An Effective Prediction Approach for Moisture Content of Tea Leaves Based on Discrete Wavelet Transforms and Bootstrap Soft Shrinkage Algorithm
, Junjie Ren 1, Fanguo Zeng 1 and Enli Lü 1,*Round 1
Reviewer 1 Report
Manuscript ID: Applsci-857533
Title: An Effective Moisture Content Prediction Approach Based on DWT and BOSS Algorithm for Yinghong NO.9 Tea Leaves.
Authors apply a combination of two novel or not widely used variable selection techniques: discrete wavelet transforms (DWT) with the bootstrap soft shrinkage algorithm (BOSS). They apply these techniques in order to select a representative group of wavelength into the Yinghong NO.9 tea leaves spectra. Then they use these representative wavelengths for developing PLS models for predicting moisture content in the samples. Moreover, they compare the results with other well-known wavelength selection methods.
For me, the proposed study is of great interest to readers of Applied Sciences. However, there are some important considerations (mainly in the MM section) that make me suggest a major revision.
Title
In my view, there are some problems with the title. You should not use abbreviations. Furthermore, it is too long because it includes information that is not really important in this study. The parameter you measure and the specific sample you use are not the main topics of this study. I propose a title similar to: “Wavelength selection by discrete wavelet transforms and bootstrap soft shrinkage algorithm in near infrared spectra of tea leaves”.
Introduction
In general, the Introduction section is clear and precise. It stablishes the importance of the product and of controlling its moisture content. In addition, it deeply describes, sometimes too deeply, some wavelength selection algorithms that are widely used. I recommend to express the aim of this work in a clearer way at the end of this section. Furthermore, some specific suggestions are:
Line 41: This sentence is grammatically incorrect, does not? I suggest: “This procedure damages the samples and it is time-consuming”.
Lines 65-67: Here and in the previous paragraph, you are assuming that all NIRS is homogeneous. There are thousands of examples where denoising and variable selection is not necessary. These procedures are highly recommendable in lots of cases, but not always.
Lines 74 and 88: It would be variable and not variables.
Material and Methods
Material and methods section is almost an extension of the introduction. On most occasions, you simply describe in theory the methods used. However, you do not describe how you implement these methods in your study. Following this material and methods section is completely impossible to replicate the study.
Section 2.1. It is necessary to deeply describe the spectral acquisition. How were the samples and the spectrograph placed? How many samples were recorded? What were the differences between the samples? How many times were each sample measured? Etc.
Line 113: When were the spectra acquired? Immediately after they were placed at the drying tank?
Line 116: Is the resolution homogeneous in all the spectral range?
Lines 116-117: What the spectral acquisition number is? You mean that you acquire 64 spectra per sample by rotating 25 g of sample? Can you explain it better?
Line 119: These groups of tea leaves have not been defined yet. The manuscript needs of a clearer description of the sample and the subsamples created. How many samples have you measured? How many spectra have you used for doing the wavelength selections?
Line 122: Were the samples also summited to spectral characterization after the drying process?
Section 2.4: In material and method section you can indicate some theoretical information about the different methods. However, the most important is to describe how you applied these methods in order to these methods can be reproducible by the readers. Please add this information. (Software, parameters, etc.). This also occurs in other subsections of the material and methods section.
Line 144: Delete “of”. “BOSS is a method using collinearity…” or “BOSS is a method that used…”
Line 147: “…random combination of variables…”
Lines 151-153: Check commas.
Lines 172-173: Check.
Sections 2.5.3. and 2.5.4: Describe how you applied GA and iPLS please.
Model Evaluation Indexes
This section is not necessary a priori, however, due to the great complexity of the chemometric methods applied, it may be useful to have these fundamental concepts as clear as possible.
Results and Analysis
Line 203: Although you are not strictly following the sections recommended by the AS guide of authors, I think it is better to use discussion in order to follow this guide.
Line 205: It is the first time you use this term in the whole article. Please define it here or where it corresponds.
Lines 208-211: It is more recommendable to use the third person in scientific papers.
Please, check the format of this section. There are several format errors.
Table 2: Keep only the first sentence as the title of the table. The remaining text is a table note. Please place it at the bottom of the table.
I have no suggestions about the scientific aspect of the Results and Discussion section. I found the results interesting. The combination of DWT and BOSS seems to be a good alternative for the selection of variables in spectral matrices. Please check the format of the whole manuscript and correct the formal aspects in the Material and Methods section.
