Hyperspectral Estimation of Nitrogen Content in Wheat Based on Fractional Difference and Continuous Wavelet Transform

Round 1

Reviewer 1 Report

Review of the manuscript: "Hyperspectral Estimation of Nitrogen Content in Wheat Based on Fractional Difference and Continuous Wavelet Transform"

 

The manuscript deals with the estimation of nitrogen content in wheat by means of ground-based hyperspectral data collected through a proximal sensor. The authors compared different pre-processing methods (computation of vegetation indices, fractional order differentiation, continuous wavelet transform) and five estimation models (SVM, ridge regression, stepwise regression, GPR, and BP neural network). The methods were evaluated at different wheat growth stages.

The topic is of interest for the scientific community and falls within the aims and scope of Agriculture-MDPI journal.

There are few points that need to be clarified; specific comments are reported below. In addition, the limits of the present work have to be better highlighted. The results are interesting but are obtained by using a reduced sample size dataset and considering only one experimental site for two years; for this reason, they cannot be generalized as the Authors did in several parts of the manuscript such as in the Abstract and the Conclusions sections.

 

Specific comments are reported below.

Line 26: “verification” or “validation”?

 

Lines 25-26 and 731-735: “Therefore, the flag-picking stage is the optimal growth period for estimating the nitrogen content of wheat”. This is the result of a specific study, it cannot be generalised since it was obtained using a reduced sample size dataset and considering only one experimental site for two years. The Authors are required to modify both these sentences and all the sentences that reported similar comments.

 

Line 35. A conclusion on the limits of the present work and future perspectives is missing and has to be included in the Abstract.

 

Lines 186: use the plural “vegetation indices” instead of “vegetation index”.

 

Line 292: check the equation for the computation of the coefficient of determination (R2). R2 is obtained as the ratio of the regression SS on the total SS.

 

Lines 287-296: The Authors have to provide details on the partitioning of the dataset in the “training” and “validation” set since this is a very important step that greatly affects the results (details the number of observations in each group; the procedure used).

 

Line 342 and …. Specify in the caption of the Tables 2, 3, 4 the number of observations used in the training set and those used in the test set.

 

Lines 403 and 405: use always the same terminology. In the caption of figure 5 and 6, the Authors used the term “fertility stages” instead of “growth stages”. The same is for other parts of the manuscript text. Check the text and correct where needed.

 

Close each sentence with a point “.” instead of a semicolon (;). Some examples are reported below:

“Therefore, the flag-picking stage is the optimal growth period for estimating the nitrogen content of wheat; For different growth stages, the optimal estimation models of nitrogen content were different. ”

“…the average reached 0.76 and 0.71, and the overall estimation effect was good; The average values of the modeling and verification R2 of the nitrogen content estimation model …”

The quality of English Language is good. There are minor spelling problems. Check the punctuation over the whole text.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

 

In the manuscript entitled “Hyperspectral estimation of nitrogen content in wheat based on fractional difference and continuous wavelet transform”, the authors used hyperspectral information from wheat canopy to determine nitrogen content during different phenological growth stages. Mathematical analyses are compared including pre-processing methods (fractional differential, continuous wavelet transform and vegetation index), and five nitrogen estimation models (SVM, ridge regression, stepwise regression, GPR, and BP neural network). The proposed methodologies are well-supported, and the mathematical analysis is robust.

Some additional observation:

·         Line 203: it is lacking the variable associated to spectral reflectance.

·         Check formula number 8.

·         Line 564: 1-9,558 nm, is it correct?

·       Results Section: The manuscript is very descriptive, but authors tend to repeat phrases in various paragraphs, comparing data from tables as percentage values. It requires better systematizing the results section.

 

 

Minor editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Well done. There are some comments. Please, Find the attached file.

Best regards,

Comments for author File: Comments.pdf

 Minor editing of English language required. Please use simple sentences. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf