A Comparative Approach of Fuzzy Object Based Image Analysis and Machine Learning Techniques Which Are Applied to Crop Residue Cover Mapping by Using Sentinel-2 Satellite and UAV Imagery

Round 1

Reviewer 1 Report

The authors have submitted a work about crop residue cover mapping through the application of FOBIA to satellite and UAV imagery. The submitted work has a very poor quality that leads me to recommend the rejection of the paper. I will explain the reasons for the rejection in the following paragraphs.

The English language and style used in the manuscript are very poor, and this is evident even for me, that I am not a native English speaker. The manuscript has mistakes even in the title, where the authors use the expression “UVA images”. Such a mistake in the title, together with the inconsistences in the text formatting, make me think that the authors were not careful enough in the manuscript preparation. This idea is confirmed by the fact that the text is poorly structured; there are discussion fragments in the results’ section, the discussion includes an abstract of the work, and the conclusion repeat ideas from the introduction and the abstract.

The submitted work seems to be a mixture of previous works from the authors (references 35 and 42 of the paper) with the only novelty of including UAV imagery. That previous works already included OBIA segmentation and fuzzy classification of satellite imagery of the same study area with the aim of residue cover mapping.

The authors did not do a thorough bibliographical research. For example, they claim that “A limited number of studies are concerned with the study of OBIA method for agricultural issues” (L121). However, there are plenty of OBIA works in agricultural environments, such as:

• de Castro, A. I., Jiménez-Brenes, F. M., Torres-Sánchez, J., Peña, J. M., Borra-Serrano, I., & López-Granados, F. (2018). 3-D Characterization of Vineyards Using a Novel UAV Imagery-Based OBIA Procedure for Precision Viticulture Applications. Remote Sensing, 10(4), 584.
• Gao, J., Liao, W., Nuyttens, D., Lootens, P., Vangeyte, J., Pižurica, A., et al. (2018). Fusion of pixel and object-based features for weed mapping using unmanned aerial vehicle imagery. International Journal of Applied Earth Observation and Geoinformation, 67, 43–53.
• Karydas, C., Gewehr, S., Iatrou, M., Iatrou, G., & Mourelatos, S. (2017). Olive Plantation Mapping on a Sub-Tree Scale with Object-Based Image Analysis of Multispectral UAV Data; Operational Potential in Tree Stress Monitoring. Journal of Imaging, 3(4), 57.
• Lam, O. H. Y., Dogotari, M., Prüm, M., Vithlani, H. N., Roers, C., Melville, B., et al. (2020). An open source workflow for weed mapping in native grassland using unmanned aerial vehicle: using Rumex obtusifolius as a case study. European Journal of Remote Sensing, 1–18.
• Li, M., Ma, L., Blaschke, T., Cheng, L., & Tiede, D. (2016). A systematic comparison of different object-based classification techniques using high spatial resolution imagery in agricultural environments. International Journal of Applied Earth Observation and Geoinformation, 49, 87–98.

The description of the materials and methods is extremely poor. It lacks information about:

• UAV flights: number of flights, hour of the day, covered surface, radiometric calibration of the images, mosaicking process, overlap among images.
• Satellite imagery: radiometric calibration, details about the sensor bands (wavelengths, resolution).
• Line-transect data: amount and location of the transects.
• Image segmentation: were all the bands included in the configuration of the MRSA?
• Membership function: the authors do not explain the parameters of the Gaussian curve (mean, standard deviation).
• Artificial neural network: Was the NN applied to pixels? Or to object or windows? How many neurons had the hidden layer?
• SVM: Was the SVM applied to pixels? Or to object or windows?
• Accuracy assessment: was the accuracy assessment based on objects or in pixels? Was the segmentation accuracy evaluated?

And, last but not least, I have doubts about the quality of the analysis carried out by the authors:

• The segmentation shown in Figure 3a does not fit to the limits of the image areas. It seems to be a random set of rectangular divisions of the image. Then, in figure 4, the object limits do fit to the different areas in the image. However, in figure 8a the final classification seems to adhere again to the limits of the segmentation shown in figure 3a.
• Segments of UAV imagery in figure 3b have different sizes across the orthomosaic and the authors did not comment anything about this fact.
• The authors did not fully justify the election of the fuzzy membership function, nor the reason for selecting five features for the classification of some classes in the UAV imagery and using a different amount in the satellite imagery.

