Near-Infrared Spectroscopic Study of Heavy-Metal-Contaminated Loess Soils in Tongguan Gold Area, Central China
Round 1
Reviewer 1 Report
The authors have tried to follow the instructions and have made a number of improvements. There are still some problems with the wording that sometimes make it difficult to understand the content of the text (cf. the revised sentence in the abstract, sentence in line 184, line 216-27,…).
Therefore again my request to revise the work linguistically. Since this is not a scientific problem, but rather a linguistic problem, it should be sufficient for the editors of the journal to evaluate a correction.
However, there are still things that should not appear in a scientific paper like the uncritical use of too many digits at some places (see fig. 3 – 5 or line 321). How could the authors state (L294) “All along the variety of the heavy metal content, the combination bands of (ν+δ)OH shifted gradually from 4530.33 cm-1 to 4559.88 cm-1.” when the spectral resolution in the shortwave infrared (region from 1300– to 2500 nm) region is 10 nm and the sampling interval is 1 nm (L136). Another detail – the spectrometer performance is provided in nanometer units but spectra in wavenumbers. But to illustrate the problem: a spectral difference of 1 nm (which is the sampling interval, NOT the spectral resolution) in this spectral range already corresponds to 2 cm-1 (2207 – 2208 nm equals 4531 – 4529 cm-1).
The algorithm of baseline correction sounds cumbersome and it is not possible to recognize the importance of the Voigt profiles.
Please let me also cautiously point out that multi-linear regression is a dangerous method. The disadvantage of the method is noticeable with collinear X variables, which are especially common with spectra. That is why most scientists prefer to use partial least squares regression (PLSR) or principal components regression (PCR).
Author Response
The authors have tried to follow the instructions and have made a number of improvements. There are still some problems with the wording that sometimes make it difficult to understand the content of the text (cf. the revised sentence in the abstract, sentence in line 184, line 216-27,…).Answer: The manuscript has been edited by a native speaker of the English editing service, MDPI.
However, there are still things that should not appear in a scientific paper like the uncritical use of too many digits at some places (see fig. 3 – 5 or line 321). How could the authors state (L294) “All along the variety of the heavy metal content, the combination bands of (ν+δ)OH shifted gradually from 4530.33 cm-1 to 4559.88 cm-1.” when the spectral resolution in the shortwave infrared (region from 1300– to 2500 nm) region is 10 nm and the sampling interval is 1 nm (L136). Another detail – the spectrometer performance is provided in nanometer units but spectra in wavenumbers. But to illustrate the problem: a spectral difference of 1 nm (which is the sampling interval, NOT the spectral resolution) in this spectral range already corresponds to 2 cm-1 (2207 – 2208 nm equals 4531 – 4529 cm-1).
Answer: We have remove some digits in Fig.3, Fig.5 and L321 according to your advice. The problem you figured out are very correct. The wavenumbers were calculated by the wavelength according to “Wavenumber= 1/Wavelength*10e7”. The sampling interval is 1 nm and the spectral resolution is 10nm. The absorption peaks were fitted using Origin software and the variation of the peak positions were from 4530.33 to 4559.88 cm-1.
The algorithm of baseline correction sounds cumbersome and it is not possible to recognize the importance of the Voigt profiles.
Answer：Our English description is not clear for many procedures in the study. The English Editor in MDPI, who is a native speaker, have helped us to correct our manuscript.
Please let me also cautiously point out that multi-linear regression is a dangerous method. The disadvantage of the method is noticeable with collinear X variables, which are especially common with spectra. That is why most scientists prefer to use partial least squares regression (PLSR) or principal components regression (PCR).
Answer: We tested the correlation between the six heavy metal variables and the 6 heavy metals using PLSR according to your advice. The result showed a relatively notable correlation.
Reviewer 2 Report
The manuscript has been improved and hence, recommend for publication except for minor English language error.
Author Response
The manuscript has been improved and hence, recommend for publication except for minor English language error.
Answer: The manuscript has been edited by a native speaker in English Editorial Service of MDPI.
Reviewer 3 Report
It is really difficult to see a correlation between the six heavy metals, the mineral profile and the NIR/MIR bends. On my opinion the work will be much clear if the authors avoid to present the metals results. From a scientific point of view heavy metal absorption and desorption processes are very complex, so during the development of this work many variables are apparently not controlled.
Author Response
It is really difficult to see a correlation between the six heavy metals, the mineral profile and the NIR/MIR bends. On my opinion the work will be much clear if the authors avoid to present the metals results. From a scientific point of view heavy metal absorption and desorption processes are very complex, so during the development of this work many variables are apparently not controlled.
Answer: We have removed the heavy metal results according to your advice. The problems you figured out are all very correct. The adsorption and desorption processes of the heavy metals in soils are very complex due to its complex containing and environmental factors. Our soil samples have low organic matter content, high clay content and little change in pH value. Therefore, the main change comes from the adsorption of clay minerals. But other factors still affect the spectral characteristics, resulting in low correlation coefficient. The aim of this study is not use this method in laboratory, but to find a rough trend between the spectra and the heavy metal, and use it in remote sensing and in-situ investigation.
Round 2
Reviewer 3 Report
In the new revised version still heavy metals results are reported and mentioned in the discussion and conclusions sections.
Author Response
In the new revised version still heavy metals results are reported and mentioned in the discussion and conclusions sections.
Answer: Thank you very much for your carefully review of our manuscript. We have removed the heavy metal results in the discussion and conclusions sections according to your advice.
