The paper focus on the potential use of commercial microwave links as scintillometers. In a first part, authors recall the basics of scintillometry, analyse and compare the scintillation signal and the deduced structure parameters from commercial microwave link with those from research reference instruments (Microwave scintillometer and Eddy Covariance data). For this first part a detailed control has to be done on text and formulas. They are both sometimes associated with small or large aperture hypothesis without specification of one or the other. Small aperture formulas are expected for Microwave links. This is confusing as the theoretical part doesn't match with the presented data. The reference cited could be more appropriate, using the original references rather than an handbook not completely dedicated to the scintillometry theory. This part ends with a clear result that should be relevant for publication showing the limitations of the CML devices to measure Cnn.  

The second part is less convincing as it is not clear at the end if the signal from the commercial link contains useful scintillation information or if the results are just a degradation of the signal from the MWS microwave research device, as the correction methods include the observed scintillation behavior from the MWS data. 

In method 1 S_noise seems to includes the power -8/3 decaying part which evolves along the day with the turbulence activity and cross wind intensity. For method 2 said to be built on method 1 it is also not clear what part of the MWS signal is included in the corrected CML data. It seems this method can be considered as a spectral modeling method using a transfer function based on theoretical spectra functions for different crosswind values, but it is not clear if the noise reduction of the first methods has been applied or not.

The overall description of both methods, even though it is "basic" methods, is not clear enough to help the reader understanding what is the quantity of the MWS signal is included in the correction (see detailed comments on the pdf). It is even not clear on which dataset the noise is caracterized, if an average noise (pre frequency bin) is removed for any half hour or if the noise is characterized for any half hour, and then not clear how the noise is removed in the 0.1 - 1 Hz frequency range using the 1Hz - 10Hz caracterisation.  

Separating the data set in two, with a calibration segment and an evaluation segment could make the study more convincing.

    

For both methods the impact of filtering and correcting should be better analysed. The reader has just some "hypothetical" plots which makes difficult the appreciation of the methods performances. For example, the noise removing procedure and sig2_noise estimation could be applied to the no-scintillation dataset (turned off transmitting antenna).

A better characterization of the noise accross frequencies for different conditions, especially with T°, windspeed, should help to consolidate method 1. 

  

For model 2, the identified cutoff frequencies have an unexpected non monotonous dependency with the crosswind values. This is not discussed. From spectra modeling, this behavior means that there is other dependencies in the dataset. More over, it is not compatible with the applied high pass filter.

The structure of the paper should be reorganised, putting together all the methodology parts, including correction methods. 

From this remarks the overall feeling is that no general procedure could be applied to the CML dataset, without a continuous MWS beside. I suggest the authors should re-organise the paper and their argumentation with the only objective to demonstrate that useful scintillation informations are included in the CML data, which I believe looking at the only spectrum presented. Then eventually demonstrate for which conditions scintillation informations are extractable from the raw data or not. 

Because important presented results deserve to be published, but thinking that methodology should be hardly consolidated, I suggest major revision for this paper. 

Extra detailed comments are included in the attached pdf.

see attached file