This manuscript describes an approach to identify absolute calibration trends of cloud radars. It is an important topic since calibration offsets cause biases in retrievals which use radar reflectivity measurements. I believe that revisions are necessary before the publication of the manuscript.

Main comments.

1. The fitting calibration technique requires the use of Dze values, which are the differences between measured and disdrometer-based reflectivities with subsequent estimation of the intercept of Dze-log(R) fits. The authors need to clarify better why the calibration offset estimated directly by comparing radar measurements when R is small (e.g., at the onset of precipitation, so WRA is negligible) is not good enough.
2. The KAZR radome is tilted by 4 deg (line 184) to remove the rain water and minimize wet-radome effects. Did this design prove to be not effective? Would your results hold for different tilted angles (e.g., reducing WRA magnitude with the tilt increase)?
3. WRA are likely dependent on the radome type, because different radomes remove rain water with different efficiency. Given this, one would expect different slopes of Dze-log(R) relations for KaSACR and KAZR. You assume that these slopes are the same (~8.6 in Figs.5). Please clarify/explain.
4. What is the reason of the large change in the KAZR calibration offset (up to ~7 dB or so) towards the end of the deployment (Fig. 8)?
5. It appears that for R > 2 mm/h, your log-linear WRA- log(R) relation deviates significantly from the R^(1/3) behaivior suggested by previous studies (Fig. 6).
6. Provide more information on frequency dependence of water absorption on a radome (line 142) as opposed to attenuation by cloud and precipitation drops, which is not scaled as the wavelength. The Bertie et al. 1996 reference (line 142) is not in the reference list.
7. Please add some discussion on applicability of your approach to different radars, different radomes and different disdrometers (e.g., PARSIVEL, Joss…).

Minor comments

1. Providing Ze – R relations for X and Ka-bands (using reflectivity as independent variable in such relations) in Fig. 3 would be informative. Using reflectivity as independent variable in such relations)
2. I believe you are assuming that rain-rate and DSDs at the disdrometer level and at the radar resolution volume are the same. Please specify.
3. In different parts of the manuscript you use “dB” units instead of “dBZ” units (e.g., lines 216-217, Y-axis in Figs, 2b, 4c, and other instances), and “dBZ” units instead of “dB” units (e.g., lines 253, 271 and other instances).
4. The manuscript could benefit from additional editing.
5. You use abbreviations “KaSACR” and “SACR” interchangeably. I suggest using KaSACR everywhere.

The manuscript by Deng et al. describes a method for monitoring trends in calibration of ARM cloud radars and the effect of wet radome attenuation (WRA). Both aspects are important because the number of cloud radars working at Ka band is increasing also because there are (at some extent), affordable radars commercially available that are used mainly for research purposes. Their performance should be monitored and the dependency of the impact on measured reflectivity factor of the WRA, which seems predominant over precipitation and gas attenuation must be predicted. The proposed method is based on a linear relation between the difference (in dB) between the Ze predicted from disdrometer measurements at ground and the one radar measured by radar at 500 meters. With proper averaging, intercept and slope of the relation express the system bias and the rainfall-dependent WRA, respectively.

The current manuscript focuses on one ARM setup. It would be beneficial to generalize the method for different radars and disdrometers. Extending the method to laser disdrometers should not be too difficult, although laser disdrometer measurements are generally considered less accurate than video disdrometer measurements. Laser disdrometers should be available in the TRACER campaign. Extending the method to other types of radomes could be challenging, as radome attenuation depends on many factors, including the hydrophobicity of the radome material, its aging, and the geometry of the radome. The Gibbs formula and the works by Gorgucci et al., Frasier et al., and Schneebeli et al. (doi:10.5194/amt-5-2183-2012) refer to an X-band radar with a hemispherical radome, which might explain the different behavior observed in Figure 6.

Minor Suggestions:

• Please replace “K” or “K” with “k” for specific attenuation to allow a better identification (e.g., in Figures 2a and 4a).
• In Figure 5, please remove the “corr” within the panels or improve the panels.
• The different uses of Ze, which can be predicted or measured with different radars, can be confusing. Consider using superscripts to differentiate them, although this is just a suggestion.