Author Response

Dear reviewer

First of all we very much appreciate of your support regarding our manuscript. We have revised our manuscript according to your comments very carefully. We did our best to improve the scientific quality of the manuscript significantly. We provided answer letter for your beneficial comments, which is attached to the revised version. We also use track change  as pdf file which enables you to identify our revise. Based on the constructive comments proposed by your kind review and our according revise, we believe that the scientific quality of the paper has improved significantly. We are very confident that you will find this revised version now worthwhile to get published. 

Author Response File: Author Response.pdf

Reviewer 2 Report

My main concern with the text are in yellow and explayed  through comments, in the annex text.

I add also a text describing the reasons of the evaluation.

Comments for author File: Comments.pdf

Author Response

Dear reviewers,

First of all we very much appreciate of your support regarding our manuscript. We have revised our manuscript according to your comments very carefully. We did our best to improve the scientific quality of the manuscript significantly. We provided answer letter for your beneficial comments, which is attached to the revised version. We also use track change  as pdf file which enables you to identify our revise. Based on the constructive comments proposed by your kind review and our according revise, we believe that the scientific quality of the paper has improved significantly. We are very confident that you will find this revised version now worthwhile to get published. 

Best regards,

Authors

Reviewer: 2    

My main concern with the text are in yellow and explained  through comments, in the annex text.

I add also a text describing the reasons of the evaluation.

Comments on PDF file

Thank you very much for your comments on the pdf file, we checked them carefully and did our best our address your valuable comments. Finally, we appreciate your support and your constructive comments. We strongly believe that we have addressed the issues concerning the logical flow of the article and have improved the overall structure significantly. We hope that you will find this revised version worthwhile to get published. 

Thank you very much again.

 The authors

Author Response File: Author Response.pdf

Reviewer 3 Report

While there is no doubt that the scientific significance of this study is of merit, there are several limitations found with this paper that cannot be overlooked and warrant its revision. 

1. There is a need for extensive English editing to properly convey the principles of this study and to link them to the significance of the results. In many cases there is an inconsistency  in the tense or missing word which breaks the reading and cannot be overlooked. In other locations this lack of editing resulted in sentences that were far too long (> 3 lines).
2. At several locations throughout the paper (e.g., line 73) acronyms were redefined. Acronyms should be defined once and used consistently throughout the paper following the MDPI guidelines.
3. Another major limitation of this paper was the severe lack of citations. There were no citations for the definition of ANN or confusion/ error matrix. Additionally, the entire discussion section (consisting of multiple pages) was written without a single citation meaning there could not have been a reflection of this work in regards to the wider body of scientific knowledge. Each section should be revisited for potential missing citations.
4. Figure 1 is rather complicated; it may be better fitting with fewer intermediate cutouts.
5. The formatting of tables 1 and 2 should be revisited to separate the text and have a more consistent appearance.
6. Figures 4, 5, and 6, based on their explanation in the text do not seem to add enough to the paper to justify their inclusion at their given locations (again breaking the reading). This may be a product of understanding of the written text. It may also be viable to include these as supplemental figures instead.
7. Lastly, another major limitation that cannot be overlooked here is the lack details for the methodology. The flight and processing parameters for the UAV were not given enough details as to understand if they were properly conducted. This could have considerable implications for the soundness of the study. In addition to this the discussion of the accuracy assessment methods and results needs to be revisited. Without an understanding of the number of reference data samples and how they were all collected it cannot be determined if this assessment is statistically valid. As it stands there is only a single 100-foot-long (must be given in metric) transect for ground reference data, which would not be suitable for an accuracy assessment of image objects. Along with this the reference data is referred to as “truth” values which is incorrect in any context, especially when considering fuzzy logic.