Finally, I think authors have to add some sections, such as Author Contributions, Funding, Acknowledgments or Conflicts of Interest. Please, check the AS guide of authors.
Author Response
Title
In my view, there are some problems with the title. You should not use abbreviations. Furthermore, it is too long because it includes information that is not really important in this study. The parameter you measure and the specific sample you use are not the main topics of this study. I propose a title similar to: “Wavelength selection by discrete wavelet transforms and bootstrap soft shrinkage algorithm in near infrared spectra of tea leaves”.
Response: Thank you for your valuable suggestions, I will revise it according to your opinion. The revised title is ‘An Effective Moisture Content Prediction Approach Based on Discrete Wavelet Transforms and Bootstrap Soft Shrinkage Algorithm of tea leaves’.
Introduction
In general, the Introduction section is clear and precise. It stablishes the importance of the product and of controlling its moisture content. In addition, it deeply describes, sometimes too deeply, some wavelength selection algorithms that are widely used. I recommend to express the aim of this work in a clearer way at the end of this section. Furthermore, some specific suggestions are:
Response: Thank you for your reminding, the aim is added at the end of this section: ’By comparison, three variable selection methods are discussed to choose the best one’.
Line 41: This sentence is grammatically incorrect, does not? I suggest: “This procedure damages the samples and it is time-consuming”.
Response: Thank you for your valuable advice, I will revise it according to your opinion. The revised version is ‘this procedure damages the samples and it is time-consuming’.
Lines 65-67: Here and in the previous paragraph, you are assuming that all NIRS is homogeneous. There are thousands of examples where denoising and variable selection is not necessary. These procedures are highly recommendable in lots of cases, but not always.
Response: We agree with reviewer that sometimes denoising and variable selection is not necessary. According to this point, we will especially emphasize the necessity by the moisture content. Because the water features in NIRS is typical and obvious, most spectral variables are redundant for water (free –OH group).
Lines 74 and 88: It would be variable and not variables.
Response: Thank you for your reminding. The text has been revised.
Material and Methods
Material and methods section is almost an extension of the introduction. On most occasions, you simply describe in theory the methods used. However, you do not describe how you implement these methods in your study. Following this material and methods section is completely impossible to replicate the study.
Section 2.1. It is necessary to deeply describe the spectral acquisition. How were the samples and the spectrograph placed? How many samples were recorded? What were the differences between the samples? How many times were each sample measured? Etc.
Response: We add some details mentioned above: ‘Samples are taken every hour from drying tank, and then divide the samples equally. At the normal time of withering, there was a total of 15 hours. The fresh tea leaves are taken 5 samples, and in the other 14 hours, the tea leaves are taken 10 samples. In total, 145 samples are obtained in this test.’
Line 113: When were the spectra acquired? Immediately after they were placed at the drying tank?
Response: In each hour, we will take some samples from the drying tank, and then acquired the spectra immediately. And then, we will get the corresponding fresh weight. After drying the samples, we will get the dry weight to calculate the moisture content.
Line 116: Is the resolution homogeneous in all the spectral range?
Response: The number of sample scans is set 64, which can rotate a circle in intergrating sphere. Therefore, the spectral range is mainly homogeneous and average.
Lines 116-117: What the spectral acquisition number is? You mean that you acquire 64 spectra per sample by rotating 25 g of sample? Can you explain it better?
Response: We have reconfirmed the concept what the spectral acquisition number is. It means the number of sample scans, we have amended this concept. The number of sample scan are 64 (can rotate a circle). Each sample is covered with 25 g in intergrating sphere.
Line 119: These groups of tea leaves have not been defined yet. The manuscript needs of a clearer description of the sample and the subsamples created. How many samples have you measured? How many spectra have you used for doing the wavelength selections?
Response: The corrected version has been highlighted in the original.’ Tea leaves samples was tested for moisture content immediately after the spectral experiments, and the samples were measured for moisture content according to GB/T 8304-2013 in Chinese. Each spectrum corresponds to a moisture content, so the number of moisture content is 145. The average moisture content per hour over a range of 15 hours is shown in Figure 1. As the drying time increases, the moisture content gradually decreases.’
Line 122: Were the samples also summited to spectral characterization after the drying process?
Response: No, the dry samples are used to get the dry weight. Using the fresh weight and dry weight to calculate the moisture content.
