This paper investigates the potential of commercial RO (Spire and GeoOptics) observations to characterize the PBLH over Arctic region via comparison with operational RO data (COSMIC-1, GRAS MetOp) and MERRA-2 reanalysis. Additionally, the study inspects the performance of commercial RO penetration capability, as well as the relationship between moisture and RO penetration probability using radiosonde profiles from MOSAiC ship campaign. Here are several suggestions and questions the authors may consider to clarify in the manuscript.

Suggestions and questions:

1. There are multiple RO missions data set available from UCAR and ROMSAF websites, including KOMPSAT-5, PAZ, PlanetiQ, TerraSAR-X, TanDem-X, Sentinel-6a and GRACE-FO. These missions are contemporaneous with Spire and GeoOptics covering global area. They also show good penetrating capability and usefulness in PBLH application. Authors need to give explanation of the reason choosing C-1 and MetOp as the counterpart in evaluation for commercial RO.  
2. The radiosonde profiles from MOSAiC expedition ship campaign provide unique and valuable information on the PBLH detection in Arctic region as independent verification data. I expect to see the radiosonde derived PBLH and the direct comparison of the collocated RO-RAOB PBLH matching pairs. But in this paper the radiosonde data was only treated as source of water vapor for exploring the relationship between moisture and RO penetration. Consider the very limited geographic coverage of the ship campaign, does the conclusion cannot be achieved by using water vapor profile from climate model data, like MERRA-2 or ERA-5?
3. As far as I know, the MERRA-2 reanalysis provides two PBLH values. One is calculated based on the total eddy diffusion coefficient of heat (K_h) , and the other one is estimated using the bulk Richardson number method. The PBLH calculated from K_h is usually higher than the one from bulk Richardson number in most regions. It would be interesting to check the difference over the Arctic area and validate it using RO derived PBLH. However, the paper did not provide much description on the MERRA-2 PBLH. Reader don't even know how the PBLH was extracted from MERRA-2 reanalysis. 
4. In the PBLH deriving method section (2.1.3), the method of first minima of the refractivity gradient to exceed -40 N-unit km-1 and the 500m threshold for RO cut-off height are chosen without justification. A sensitivity study for the threshold, comparison of different methods (minimum gradient, wavelet covariance transformation etc.) and variables (refractivity based, bending angle based) are recommended, according to the discussion for figure 6(b) and 6(c).
5. The seasonal variation of RO penetration probability is displayed and discussed, whereas the more important seasonal variation of PBLH was not provided. In my opinion, a big picture of PBLH in north pole region is desirable (seasonal variation, diurnal circle if any, longitudinal variability related to the Atlantic Ocean current and sea ice distribution etc.) in the section 1.1, then a statement of how commercial RO can improve the understanding in section 3.3. 
6. In figure 1, the different penetration probability of Spire NOAA and Spire NASA may contributed by the sample noncoincidence, because Spire NOAA is a small subset of Spire NASA (~3500 out of ~12000 in one day). Whereas for GeoOptics, NOAA and NASA are basically covering the same observations. Therefor the explanation of the discrepancy of orange/red lines may completely different. Since the paper introduced NASA purchased commercial RO, which is processed by vendor, and NOAA purchased commercial RO, which is processed by UCAR, it's ideal to derive PBLH using both NASA and NOAA commercial RO, to help understanding the factors affecting RO penetration. 

Based on the discussion above, I think this manuscript needs substantial re-work to make it acceptable for publication.