Forecasting Table Beet Root Yield Using Spectral and Textural Features from Hyperspectral UAS Imagery

Round 1

Reviewer 1 Report

Dear Authors,

The manuscript presents a very interesting study on the application of novel methods in processing UAV hyperspectral imaging data for yield forecasting of crop plants, here of red beet. Given the plant species studied, this is a unique study.

The experiment was performed carefully and correctly, with the presentation of their results, discussion and conclusions.

The chapters of the manuscript Introduction, Material and methods, Conclusions were prepared properly.

The results and their discussion were described and presented comprehensively, but in several places the Discussion chapter contains elements of the Results. The Discussion and Results chapter should be rewritten with this in mind.

The manuscript was prepared carefully.

 

Minor comments and suggestions:

L 105: 60? × 100? > 60 ? × 100 ?

L 114: „different stages during the growing season”

If there were different stages of development then please express them in these stages, e.g. BBCH. Justify why you have chosen those dates given in Table 1. If there are images you can easily determine the BBCH stage. Expressing in BBCH stages (or others that are standard in your particular region) will be clearer to readers positioned more in agronomic/biological disciplines than in engineering, IT or RS.

L120: “Headwall’s Hyperspec III SpectralView software”. Specify software versions.

L121 „Pix4D” . Specify software versions.

L294: "Figure 2." Figures and text are very small, unreadable. It is worth considering re-editing. Perhaps a vertical layout of the figures, 2 x 5 , rather than horizontal 5 x 2? Similarly, figure 3. and 4.

L394: Fig. 4 and its description in the paragraph above are probably part of the results, not the discussion. Similarly, Figs. 5. and 6.

Author Response

Dear Reviewer,

We would like to thank you for your sincere and thorough review of the manuscript. Your comments and suggestions were very helpful and we have updated our manuscript to incorporate your suggestions. 

While we’ve moved figure 6 and its corresponding section 4.3 over to the Results section from the Discussion section, we did not do it for the other two figures; we felt they are more appropriate in their current chapter providing context to our discussion.

Sincerely,

Mohammad S Saif

Reviewer 2 Report

The authors propose a feature selection procedure to identify the most informative spectral and textural features in a specific application of precision crop management. 

To evaluate the feasibility of predicting  beet roots weight, they use hyperspectral images acquired in several flights  corresponding to different stages of the growing season and derive normalized indices for reflectance and texture differences in spectral bands pairs.   The manuscript is very well written, the methodology thoroughly described and the results widely discussed.  The main limitations of the proposed procedure, essentially related to the single specific data set that has been analyzed and the need for further validation, are also clearly stated.

Author Response

Dear Reviewer,

We would like to thank you for your sincere and thorough review of the manuscript. We appreciate your kind feedback. While reviewing again, we spotted some minor English errors and have rectified them.

Sincerely,

Mohammad S Saif

Reviewer 3 Report

The manuscript entitled “Forecasting Table Beet Root Yield from Spectral and Textural Features from Hyperspectral UAS Imagery”, mainly focuses on the determination of table beet root yield using hyperspectral UAS data, especially spectral and textural features. The study focused on evaluating the results by exploring different growth stages of the considered crop. However, a presentation of the life cycle as well as these stages will be a considerable point for the reader. The study also presents a selection of specific wavelengths which have high prediction ability. More detailed remarks result after reading the article, in particular:

 1.    It would be preferable to change from in the title to avoid repitition.

2.    Line20: indicate the abbreviation of LOOCV.

3.    The prediction algorithm or method was not presented in the abstract, only Hyperspectral data are mentioned.

4.    Authors have mentioned in their abstract growth stages of table been. It would be better to describe them in details in the manuscript.

5.    In related works, only one study related to beet root was included. It will be more interesting to give the state of the art of using imagery (UAV/multispectral/hyperspectral…) for beet root.

6.    Line 77: replace did extract with extracted.

7.    Line 77: how about their results?

8.    Line 79: which type of crops? Provide them.

9.    Line 88: provide the abbreviation of LAI, this is the first mention.

10.     Line 104: write (cv. Ruby Queen) in italic.

11.     In your manuscript, sometimes the authors use yield sometimes weight. Explain the relation at the beginning and choose the right one in each specific sentence.

12.     In the data collection sub-section, it would be interesting to get a photo in order to show the plots of your study.

13.     Which criteria did you consider to choose the dates mentioned in table 1 to conduct your flights?

14.     Line 281: Explain VIF. It will be better to provide all the abbreviations in their first mention in the article.

15.     The order of the information provided in the manuscript should be revised, many terms are mentioned in the top of the article. However, their explanation is done in the following sections.

16.     Line 333: why did you choose the alpha value of 0.5 for the random stepwise regression?

17.     Line 336: it is better to express the value by kg /unit area.

18.     Line 372: VDVI ?

19.     A part from the results is indicated in the discussion part (Figure 6)

20.     The conclusion is too detailed and repeats what was described before.

 

Author Response

Dear Reviewer,

We would like to thank you for your thorough feedback on our manuscript. Your comments were constructive and we have revised the manuscript accordingly, incorporating your suggestions.

Below we have detailed our responses for each issue that you have raised.

1. The title has been changed.
2. We have added the abbreviations.
3. We have now mentioned using multivariate linear regression in the abstract.
4. We have added a growth stage column in Table 1 for the understanding of our readers.
5. Unfortunately, not many studies are available in the literature that predicts beet root weight. To the best of our knowledge, we have included all the studies related to the prediction of red beet and sugar beet root weight from airborne sensing. In the discussion section (line 557- line 570) we stated the performances of other available methods in the literature related to beets and other root-type crops to provide context for our results. But we did not perform a direct comparison, as the results were based on different data and metrics, e.g., we used loocv with 18 data points.   
6. Corrected,
7. Results added
8. Type of data mentioned
9. Abbreviation added.
10. Rectified
11. We stated our definition of yield (in line 18 and line 109), and used the appropriate term throughout the manuscript.
12. A sample airborne hyperspectral image for one of the flight of all 18 plot has been added (figure 1).
13. We mentioned the circumstances for the data collection dates (line 129 - 132).
14. VIF explanation provided.
15. Reordered.
16. This was a typing mistake; we used 0.05, which is standard across the literature.
17. We have updated our yield units to kg/m
18. Abbreviation added
19. Figure 6 shifted into the result section.
20. We removed similar statements and made the conclusion more succinct.