Review of “Atmospheric sounding of the boundary layer over alpine glaciers using fixed-wind UAVs” by Groos et al.

This manuscript describes the use of a fixed wing UAS to profile the lower atmosphere over a mountain glacier environment. Since data from only a single day is shown the results here are illustrative of the types of features that could be observed with a UAS field campaign but do not allow for any broader conclusions about glacial meteorology. While the results presented will be of interest as an illustration of the potential research applications of using a small UAS to study alpine glacier meteorology the presentation requires major revisions as described in the comments below. Once these major revisions are completed the manuscript will be suitable for publication in Atmospheric Measurement Techniques.

Major comments

A figure or series of figures illustrating the data processing described on pages 9 and 10 should be included to illustrate what the raw, unprocessed data from the UAS looks like and how that is modified prior to further scientific analysis. This figure(s) should show:

-  unprocessed T profiles and the final smoothed profiles in 1 m bins.

-  temperature bias between ascent / descent legs averaged to 1 m bins

- illustrate how lapse rates and inversion height are calculated by showing profile of T’(z). In particular I am interested in seeing how noisy the T’(z) profile is and what impact this has on identifying the height of the SBI.

- show RH, T and derived q profiles

- show profile or time series of original resolution roll rate and derived turbulence intensity proxy

These figures showing original data and derived data used in the results section will allow the reader to clearly see how the data was modified to allow for subsequent scientific analysis.

I found the color shaded time-height plots to be attractive but ultimately not very useful for understanding the features present in the UAS observations. I strongly suggest that the authors replace these figures with single plots for each each variable (T, q, wind speed) showing all descent profiles from all flights. By showing all of the profiles on a single plot it will make it easier to see details in the change in magnitude and vertical structure over the course of the day  than the color shaded cross-sections currently shown. To help interpret the time evolution shown in this plot each descent profile should be shown in a different color (maybe ranging from blue to red with increasing time of day).

Showing profiles of wind direction, in addition to the wind roses shown in Figure 12, would make it easier for the reader to see the relationship between the switch from down glacier to up valley wind  direction and differences in speed.

It would be useful to show a synthesis plot at a representative time showing profiles of all of the analyzed variables together to illustrate how the different variables and their profiles relate to each other.

What is the explanation for the nearly linear lapse rate for profile 1  down to the lowest observed height in the 16:05 sounding in Figure 6? This differs from all of the other profiles and is markedly different from the profiles at adjacent times. Is this an observational error or a real feature of the atmosphere?

Uncertainty in the observed quantities and the impact on interpretation of the results needs to be included. In particular, what is the uncertainty in the derived wind speed and direction and does this alter the interpretation of the results. In particular I am wondering about the rapid shift in wind direction and how this is handled if the spiral path used to calculate wind speed and direction spans both down and up valley wind directions. Does this account for the low wind speed at the height of the change in wind direction (i.e. it is an artifact of how the wind is derived rather than a true feature of the wind profile?).

Minor comment

Lines 110, 123: Figure 2 should be figure 3

The manuscript presents one day of profile measurements (8 individual measurement flights with 2 profiles each to an altitude of max. 400 m above the ground) with a fixed wing UAS over a Glacier surface. One of the stated main goals of the study is the characterization and investigation of the structure and development of the katabatic jet over a glacier, a phenomenon that typically extents only a few tens of meters above the surface, and can by that not appropriately probed by profile measurements with a fixed wing UAS. Here multi-rotor drone systems (preferably with forced ventilation of the temperature and humidity sensors) with their capability of hovering and ascending/descending very slowly, would be the by far better choice. The manuscript covers a lot of different topics (a bit of system description, a bit of methodology, a bit of scientific evaluation), but is not going deep in any of them. The most interesting and novel part of the study, the estimation of a turbulence proxy, is again only touched and not described (it is referred to a separate publication). Integrating that part in the presented manuscript would clearly help to make it publishable, in its present form it is too thin and lacking novelty. 

However, I see a clear potential of combining the presented one day of measurements with additional measurements in comparable environments, or as mentioned above include the detailed description of the turbulent proxy algorithm.

Some additional comments that hopefully can support a new submission of the manuscript:

Introduction: there is a vast discrepancy between the identified scientific gaps and goals and the very limited measurement capabilities of the presented/used UAS; as mentioned in the general comments fixed wing is not appropriate for the shallow katabatic jet; and the ceiling altitude of max. 400 m is far from sufficient to investigate the larger scale valley wind systems in the study area

line 138: "the sink rate was low to minimize....."; you should quantify this; I would even suggest that you also plot the profiles of vertical velocity for all descents to get a feeling how constant this is and which effect it could have on the measurements due to the sensor time constant

data processing and analysis: You state that met data and flight data are collected/stored separately. Are those data sets in any way synchronized by a common clock/timer?

Line 161 and Appendix A1: you apply a simple shift in temperature (of about 0,4 C) that you gained from one day of parallel measurements at an ambient temperature of around 25 C; there should be at least a second comparison been done at a distinctly (ideally close to operational conditions over the glacier) lower temperature to check for potential gain errors, that would cause a temperature dependent change of the differences.

Figures 6/9/11: it would be so great to have a ground value (e.g. by a simple weather station placed close to the UAS start/landing site) to verify/validate the very strong gradients that often occur close to the ground in your observations

Figure 12: it feels inconsistent to present the wind information as wind roses, while all other parameters are given as profiles

References: Groos et al (line 429); inconsistency in abbreviation of journal name