Soil Moisture Calibration Equations for Active Layer GPR Detection—a Case Study Specially for the Qinghai–Tibet Plateau Permafrost Regions

Round 1

Reviewer 1 Report

What to you know!?  It is all about water.  But great data, especially the data on organics and salts.  I am not exactly sure HOW this can be used now, but perhaps you will get to that later on. Or is this just some data presentation showing these relationships.   I would love a paragraph or two on how all this can be used.  

Author Response

We are very sorry for making you so confused due to the flaws of our manuscript, and sincerely thanks that you still provide us this precious chance to reply these confusions. The manuscript has been greatly improved with our best trying. The original title has been changed to avoid misleading readers. Abstract was rewritten again. At the end of introduction section, a general framework of this paper was added. In the discussion section, Figure 14 was added to further explain how the soil organics data were used.

The main purpose of this research paper is to provide an effective active layer soil moisture calibration equation that is suitable for GPR detection on Qinghai-Tibet plateau permafrost regions. In our research, a simplified Complex Refractive Index Model was employed under the effective hypothesis at GPR frequency range (about 100MHz). There are four unknown parameters within CRIM, including soil water permittivity (ɛ_w), soil geometry exponent (n), soil matrix permittivity (ɛ_s) and soil porosity (φ). These parameters cannot be estimated wholly with our field investigated data. The ɛ_w within active layer was firstly calculated based on the relationship between water dielectric constant, temperature and salinity. Then the second main parameter, n, was deduced utilizing the calculated soil water dielectric constants and the linear fitting slopes of field investigated active layer soil moisture and dielectric constant data set under different exponential case from -1 to 1. Finally, a simplified Complex Refractive Index Model was acquired through neglecting the influence of ɛ_s and φ variations on moisture estimation at regional scale. The main shortage of CRIM calibration equation is that its calculated soil moisture error will gradually increase with decreasing of GPR velocity and increasing of GPR velocity interpretation error. To avoid this shortage, a direct linear fitting calibration equation, named as υ-fitting, was acquired based on statistical relationship between active layer soil moisture and GPR velocity with our field investigated data set. When GPR velocity interpretation error is within ±0.004m/ns, the maximum moisture error calculated by CRIM is within 0.08m³/m³. While when GPR velocity interpretation error is larger than ±0.004m/ns, a piecewise formula calculation method, combined by υ-fitting equation when GPR velocity is lower than 0.07m/ns and CRIM equation when GPR velocity is larger than 0.07m/ns, was recommended for the active layer moisture estimation with GPR detection on Qinghai-Tibet plateau permafrost regions.

Reviewer 2 Report

I have been all the way through the paper. It is a good paper and a lot of work. The manuscript is well-written,

well-presented, and it follows a sound and logical approach

to process to collect and process data and generate conclusions. Figures and tables are clearly presented.

The manuscript is worth publishing. However, my major concerns:

1.Please demonstrate gap of literature and contribution of this research

2. Please explain why the improved technique is appropriate and, at least,compare it to a benchmark method(s). 

3. Please show why this technique more suitable for Qinghai-Tibet plateau permafrost not other regions.

4. Please present potential drawbacks of the proposed technique, and how it could be improved in the future studies.

Author Response

Thank you very much for your praise of our work and these important questions. We have studied these questions carefully and have made corresponding reply and correction within the revised version.

please see the attachment (Filename: Reply for reviewer 2) for details!

Author Response File: Author Response.docx

Reviewer 3 Report

Comments on “Complex refractive index model for active layer soil moisture measurement with ground penetrating radar on the Qinghai-Tibet plateau permafrost regions”

The manuscript presents many things: velocity estimation from GPR data and evaluation of its error, estimation of CRIM parameters, conversion from velocity to water content and its error analysis. Personally, the approaches used in estimating CRIM parameters and conversion from velocity to water content are interesting. However, covering all these points seems too much and made me confused. There are no new findings in the part that GPR measurement is described. The evaluation and analysis of error are not convincing enough. The material the authors are measuring is basically inhomogeneous, and the estimation error cannot be discussed without considering the inhomogeneity, in my opinion. The inhomogeneity, as well as some methods used in the paper, seem site-dependent and the authors must clearly state the fact in order not to give an impression that the method is general to readers. Revising the paper focusing on these points makes the paper more like a case study and it should be reorganised and rewritten from a bit different aspects. Therefore, my recommendation for this submission is rejection, but I encourage the authors to rewrite and resubmit the revised version considering the following comments.

The manuscript does not seem to be well organised, especially the sectioning and section/subsection headers. For example, “Materials and methods” section seems to have measurement results. Also, Section 3.1 has a header that sounds very general and does not represent what is described in the section.

