Dear Authors,
both the reviewers have made a thorough review of your manuscript, and both raised serious doubts about the correctness of volumetric water content data indirectly retrieved through dielectric permittivity measurements when part of the water is forzen, as well as about the discussion of the obtained results, in which no clear distinction between liquid and frozen water was made.
At the same time, both the reviewers recognize some merit in the manuscript, so my decision is to reconsider it after major revisions.
I have read your point-to-point answers, and I can see that you are going to address all the issues raised by the reviewers.
I add here a few comments of my own:
1. Regarding the calibration of indirect measurements of soil water contant based upon dielectric permittivity, you should consider that also the permittivity of liquid water is sensitive to temperature variations: the value of 80 which applies at 20°C grows up to 88 at 0°C. This implies that both the specific calibration of FDR and Topp's polynomial that you use for TDR have additional uncertainty when applied at low temperatures.
2. I still find difficult to understand the supposed process of melting from below: although temperature variations in the soil are obviously more delayed at higher depths, it remains true that the amplitude of the variations decreases with depth. So, it is possible to have temperatures above melting point in depth and below freezing point in shallower layers, but only when the soil is refreezing after the summer season (the temperature is decreasing above, and is still high, or maybe even increasing, below). But the melting process in any case starts from above, and it should be clear in the discussion.
3. About your statement, criticized by both the reviewers, that under dry conditions you observed quicker infiltration, I would remind that the infiltration capacity of a soil is controlled not only by the unsaturated hydraulic conductivity (obviously smaller in drier conditions), but also on the vertical gradient of water potential, which can be very high downward when the soil is dry, so that usually the infiltration capacity is maximum when a soil is dry, and then reduces. So, before invoking preferential flows, often seen as the "miracle cure" to all problems, I would carefully check if the measurements (and the characteristics and even the visual observation of the topsoil) actually give evidence of the possible activation of by-pass flow. In other words, I would better motivate statements as "indicative for preferential flow processes", and similar.
Having said this, that I hope can help while writing the revised manuscript, I look forward to receiving it.
Best regards
Roberto Greco

Dear Authors,
both Referees acknowledge the substantial improvment of the revised manuscript, but while Ref #2 recommends publication, Ref #1 still raises several issues.
In fact, I agree that the structure (section 3), the language (overall), and some parts of the discussion (e.g. the definition of the threshold between frozen/unfrozen conditions, the discussion of the accuracy of the measurements...) still need to be improved before publication.
In particular, as alternative explanation of the different behavior observed at DRE is also plausible (as suggested by Ref #1), I invite you to revise that part of the discussion, to better support (or, even, partly correct) your interpretation.
However, I am quite confident that such further improvement is feasible, and I look forward to receiving the new version.
Best regards,
Roberto Greco

Dear Dr. Pellet,
I am pleased to inform you that, on the basis of the positive evaluation of the Referee, I recommend to publish your revised manuscript as it is.
Best regards,
Roberto Greco