This paper provides an overview of the IWV, LWP, and thermodynamic profiles retrieved from two ground-based multi-channel microwave radiometers and the passive observations from two single-frequency cloud radars.  Details are provided on the retrieval methods.  A detailed analysis and comparison of the IWV and LWP, including a discussion of the sensitivity of the different systems and potential sources of bias, was included.  Overall, I find this paper to be well-constructed, the writing clear and precise, and that the paper should be accepted after some relatively minor suggestions are addressed.   Nice job, authors!

• Line 193: Would you please provide motivation to why you selected these thresholds to use in your clipping? 
• L193: And presumably, “clipping” means that data above these thresholds are considered “bad” and not used in future analyses (or included in the dataset you’ve archived)?
• L293: I believe you mean it is clear if there is no observed backscatter above some threshold; please indicate that threshold.  Also, what is the model of the ceilometer used in this study, and is the same model used on all three locations (BCO as well as the two RVs)?
• L295: I believe you again mean “flagged clear if the cloud radar does not indicate any reflectivity above X dBZ between 500m…”
• L301: for confidently cloudy, do you require both the radar and ceilometer see the cloud at (approximately) the same height?
• General question: how do you identify cases where the radar sees an ice cloud, and prevent these occurrences from affecting your statistics?
• L347: it would be good to reference a paper here that shows that the visible, infrared, and microwave measurements have different sensitivities to liquid clouds.  One possible example from the ARM program is Turner et al. BAMS 2007.
• L362: the single-channel retrievals have two aspects that result in the lower LWP noise values: (a) the IWV is fixed, and (b) the observation frequency is in the W-band, where the strength of the WV absorption is stronger.  You mention the latter at L381, but it could be mentioned here too.
• L383, and also at L362: if you included realistic uncertainties in the IWV used in the single channel LWP retrieval from the radars, it would result in larger uncertainties in the retrieved LWP.  But how much larger?  Would you please quantify that?
• L429: you need to include a reference to the 2 degrees of freedom for MWR temperature retrievals.  There are several papers by coauthor Löhnert that could be referenced here.
• L437: Would using a physical-iterative retrieval result in improved statistics, relative to your statistical retrieval, at this site?  It might be interesting to include a sentence about that (either here or in the discussion)

Again, very nice paper.