Section 2.4: In material and method section you can indicate some theoretical information about the different methods. However, the most important is to describe how you applied these methods in order to these methods can be reproducible by the readers. Please add this information. (Software, parameters, etc.). This also occurs in other subsections of the material and methods section.
Response: Thank you for your reminding. The text has been revised according to your valuable suggestion.
Line 144: Delete “of”. “BOSS is a method using collinearity…” or “BOSS is a method that used…”
Response: Thank you for your reminding. The text has been revised.
Line 147: “…random combination of variables…”
Response: Thank you for your reminding. The text has been revised.
Lines 151-153: Check commas.
Response: Thank you for your reminding. The text has been revised according to your valuable suggestion.
Lines 172-173: Check.
Response: Thank you for your reminding. The text has been revised according to your valuable suggestion.
Sections 2.5.3. and 2.5.4: Describe how you applied GA and iPLS please.
Response: Thank you for your reminding. This part is added.
Model Evaluation Indexes
This section is not necessary a priori, however, due to the great complexity of the chemometric methods applied, it may be useful to have these fundamental concepts as clear as possible.
Response: This part is moved to section2.4 with deleting the Table.1. And the title has been revised to’ Analysis of PLS Model ’.
Results and Analysis
Line 203: Although you are not strictly following the sections recommended by the AS guide of authors, I think it is better to use discussion in order to follow this guide.
Response: Thank you for your valuable advice, I have revised it according to your opinion. The revised version is ‘Discussion and Result’.
Line 205: It is the first time you use this term in the whole article. Please define it here or where it corresponds.
Response: Thank you for your valuable advice, I have revised it according to your opinion.
Lines 208-211: It is more recommendable to use the third person in scientific papers.
Response: Thank you for your valuable advice, I have revised it according to your opinion.
Please, check the format of this section. There are several format errors.
Response: Thank you for your valuable advice, I have checked the format of this section and revised.
Table 2: Keep only the first sentence as the title of the table. The remaining text is a table note. Please place it at the bottom of the table.
Response: Thank you for your valuable advice, I have revised it according to your opinion.
I have no suggestions about the scientific aspect of the Results and Discussion section. I found the results interesting. The combination of DWT and BOSS seems to be a good alternative for the selection of variables in spectral matrices. Please check the format of the whole manuscript and correct the formal aspects in the Material and Methods section.
Finally, I think authors have to add some sections, such as Author Contributions, Funding, Acknowledgments or Conflicts of Interest. Please, check the AS guide of authors.
Response: Thank you very much for your valuable suggestion, this part have been added before the References.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear authors,
Thank you for the interesting manuscript.
The topic is important ans the methods are
among the best ones, no doubt.
Here some comments on the manuscript:
Title:
Before reading the rest:
Contains acronyms and too much details.
After reader the manuscript:
Yes, the title could be more general.
Abstract:
Quite goood: tells what has been done.
Could also tell something about the results by numbers.
Only a couple of language points:
Line 25: 'validate the effectiveness' (grammar)
Line 28: 'for the prediction' (also grammar)
Keywords:
Like title: avoid acronyms,
try to write a better title and corresponding keywords.
Introduction:
Line 33 Sentence 1: 'Tea tree is a kind of camellia'
please use a more scientific expression like:
'Processed leaves and leaf buds of tea tree (<i>Camellia sinensis</i> (L.) Kuntze) are used to produce tea using varying levels of
oxidation.'
Line 70: 'parts' better 'types'
Line 74 (et others): 'wavelength selection' (not 'wavelengths selection' so singular not plural: consistently)
Line 104: '...built to validate based on tea leaves using...'
What the authors are trying to say by this?
Could you reformulate this: some missing words or grammar problem???
2. Material and Methods
Line 112: 'about 4cm thick'
Which is 4cm thick? Obviously the tea leaf bed, not the tank, yes?
Lines 116-117: the authors could give a drawing of the sampling arrangement
Line 119: singular/plural (only once): 'tea leave samples'
Lines 120-121: 'in Chinese' was the reference in Chinese?
Line 121: 'teal' is color (or bird of teal color)
Figure 1: Use scientific notation on axis:
x-axis: marks 0, 1, 2, 3, ... [h] (BTW, only 14 hours!?)
Put Origin at point (0,0)
y-axis: what is the unit? Perhaps you should alse define
what you actually mean by 'moisture content'.
Figure 2: Avoid unnecessary background colors
(they actually make copying more difficult while not
giving any extra information).
