Vis-NIR Reflectance Spectroscopy and PLSR to Predict PCB Content in Severely Contaminated Soils: A Perspective Approach

Round 1

Reviewer 1 Report

This paper presents a fascinating and valuable method of predicting PCB levels in the soil. Along with the objectives, there is a clear definition of the problem. The novelty of this paper can be fully justified. Some parts of the materials and methods need to be explained more clearly, as suggested in the comments. Findings become more interesting if they follow discussions from other similar papers. In general, the article is worth publishing after a few minor revisions. 

Comments:

Title: Vis-NIR reflectance spectroscopy and PLSR to predict PCB content in severely contaminated soils. A perspective approach. Could you please double-check the "." And change it to ":"? Also, perspective or perspective?

Line 26: Should it be "prediction of the considered contaminants" or "prediction of the considered contaminant"? Please write more precisely.

Line 37: Please first state the full name and then the abbreviated form. The manuscript should be checked thoroughly for such an issue as it is repeated frequently.

Line 42: Could the authors address MATRA? Is it the name of the place? Maybe the first full name should be stated if this is not a well-known abbreviation.

Line 45: Please start the new paragraph for "Usually, the …"

Lines 70-75: The connection of the use of different calibration techniques to Vis-NIR reflectance spectra is unclear. Could the authors try to improve the paragraph for more celerity? Also, the choice of PLSR over other methods must be discussed.

Figure 1: Can the authors replace the map with the colorful one?

Section 2.1: The information regarding the physical and chemical properties of the soil is missing. More importantly, what was the type of the soil based on the international classification?

Section 2.2. Please add the name of the software.

Lines 128-129: Please avoid making any paragraph containing only one sentence.

Section 2.4: It is unclear how this section addressed Soil properties as mentioned in the subheading title.

Lines 213-216: It is suggested to transfer this information into materials and methods.

The results of the extraction of the PCs, such as Eigenvalues and Percentage of Variance %, must be summarized in one Table before discussing the results of PCA in Figure 2.

Lines 232-242: Please discuss the reasons for your observation from the literature.

Caption in Figure 4 & Figure 5: Please avoid using Figure 3 in the caption.

Figure 4: Please highlight the regions where differences in the overall reflectance are noticeable by creating boxes or arrows.

Figures 5 b and c: Please use the unit and heading for the horizontal and vertical axis.

Lines 284-289: Please cite properly for any assignment of the peaks.

It would be more interesting if the authors discussed the possibility of using their suggested method for other case studies? Is it only working for severely contaminated soils? Discussing how this model could be relevant to the broader context would add extra benefits to the paper.

Section 4. Conclusions: Please remove all the references from the conclusion section. The reduction in the length of this section is highly required.

Author Response

see the attached file

Author Response File: Author Response.docx

Reviewer 2 Report

Comments on Manuscript applsci-1787270 “Vis-NIR reflectance spectroscopy and PLSR to predict PCB content in severely contaminated soils. A perspective approach”

 

This manuscript by Leone et al., describes the development of an infrared spectrometry method to measure PCBs in soil samples. Although the idea of the study is interesting and could be a major change on how we detect and measure PCBs in the environment, there are so many variables in soil that could affect the final results, at least with the samples used. Thus, the described method cannot be generalized, which I believe is one of the main goals of this study. For example, what about if the soil is also contaminated with other halogenic compounds, how can the method separate those? The PCBs18 and EOX results look very similar, but is it due to only PCBs or are there other compounds driving these results? I’m also skeptical of the results due to the low number of individual PCB congeners used in this study. The authors only measured 18 individual PCB congeners, which is roughly 10% of the total 209 congeners. Is that a representative number of congeners? Further, if the authors are trying to develop a new method, they should try to reduce the number of variables or set up more controlled experiments. A better study design would have been to spike “clean” soil with different known PCB concentrations and develop the method. Furthermore, there is no discussion about detection limits, or how low the PCB concentration in the soil this method could detect. And what about the distribution of PCBs in the soils? It seems that the method works better for high chlorinated congeners, but if the soil is contaminated with a “light” Aroclor (i.e., 1016), or even a dechlorinated Aroclor, will it work? I think the results presented here are very specific for the soil used and it is not clear if it will work for other soils and other type of contamination. The only caveat that I found in the manuscript was the low number of samples, but as I listed here, there are so many more than that. Thus, I cannot recommend this manuscript for publication. I also have the following small comments:

 

Line 45. I’m not familiar with the term pedo-environment. Perhaps the authors could elaborate more on this or use another word.

 

Line 108. “376” seems to be a typo.

 

Lines 184 – 186. Perhaps the authors could include the equation.

 

Line 203. I would suggest to the authors to use no more than 3 significant figures in their numbers.

 

Line 207. Perhaps the authors could elaborate more on what they mean with “soil surface movement”.

 

Table 1. If the authors are only using 18 individual PCB congeners, it is not correct to talk about homolog groups. There are too many missing congeners in each of those homolog groups. Perhaps they can include the number of congeners in each of the homolog groups listed in the table.

 

Section 3.2. It is not clear how the data were used in the PCA. Is it distribution (e.g., fraction or %) or concentration? It seems that the data are concentrations, just looking at PCBs18 (total concentration). I have always seen that when PCA is performed with PCB data, the fraction or distribution is used, so the samples are normalized and the sum of the congeners in each sample is 1. Perhaps the authors need to explain why they used concentration instead of fraction or percentage.

Line 252. “Hesa” should be Hexa.

Line 293. Eviadently should be Evidently.

Figure 6. Some plots have log10 scale and other normal scale in the x-axis. This looks like the authors selected the scale to yield the best fit of the regression. This is not very scientific and shows that the regression works depending on how the data are manipulated. I recommend to the authors to explain why they did that and just use one scale.

Line 451. Here is the only caveat of the method described by the authors, low number of samples. As I have mentioned above, there are many more caveats of this study that at least need to be included in this manuscript.

Author Response

see the attached file

Author Response File: Author Response.docx

Reviewer 3 Report

The quality of presented manuscript is quite high and the obtained results are appropriate explained and interpreted in a comprehensible manner.

Page 4 line 131-132 "two days at a room temperature of about 24 degrees to eliminate any possible residual moisture..." I am not sure that this formulation is clear. I guess that this is not a way to eliminate moisture but it is a way how to reduce moisture to an acceptable level.

Table 1

PCBs18 MAX 16991.28  two decimal plates for this value make no sense with respect to uncertainity. The same for some other values in table 1.

Author Response

see the attached file

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors have addressed my observations. However, I'll have to disagree with them about only measuring 18 congeners. As I mentioned before, there are so many more congeners, including non-Aroclors or congeners not present in commercial European mixtures. It is my opinion that measuring just one or two congeners per homolog group and then reporting them as Tri-, Tetra- is not correct. That is why I indicated to include the numbers of congeners per homolog group reported in the manuscript. 

And the same way the authors responded to one of my observations, please google "significant figures".

Author Response

See the attached file.

Author Response File: Author Response.docx