The paper represents a rocket-borne technique for measurement of winds and temperature in the middle atmosphere between 20 and approx. 55 km.

The temperature measurement technique uses a thermistor and has been known since the early 1960th and was widely used in the last century. 

Wind measurements are made by making use of the GNSS BeiDou.

Additionally, the system measures atmospheric pressure by a "silicon piezoresistive manometer". The measured data is transmitted to the ground by a built-in telemetry system. Authors claim that the density can also be derived from those measurements by using the ideal gas law: p=nkT, where p and T are the measured pressure and temperature, k is the Boltzmann constant and n in the number density of the ambient atmosphere. However, the density measurement results do not appear in the paper.

The paper is difficult to read because of a very specific language used alongside not commonly accepted terms. The results are clearly summarized, however not well-supported by the arguments.

After a brief introduction of the measuring instruments in Sec. 2, temperature retrieval including error analysis, is described in detail. Also, measurement error for wind soundings is addressed in this section. 

After that, the paper is focused on derivation of small-scale temperature and wind fluctuations and comparison with the satellite-borne SABER measurements and MERRA 2 reanalysis data.

Since the manuscript is submitted to the AMT, I expected to learn more about the measurement technique itself, the instruments and data retrieval. A detailed description of the instrument is not present in the paper. In particular, which kind of sensors (thermistor, pressure sensor) were used. Either these were commercially available parts or self-made sensors. This knowledge will further help to clarify two important questions: resolution and sensitivity of the measurements. The paper by Wagner, 1961 sited in this manuscript, concludes that "Time constant and fall velocity considerations prohibit the measurement of small-scale temperature fluctuations".  I accept that the new sensors can be more sensitive and much faster. However, this should be shown in this manuscript, since the main focus is made on the fluctuation measurements. Also, no information on the telemetry requirements  (transmitter, receiver, bandwidth, sampling rate, etc.) appears in the manuscript.

The main focus in the entire paper is made on the fluctuation analysis, with an attempt to demonstrate that these measurements are suitable for investigation of small-scale processes like gravity waves and turbulence. This part looks to me neither consistent nor complete.

In Sec. 5 authors focus on deviations of measurements (wind and temperature) from MERRA2, SABER, and MSIS. This makes no sense to me. The general comparison in the left panels of Figs. 5 & 6 looks very convincing and sufficient. 

However, the next sections deal with fluctuations. Their extraction method is described in the text: 20-point sliding average, interpolated profile with 50 m interval, and fitted fifth-order polynomial to get the background profile. Additionally, high-pass filtering with a cut-off wavelength of 10 km is applied.

This decomposition into background field and fluctuations needs a demonstration by proper plots, which is not shown in the paper (neither background nor fluctuations). Also, the missing data on resolution and sensitivity of the measurements would help to judge on the quality of the derived fluctuations.

Next, spectral analysis is applied to the derived fluctuations. The way it is done looks strange to me. The same time series was analyzed by both the Lomb-Scargle (Fig. 10) and the Fourier (lowest panels of Fig. 11) analyzes. They should demonstrate quite similar picture, which is not the case here. It is not clear to me where such a difference comes from and why authors need the Lomb-Scargle analysis. Why, in particular, the Fourier spectra do not reveal the same dominating wavelengths as shows the Lomb-Scargle. Also, the speculations about dominating slope in the spectra do not look convincing: there are different slopes in those spectra, better pronounced in the right panel. Another question is: what the noise of the instrument itself look like? If the instrument (thermistor) reveals, e.g., red noise, the slopes in spectra demonstrated in the manuscript cannot be interpreted so directly.

It is also quite commonly accepted now, that the gravity waves often propagate as wave packets along very different pathways (directions, inclination angles, etc.). This leads to an observational challenge, that they rather appear in a small part of the measured altitude profile. These wave packets can better be localized by the wavelet analysis.

To my knowledge, the Stokes parameters derived in the Sec. 7.2 yield a relatively coarse description of the gravity waves in comparison with other methods. The measurements described in this paper (zonal and meridional winds and temperature) allow applying a more advance and precise analysis technique, the hodograph method. 

To summarize. I suggest improving the description of the instrument by adding more technical details. If authors want to make a fluctuation analysis, important parameters are the resolution and sensitivity of the measurements, as well as the deduction method of fluctuations. 

See attachment.