Round 3
Reviewer 3 Report
On my opinion the work could be further improved giving more details about the analytical methods employed and going deeper in the discussion trying to better clarify the chemical processes that are related to the mineralogy of the studied soils, the capability of those to retain contaminants and clarify how the anthopogenic activities are contributing to the formation and pollution of the stratigraphic layers.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
The article 7637018 by Min Yang*, Youning Xu, Jianghua Zhang, Huaqing Chen, San Liu, Weiliang Li and Ying Hao presents a spectroscopic study on loess soils from the Northwest of China. The study aims to identify the adsorption of heavy metal cations, e.g Cu, Pb, Zn, Cr, Cd and Mo onto illite, smectite and illite-smectite (I/S) mixed layers by NIRS. For the classification of soils into the categories moderate, medium and highly polluted by the heavy metal cations the Nemerow index is used. The authors characterize the soils by X-ray diffraction (XRD), Mid infrared spectroscopy (MIR), and Inductively Coupled Plasma Mass Spectrometry (ICP-MS) to explain to explain the spectral features observed by NIRS.
In the introduction the authors refer to various modern data analytical tools, however the evaluation of spectral data is limited to non-linear curve fit and the overall discussion of the results relies on linear regression. From my point of view, plenty of information has been gained, but neither evaluated nor discussed adequately.
This article cannot be published in its present form but needs major improvement concerning both, presentation and analysis of data. I would expect sort of global data analysis including results from all of experiments. Moreover, style of writing and language must urgently be improved.
Details:
The results of XRD-analysis are not discussed at all, the relation to the N(IR) data is missing The heavy metal element analysis ends by the calculation of the Nemerow index. I would like to see at least the trial to correlate individual results with the other spectroscopic data. Do the samples contain organic matter? pls. provide data on organic carbon content of the samples. In case organic matter is contained it will contribute to the NIR spectra as well. Moreover, both NIR and MIR spectra will be affected by the protonation state of the material (i.e. the pH values of preparation) Section 2.5: this is the basis for all spectral analysis and must contain more details on the type of baseline, smoothing algorithms, number of smoothing points, degree of polynomial etc. Why did the authors use a Gaussian model instead of Voigt profiles? How was the fit initialized, how many fitting steps were used? pls. provide residuals. 3: Could you attribute the obtained profiles to certain vibrations? Pls. start numbering the spectral axis at high values, i.e. from 4000 – 1000 cm-1. 3 – Fig. 5: Frequency is not the correct label! P6, l184-186: pls. provide a helpful illustration for your statement Table 1. It would be helpful for the reader if you found more descriptive names for your samples. If you like to stay with numbers, for example you could call it S01 – S25. The names should further include information on the material (illit, smectite, I/S mix). Pls. provide an overview on all spectra for each method in the Supplement. 6: grey print has poor quality, pls. improve graphical quality AND provide more explanation in legend; a figure should be understood without having to refer to the main part of the article. 7: Linear correlation (not relationship) of band frequency and Nemerow index. As pointed out previously, I suggest a more elaborate evaluation of data. The result looks perhaps like a trend but NOT like a true correlation. If you leave out the extreme value at 4530 cm-1, you will end with a cloud of data and R2<0,5. Instead of the Nemerov index it could be worthwhile to test for the individual cations as well (with a multivariate approach that will be possible). L306-308: wrong statement, both MIR and NIR are nondestructive!Minor remarks
P2,l 68: absorbability? P3,l103: pls reconsider mesh size, number of valid digits Section 2.1: is this a standard procedure published elsewhere? If yes, pls. provide reference Section 2.3: “shortwave infrared 1300-2500 nm” (unusual expression): 4000 – 28571 cm-1: this is NIR + Vis P 4, l148: pls explain what is meant by “objective” concentration, this is important to understand the Nemerow indexReviewer 2 Report
Good manuscript but needs improvement on the scientific content. Here are some of the issues of concerns:
The introduction is very shallow and needs to be improved particularly on the subject matter (Loess soil) with their unique characteristics. The objective concentrations were not defined for the elements selected in the study. What is the justification for the selection of the studied elements? How many times where the soil sampled and at what depth. Kindly revisit table 1 and presents the mean value ± SD/SE. Also, what are the quality assurance and control measures employed in this study? In table 1, sites C170410003, C170860007, and C170440003 were missing but were presented in figure 2, why?
The results and discussion need to be improved. There are a lot of assumption and speculations which needs to scientifically justify. Some data are missing in table 1 and some are presented.
As frequently mentioned in the manuscript, that at higher frequencies, there is a decreasing of the Numerous index, this was not established in the study. The correlation coefficient was 0.4 and shows a weak positive correlation between the parameters.
A statement such as "The regression result confirmed that the hydroxyl combination band in NIR spectra((ν+δ)OH) near 4548 cm-1 can be attributed to the combination of OH stretching and bending in [AlAl]O-OH group (ν[AlAl]O-OH]) and [AlAl]O-OH group (δ[AlAl]O-OH) in the MIR spectra" need to be rephrased.
The conclusion must be clear and concise, please address that.
Reviewer 3 Report
I have some main concerns, it is really difficult to understand how the IR spectra are related to the ICP-MS results. IR analyses are not able to see soil metal compounds at ppm levels as it is shown in this work. The only reference in the discussion to metals is the Fe-Mn oxides. On my opinion it is really important to make the readers understand how IR spectra are related with the metals amount.
