES DISC), Accessed , https://doi.org/10.5067/Aura/OMI/DATA3002 

For OMI instrument: https://aura.gsfc.nasa.gov/omi.html

Figure 7 and Tables 3,4:  Explain significant differences in AOD comparisons at 1020nm during TRACER and EPCAPE.

Figure 8: Extend Y axis to show full range of DDR variability.

Summary:

This manuscript presents an instrument designated ‘ Shortwave Array Spectroradiometer-Hemispheric (SAS-He)’  with results from the TCAP, TRACER, and EPCAPE ARM campaign, showing AOD, direct to diffuse irradiance ratio, and downwelling irradiance. The authors present a comparison of their measurements during these campaigns, focusing on specific spectral bands. The document includes tables presenting the evaluation results for each spectroradiometer in terms of their performance metrics. The authors conclude that while all three instruments show good agreement overall, there are some differences observed within specific spectral bands and under certain atmospheric conditions. They suggest further investigation to understand the causes behind these discrepancies and potential improvements to enhance inter-instrument consistency. Notable is the description of the instrument correction factors and calibrations that have been derived. 

This is a great manuscript to read, however there are a few minor comments to address, mostly on some clarification of some points (see list below). After these minor comments are addressed, it is recommended for publication in AMT. 

General Comments:

1. The direct to diffuse ratio is both used for quantifying the non-linearity of the spectrometers and the comparison to MFRSR. It is not evident if the subset of data is used is the same for non-linearity correction and comparison to MFRSR, and if there is a circularity in these comparisons. 

Specific Comments:

1. Line 24-25 (repeated in the summary): How does the uncertainty of the cimel sunphotometer translate to the uncertainty metric from this method 0.01-0.02 root mean square error, may not exactly equate to the accuracy uncertainty. Some refinement in this statement to differentiate uncertainty in accuracy and root mean square.
2. Line 60-61: How does the SWS influence the development of the SAS-He? Was it at all influence? If so then a citation might be adequate here.
3. Line 88: if it is a single core fiber, how does the coupling work to split exactly 50/50 in the Y fiber optic?
4. Line 93: Wouldn’t a celsius scale be more appropriate to highlight the stability of the thermal control?
5. Spectral Registration: How often is this procedure run? since 3 campaigns are shown, has this been done only once, or is there a time series of spectral calibration to ensure spectrometer stability?
6. Spectral Resolution: How was this measured?
7. cosine correction: How far off from cosine was the lab measurements returned?, and is it spectrally neutral? While most of this can be corrected, if substantial, it would indicate that sampling is uneven, so there is increased error in the hemispherical measurements.
8. Spectrometer non-linearity correction: There seems to be the whole range of potential non-linearity correction factors at low direct to diffuse ratios (vertical color range in figure 3a) Does this mean that the correction is limited in efficacity at the lowest end of the ratio? There is no mention that there are limits to the correction factor at low ratios.  This seems potentially problematic to accurate measurements.
9. Line 196: degree Celsius or Fahrenheit?
10. Line 198: How much variation is observed in the 30 second dark measurements?
11. Line 199: What is the temperature of the minimum in temperature sensitivity?
12. Figure 5: What is the standard deviation from the fit of the langley calibration presented here? And the resulting impact to the expected accuracy in AOD?
13. Figure 6: How about the wavelengths near 1220 nm for AOD?
14. Line 299: Is ‘as’ missing for ‘such as’?

Referee comment on “Shortwave Array Spectroradiometer-Hemispheric (SAS-He): Design and Evaluation”

Anonymous Referee

General comments:

Overall, this paper provides a well-written description of the Shortwave Array Spectroradiometer-Hemispheric (SAS-He), a novel ground-based radiometer capable of providing spectrally resolved total irradiance measurements. These measurements can be utilized to calculate both the total and aerosol optical depths (TOD and AOD), as well as the direct-to-diffuse ratio (DDR). Despite delving into the technical details regarding the design, calibration, and correction of this new instrument, the paper also includes a meticulous evaluation using independent, collocated instrumentation across a diverse range of observational conditions, such as those encountered in the TCAP, TRACER, and EPCAPE measurement campaigns.

I believe that this manuscript aligns perfectly with the scope of AMT, and the presented results are indeed relevant. There are only a few minor technical remarks to address.

Minor/technical comments:

Page 2, line 47: Following Giles et al. (2019), AERONET derives AOD at 9 different spectral bands although the AOD at 935 nm is extrapolated based on the Ångström Exponent. Why did the authors state in the text that AERONET provides information at seven spectral bands?

4: There is a typo in the y-axis? Are these values unitless?

Section 2.2.8 and Figure 4: What atmospheric conditions are considered "good" by the authors in the Figure? I believe that some explanation about this classification in the text would be necessary. Additionally, are the spectral bands not included (marked with asterisks) due to the influence of atmospheric gas absorption? Have the authors accounted for these absorption processes in their calculations?

Figure 5 and Section 2.2.9: Are the authors applying the Langley-Plot method between air masses from 1.X to 6, as stated in the Figure?

Page 9, line 220: I consider it will be highlighting to include the number of Langleys performed (and maybe the time interval?). I haven’t read this information in the text or maybe I have missed this number. Furthermore, the authors set a threshold in the text to define those stable Langleys performed in the whole time series: below 1% per day. Are the authors talking about the standard deviation of the fitting?

Figure 6: Is Optical Depth (OD), TOD (as expressed in the caption) or AOD (as expressed in the legend)? Please clarify. I understand that with “good AOD”, the authors are referring to those spectral bands that can be used to retrieve effectively AOD and TOD from the SAS-He. Maybe the term “good AOD” is not the best one to be included in the legend. Why not include the rest of the absorption from NO2, CH4, O2, CO2, etc that can impact your measurements in your spectral range?

Page 10, line 250: It is important to highlight that you are comparing with independent instruments!

Page 14, first paragraph: Why do you think the 1020nm spectral band presents worse results (in addition to 380nm in TRACER)?