Author Response

Dear reviewer

First of all we very much appreciate of your support regarding our manuscript. We have revised our manuscript according to your comments very carefully. We did our best to improve the scientific quality of the manuscript significantly. We provided answer letter for your beneficial comments, which is attached to the revised version. We also use track change  as pdf file which enables you to identify our revise. Based on the constructive comments proposed by your kind review and our according revise, we believe that the scientific quality of the paper has improved significantly. We are very confident that you will find this revised version now worthwhile to get published. 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper has been considerably improved after the first round of review. Now, it is worthy of publication. 

Author Response

Many thanks for the positive feedback regarding the revised version! We are happy to you agreed with our revised and recommended the paper to be accepted!

With best regards,

Authors

Reviewer 2 Report

Please see my comments into the text. I have one comment about DN (Dgital number and reflectance).

A comparative approach of fuzzy object based analysis and machine learning techniques applied to crop residue cover mapping using multispectral Sentinel-2 and digital UVA images. (V2)

The novelty of this study is investigation of the potential of semi-automated fuzzy OBIA in detection of CRC and classification of tillage intensity.

The main objectives of this study are: a) to propose a novel methodology for CRC monitoring using a semi-automated fuzzy OBIA and as well as compering its efficiency against the two well-known Machine learning techniques (SVM and ANN), and b) to investigate the potential of UAV based digital images and multispectral satellite images for detecting CRC.

The article format improve a lot.

The methodology still presents some gaps as poor descriptions of the UAV images in terms of use or not the physical measures (Reflectance or DN). At least how the authors did the correction of radiation during acquisition (11:00 am until 3:00 pm). There is differences in intensity and angle.

The Discussion improved as well as the conclusion.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

Many thanks for the positive feedback regarding the revised version! We have revised the paper once again according to your comments and we hope that  you will find the new version to be accepted!

With best regards,

Authors

Author Response File: Author Response.docx

Reviewer 3 Report

You have no doubt taken great strides to improve the comprehension of this paper. There are still, however, several shortcomings for this paper which must be addressed before it can be recommended for publication. 

The corrections you have made to the English grammar and sentence structure have made the paper more accessible but there are still sections that must be revisited (e.g., lines 87, 106-107, 152-153, 422-424, and 432-436). For this last example specifically, this seems like a stepwise function and could be explained much more simply. 

Each of the tables and figures are much more clearly presented now, thank you for those corrections. 

Following my initial review and the responses of other reviewers it seems that the novelty of this paper is just starting to come to light. In your introduction (lines 119-122) you have changed the text to address the merit of this study within the larger body of work however this paragraph still does not reflect the new insights that this study brings. This section of your introduction should explicitly detail yours (and others) past work on this very subject (e.g., citations 35 and 42). This will no doubt change the perspective of the paper to speak specifically to what methods and results are novel here.it would also change the required amount of detail given on certain subjects in the methods and results. 

On a similar note, as with my initial review, there is still a tendency to not provide proper citations. There are no citations given in the accuracy assessment section of the methodology. At several locations throughout the paper you speak directly about given definitions or the acceptance of methods without providing citations (e.g., lines 62-63, 485-486, 557-558). 

Lastly, this revision made no comment or amendment to use the of "truth" data in the accuracy assessment section and discussion. This section cannot be accepted with such a reflection on the accuracy assessment process. 

Author Response

Dear Reviewer,

Many thanks for the positive feedback regarding the revised version! We have revised the paper once again according to your comments and we hope that  you will find the new version to be accepted!

With best regards,

Authors

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

The authors should to show How the DN agree with reflectance in case of Drone. It is not correct to use DN instead of reflectance, indenpendent the bibliography (60 and 62). 

Author Response

We revised the paper according to comments.

Reviewer 3 Report

The revisions that you have made for this version of the manuscript demonstrate a remarkable improvement in its quality. While there are still more minor revisions that can be noted (referenced below) the majority of the original suggestions have been remedied. As such I can now recommend this paper for publication without complaint. Thank you for your dedication to these revisions.

1. Minor grammar issues were highlighted (yellow) in the attached PDF. This includes a single confusingly long sentence in the discussion section.
2. A few acronyms were not defined in the body of the text. These include: [Landsat] TM, [Landsat] OLI, and UAV.
3. Citation #49 and #94 are for the same reference.

Comments for author File: Comments.pdf

Author Response

We revised the paper according to comments. 

Author Response File: Author Response.docx