Line 172: better expression: 'delta_y1 means the standard deviation of y1'
and similarly for delta_y2
Lines 175-176: GA is a global optimisation method
(it finds the best among very many alternatives)
4. Results and Analysis
Figure 4: using (thin) colors in graphs is not wery good.
Show by labels all graphs (now only L7 and L8)
and turn the arrows from label to curve.
You might use different line types to emphasize original spectra etc.
Figure 5: Using color in bars is not good for copying.
Use just black lines.
Figure 6: Color coding bad for copying/printing.
How about using shapes or size?
The yxis unit? Please clarify your graphics.
Table 2:
Difficult to read. It took time to decode what is actually given
with what.
Use horizontal lines or something and think number of decimals,
font size etc to make one liner rows.
Line 293: 'variable selection'
Figure 7:
Caption: In English it is always good to give the most important
words first in a sentence:
'Note: The selected wavelengths reach above the dotted threshold line.'
('Above' is not the most important word, but I think the 'selected wavelengths' are, right?)
Figure 8: difficult to print/copy.
Why use color filling in bars when it is enough to plot just the ends,
which gives a dotted curve...
Think your graphics also here.
Caption also has the above style problem:
'Note: The preferred interval is given in green.'
BTW, also here color is not good for copying/printing...
5. Conclusions
Line 332: 'this research are summarised' (typo: 're' -> 'are')
Line 333: 'the noise was considarably removed':
The authors could give some quntitative measures like SNR here (and earlier).
References:
Be consistent with your format:
AT least the following references are not consistent with most of the
references:
2,14,18,28,32,38 (Upper case title initial letters)
5,10,11,19,31,33 (Repeated journal name)
16 (All uppercase journal name, Use e.g. small caps font)
24 (All uppercase title)
30,34 (Scientific name by italics font)
41 (NIR in uppercase)
Author Response
Title:
Before reading the rest: Contains acronyms and too much details.
After reader the manuscript: Yes, the title could be more general.
Response: Thank you for your valuable suggestions, I will revise it according to your opinion. The revised title is ‘An Effective Moisture Content Prediction Approach Based on Discrete Wavelet Transforms and Bootstrap Soft Shrinkage Algorithm of tea leaves’.
Abstract:
Quite goood: tells what has been done.
Could also tell something about the results by numbers.
Only a couple of language points:
Line 25: 'validate the effectiveness' (grammar)
Line 28: 'for the prediction' (also grammar)
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Keywords:
Like title: avoid acronyms,
try to write a better title and corresponding keywords.
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Introduction:
Line 33 Sentence 1: 'Tea tree is a kind of camellia'
please use a more scientific expression like:
'Processed leaves and leaf buds of tea tree (<i>Camellia sinensis</i> (L.) Kuntze) are used to produce tea using varying levels of oxidation.'
Response: Thank you for your valuable suggestions, I have revised it according to your opinion. The sentence changed to ‘Processed leaves and leaf buds of tea tree are used to produce tea, which are popular in many parts of the world’.
Line 70: 'parts' better 'types'
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Line 74 (et others): 'wavelength selection' (not 'wavelengths selection' so singular not plural: consistently)
Response: Thank you for your valuable suggestions, I have revised it according to your opinion. All ‘variables selection’ is changed to variable selection.
Line 104: '...built to validate based on tea leaves using...'
What the authors are trying to say by this?
Could you reformulate this: some missing words or grammar problem???
Response: The sentence has been revised: ‘After previous variable selection, DWT-BOSS-PLS, GA-PLS, and iPLS models were established, which are based on the NIRS data and moisture content.’ And the propose is added in finally: ‘By comparison, three variable selection methods are discussed to choose the best one.’
- Material and Methods
Line 112: 'about 4cm thick'
Which is 4cm thick? Obviously the tea leaf bed, not the tank, yes?
Response: Thank you for your gentle remind. I have checked this word, the scientific wording is ‘withering trough’. I have revised the word over the manuscript. 4cm means the thickness of leaves which placing on the withering through, I have revised it.
Lines 116-117: the authors could give a drawing of the sampling arrangement
Response: Thank you for your gentle remind. In this part, I have added some detail to explain the arrangement, including How were the samples and the spectrograph placed? How many samples were recorded? How many times were each sample measured? Etc.
Line 119: singular/plural (only once): 'tea leave samples'
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Lines 120-121: 'in Chinese' was the reference in Chinese?
Response: The ‘GB/T 8304-2013’ in Chinses. (government standard)
Line 121: 'teal' is color (or bird of teal color)
Response: Thank you for your valuable advice. This part has been a major revision.