Abstract: The first half of the abstract does sound introduction and essentially unnecessary. The second half sounds more abstract, but too abstract and cannot imagine what is presented in the paper. The authors should highlight more about the significance of the work, methods and results.

Introduction: The part sufficiently presents the background of the work, but it lacks what methods are used in what way to find what. Because there is no information about those, it is difficult to read and understand the following section. You should provide a rough framework of the work in the introduction to give readers an idea of what follows.

Table 1: There is no explanation of how “epsilon” in the table was obtained.

P4, L145: What the authors call BFWARR is essentially no difference to WARR. The point of the measurement is to use a fixed transmitting antenna and to scan only receiving antenna. This is not an innovation. Having said that, the part can be removed and a reference can just be provided.

P4, L148: Please explain why the authors scan the receiving antenna backwards in the second half (the part antennas goes away). It is of course up to you how you collect data, but if the way of scanning influences (e.g., improves) the results, the authors must explain why. If it is just the way that authors prefer with no reason, you should not mention it to keep the clarity of a method. It seems to me not to make any difference and only the first half should be shown to avoid confusion.

Figure 2a: The interface of the layers drawn in the figure is not flat. It must be modified to meet the assumption of WARR measurement and data analysis. Also, the assumption is clearly stated not to mislead readers.

Figure 2b: Why the data contain two sets of measurements? Please explain if it means something. Otherwise, please show only one set—it is just confusing.

Equation 2: it seems erroneous

P5, L163-: If a pit is dug, the soil is disturbed and it usually does not give the same permittivity any more. Therefore, the way of evaluating the error seems pointless to me.

How do you estimate the velocity from the reflection from a metal plate?  A hyperbolic curve is picked in Figure 3 and I assume that velocity is estimated based on the hyperbolic curve. Do you consider the size of a metal plate? Are you aware it influences the shape?

As the authors pointed out, one of the major sources of error in estimating velocity from GPR data is where to define the time zero. The problem exists in all WARR, multi offset and metal plate reflection. So, what is the point of comparing the results of these three and calculate error? I am afraid it does not represent the error of velocity estimation by GPR measurement.

Section 3.1: The authors tried to model the temperature variation of permittivity in order to obtain the parameters in CRIM as accurate as possible. Are the authors aware that permittivity is also dependent on frequency? The authors used values found in literature for modelling. Did you check the frequency the literature used and values are valid for the frequency (range) of your GPR measurements?

Equation 10: The authors call the equation CRIM, but it is no longer CRIM. The point of CRIM, in my opinion, is that it consists of the volume fraction of each phase. Also, as the name indicates, it is valid to complex permittivity, but Equation 10 is not.

Section 3.2: To be honest, I do not see the point of the subsection. You had to consider the error of estimating velocity because it is from GPR measurements. However, you dug pits and sampled soils, why do not use permittivity of sampled soils measured by a more accurate way. It should be or is assumed to be exact, unlike one from GPR.

Before doing error analysis, you should state how much accuracy is required, in this case, in studying permafrost. However, the authors need to consider inhomogeneity of soils, which gives the spatial variation of velocity/permittivity/water content and it should be considered different to measurement error. My impression is the inhomogeneity gives variation in water content more than variation caused by error in the estimation of permittivity and CRIM parameters. In such a case, it does not make much sense to accurately estimate CRIM parameters, unfortunately.

As the authors are probably aware, velocity and water content two are not directly related—velocity is nonlinearly related to permittivity, and permittivity is nonlinearly related to water content according to CRIM for example. The authors obtained a linear relationship reasonably, because, I assume, these two nonlinearities coincidently cancelled out. That means that the linear relationship is coincidently obtained for your samples, and it most likely still nonlinear for soil samples from other places. Therefore, the method may be valid for this specific site, but it is not valid in general. The authors should present the results with care and explicit explanations in order not to mislead readers.

Figure 9: It is not clear how they are calculated and what they mean.

Section 4: The discussion on organic content sounds reasonable. But, we all know the parameters of CRIM are very sensitive to soil texture and grain size distribution. Please add some additional discussion from these aspects to make it more comprehensive.

English is good enough to understand what the authors want to say in most cases. There are some part that difficult to understand because of English expressions. It would make a much well-written paper if language editing service is employed.

Author Response

I’m very thanks that you must spend a lot of precious time to review this manuscript. You also provide many important and professional questions and suggestions. We have studied these comments with our best trying and have made corresponding answers and corrections within the revised manuscript, which we hope can meet your expectation.

Please see the attachment (Filename: Reply for reviewer 3) for details!

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

    I should extend my thanks to authors for taking into accounts my comments. The paper has been improved substantially, hence, I would suggest "accepted in preset form".