Designing Unmanned Aerial Survey Monitoring Program to Assess Floating Litter Contamination

Round 1

Reviewer 1 Report

Dear authors

I have read with great interest your paper dealing on the feasibility to monitor floating litter using UAV images and on the operational advantages and disadvantages of three imagery processing and analysis strategies: i) Manual Count; ii) Pixel Based Detection and; iii) Machine Learning. The manuscript is well written and organized but present some issues that should be addressed and in particular the discussion 

Major-Moderate issues.

Results

Could you indicate which items were correctly identified and which were not for each method?

Were all the items identified with visual observation (manual count)?

This information are important and should be provided

Discussion

Most of your discussion is not directly based on your results and should be therefore rewritten.

Did you observe difference in the detection of the ML items in relation to the flight height? Your analysis does not consider flight height or weather parameters. Indeed, it is not clear on what results did you based section 4.1 (discussion not based on your results 364-418 for instance).

Maybe you should also explain in the method section why you used two image sets.

l.428-439 already mentioned before.

As a consequence, the conclusion will also need to be partly rewritten after modifying the discussion

Minor comments
the introduction can be reduced focusing on the main scope of the paper

Method section: did you perform different tests with different flight height (10, 20 and 30m)? Please specify the image resolution (cm/px)

Could you please add a description of the items used in the method section. How many items did you used? Did you use only plastic items?

Figure 2 is not clearly explained

Some elements of the result's section should be moved to the discussion (L.328-336: limitation of the ML approach)

The two first paragraphs of the discussion (l.337-363) is very general and is not directly related/linked to the discussion of your results, maybe more useful for the introduction. The introduction, however, should be reduce focusing on the aim of your study.

4.1: Operational and Analytical Considerations     you may remove the section or you should use different subsection.

Best regards

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors address the issue of the floating litter monitoring by means of aerial surveys using unmanned aircrafts. They thoroughly analyze the problem and consider various methods of image analysis. It turned out that the algorithm based on the machine learning exposed poorer efficiency than visual inspection. This results indicate the necessity of further development of machine learning algorithms for automated object detection. Despite the manuscript  rather asks questions than answers them, I recommend publication after minor editing of English, mainly in terms of the vocabulary usage.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

MS: Designing unmanned aerial survey monitoring program to assess floating litter contamination

The paper compares a couple of remote sensing methods to identify floating objects from a sea water using high resolution UAS imagery. It is very valuable to discriminate litter objects from other background features using high resolution UAV data. As I aware it is bit hard to mosaic UAV data without having good ground control points. Therefore it is useful to provide some details on processing UAV data and limitation of the study. Here are a few comments to improve the quality of the article and some comments are made using sticky notes in the pdf.

Abstract

L 7: Collected UAS images

1.       Introduction

Generally, introduction is well written providing good background knowledge on micro plastic debris in the ocean, available remote sensing data such as Landsat images and novel UAV application on coastal science research. The objectives have given but not included any research questions going to be address in this research.

L 114: Include a reference for habitat vulnerability mapping in the coastal regions using sUAV.

Here is an example:

Madurapperuma, B., Lamping, J., McDermott, M., Murphy, B., McFarland, J., Deyoung, K., ... & Mitchell, S. (2020). Factors Influencing Movement of the Manila Dunes and Its Impact on Establishing Non-Native Species. Remote Sensing, 12(10), 1536.

2.       Materials and Methods

L 169: I believe the numbers given for the rows and columns of a scene. The resolution may fall between 3-10 cm I believe

L 169: space between number and unit 10 – 30 meters

L 172: What percentage set for side and forward overlapping of collecting UAV imagery.

L 184: valid UAS images

L 216 &L 219: color? Or colour? Use whichever the style throughout.

UAS data preprocessing information’s are lacking in this manuscript. The supplementary documents have some information. Are there any issues encountered during mosaic images (i.e. create empty spaces) since it create problems when identify features from open sea during mosaicking using structure from motion techniques.

L 194: Capitalize each word

L 206: Delete period

3.       Results

L 276: Spelling mistake (T)he

L 292: space between parenthesis and word

L 296: Figure 3: You can label these two graphs as a) and (b) and then give a detail in Fig 2 caption.

Add some results of feature extraction of debris in different methods

How do you validate your classification methods

4.       Discussion

The discussion is well written with supporting literature.

L 438: Make lowercase which. Delete period and add comma before which

5.       Conclusion

Conclusion may improve summarizing the main findings.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors addressed almost all comments suggested and thank you. The quality of the article has improved with the inputs for reviewer's comments.