Figure 1: Use scientific notation on axis:
x-axis: marks 0, 1, 2, 3, ... [h] (BTW, only 14 hours!?)
Put Origin at point (0,0)
y-axis: what is the unit? Perhaps you should alse define
what you actually mean by 'moisture content'.
Response: After14 hours, the moisture content is decrease to about 0.46, the tea leaves are needed to push the next step (tea rolling). So we think it is no significant to extend the withering time.
Thank you for your reminding. The concept of moisture content is added in the manuscript. And Figure.1 has been improved.
Figure 2: Avoid unnecessary background colors
(they actually make copying more difficult while not
giving any extra information).
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Line 172: better expression: 'delta_y1 means the standard deviation of y1'
and similarly for delta_y2
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Lines 175-176: GA is a global optimisation method
(it finds the best among very many alternatives)
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
- Results and Analysis
Figure 4: using (thin) colors in graphs is not wery good.
Show by labels all graphs (now only L7 and L8)
and turn the arrows from label to curve.
You might use different line types to emphasize original spectra etc.
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Figure 5: Using color in bars is not good for copying.
Use just black lines.
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Figure 6: Color coding bad for copying/printing.
How about using shapes or size?
The yxis unit? Please clarify your graphics.
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Table 2:
Difficult to read. It took time to decode what is actually given
with what.
Use horizontal lines or something and think number of decimals,
font size etc to make one liner rows.
Response: Thank you for your valuable suggestions, I have revised it according to your opinion.
Line 293: 'variable selection'
Response: Thank you for your valuable suggestions, I have revised it in the whole manuscript.
Figure 7:
Caption: In English it is always good to give the most important
words first in a sentence:
'Note: The selected wavelengths reach above the dotted threshold line.'
('Above' is not the most important word, but I think the 'selected wavelengths' are, right?)
Response: Thank you for your valuable suggestions, this sentence has been changed to ’ Note: The selected wavelength is above the dotted line.’
Figure 8: difficult to print/copy.
Why use color filling in bars when it is enough to plot just the ends,
which gives a dotted curve...
Think your graphics also here.
Caption also has the above style problem:
'Note: The preferred interval is given in green.'
BTW, also here color is not good for copying/printing...
Response: The figure and note have been revised. ’Note: The preferred interval is given in black.’
- Conclusions
Line 332: 'this research are summarised' (typo: 're' -> 'are')
Response: Thank you for your careful check, I have revised it.
Line 333: 'the noise was considarably removed':
The authors could give some quntitative measures like SNR here (and earlier).
Response: This part has used Figure.4 and 5 to express ‘DWT is a good way to remove of the noise in NIR’, is it necessary to add additional instructions?
References:
Be consistent with your format:
AT least the following references are not consistent with most of the
references:
2,14,18,28,32,38 (Upper case title initial letters)
5,10,11,19,31,33 (Repeated journal name)
16 (All uppercase journal name, Use e.g. small caps font)
24 (All uppercase title)
30,34 (Scientific name by italics font)
41 (NIR in uppercase)
Response: Thank you very much for your careful review and the full text has been revised according to your suggestion.
Author Response File: Author Response.docx
Reviewer 3 Report
Manuscript Number: Applsci-857533
Decision Letter of Article: An Effective Moisture Content Prediction Approach Based on DWT and BOSS Algorithm for Yinghong NO.9 Tea Leaves
This paper showed a study for a development of an effective and non-destructive prediction method based on near-infrared spectroscopy based on the method that combines discrete wavelet transforms (DWT) with the bootstrap soft shrinkage algorithm (BOSS). The study is interesting, but the paper presents many errors and fails which that must be corrected. The authors must also explain some questions that were raised:
1-The structure of the paper must be reformulated according to the Instructions to Authors and the English in the text must be revised;
2-The tea leas were previously macerated? The authors must be indicated which is the experimental procedure/processing of vegetal material. The material treatment is important for a suitable and consistent spectra acquisition;
3-The authors used any preprocessing methodology for spectra data normalization, such as the Standard Normal Variate (SNV), multiplicative scatter correction (MSC), etc? What is the opinion about those procedure? The first and second derivative can be used in this preprocessing procedure?
4-The authors must compare this procedure with other processing data. They must present the advantages and disadvantages about that.
5-The table 1 must be removed. The information must be placed along the text.
6-In the section 2.2 the moisture concentration was compared using another experimental procedure? The figure 1 must be improved. Please use another notification about the time gap. Please the moisture content must have unities. Please correct it.
7-In the section 2.4.5 iPLS – Which is the criteria used for the spectra split? The authors did used another procedure?
8-All pictures/figures must be improved.
9-The references must be according to author instructions.
The work must be reformulated and improved due to the fails and doubts. For those reasons, I suggest a major revision. The questions that were raised must be carefully answered.
Comments for author File: 
Comments.pdf
Author Response
- The structure of the paper must be reformulated according to the Instructions to Authors and the English in the text must be revised;
Response: Thank you for your valuable suggestions, I will revise it according to your opinion. And the English has been revised by MDPI English editor.
- The tea leas were previously macerated? The authors must be indicated which is the experimental procedure/processing of vegetal material. The material treatment is important for a suitable and consistent spectra acquisition;
Response: The fresh tea leaves are non-destructive during the experiment. The part of ‘Material and Methods’ is added some details, including How were the samples and the spectrograph placed? How many samples were recorded? How many times were each sample measured? Etc.
- The authors used any preprocessing methodology for spectra data normalization, such as the Standard Normal Variate (SNV), multiplicative scatter correction (MSC), etc? What is the opinion about those procedure? The first and second derivative can be used in this preprocessing procedure?
Response: I have used those methods to establish the PLS model, however, the new method is better than those traditional preprocessing methods. The result is as follow:
|
Variable set |
RMSEC |
RMSECV |
RMSEP |
n_VAR |
|||
|
SNV |
0.8266 |
0.4171 |
0.8175 |
0.4279 |
0.8396 |
0.4127 |
3112 |
|
MSC |
0.8261 |
0.4177 |
0.8170 |
0.4286 |
0.8392 |
0.4134 |
3112 |
|
Der1st |
0.8693 |
0.3622 |
0.8573 |
0.3785 |
0.9329 |
0.2660 |
3112 |
|
Der2nd |
0.8597 |
0.3953 |
0.8444 |
0.3013 |
0.9076 |
0.3013 |
3112 |
Table.1 The result of different preprocessing methods.
- The authors must compare this procedure with other processing data. They must present the advantages and disadvantages about that.
Response: Thank you for your suggestion, those preprocessing methods are applied to NIRS study. However, this paper emphasizes the selection of variables. Before writing this paper, I tried the classical pretreatment methods (SNV/MSC/1st/2nd) with unsatisfactory results, so I derived the idea of variable selection to improve the accuracy of the model. Other classical variable selection methods (GA/iPLS) have been introduced in this paper, so do we need to introduce other non-variable selection methods for comparison?
- The table 1 must be removed. The information must be placed along the text.
Response: Thank you for your reminding. The text has been revised according to your valuable suggestion.
- In the section 2.2 the moisture concentration was compared using another experimental procedure? The figure 1 must be improved. Please use another notification about the time gap. Please the moisture content must have unities. Please correct it.
Response: We have added some details about this experiment: ‘Tea leaves samples were tested for moisture content immediately after the spectral experiments, and measured for moisture content according to GB/T 8304-2013 in Chinese. Each spectrum corresponds to a moisture content, so the number of moisture content is 145. The average moisture content per hour over a range of 15 hours is shown in Figure 1. As the drying time increases, the moisture content gradually decreases. Additional details have been highlighted. The concept of moisture content is added in the manuscript. And Figure.1 has been improved.
- In the section 2.4.5 iPLS – Which is the criteria used for the spectra split? The authors did used another procedure?
Response: We have changed the interval size to find the best result in iPLS. The interval sizes are set 10, 20, 30, 40, 50, 60 to find the lowest RMSECV. The lowest RMSECV of interval size is 20. The result is as follow:（figure）
- All pictures/figures must be improved.
Response: Thank you for your valuable suggestions, we have revised it according to your opinion.
- The references must be according to author instructions.
Response: Thank you for your valuable suggestions, we have revised it according to your opinion.
The work must be reformulated and improved due to the fails and doubts. For those reasons, I suggest a major revision. The questions that were raised must be carefully answered.
Thank you very much for your careful review and the full text has been revised according to your suggestion.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
I thank the authors for their efforts to improve the quality of the manuscript. I consider the manuscript in its present form to be acceptable for publication.
Reviewer 3 Report
The suggestions were performed according to the reviewer's revision. I suggest accepting this work.
