Analysis of Shoreline Dynamics and Beach Profile Evolution over More than a Decade: Satellite Image Characterization and Machine Learning Modeling

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper entitled "Analysis of Shoreline Dynamics and Beach Profile Evolution Over More Than a Decade: Satellite Imagen Characterisation and Machine Learning Modelling" presents the evolution of the shoreline in the Colombia's Caribbean coasts. Traditional methods, like Dean and Bodge formulations, were compared to 28 machine learning algorithms from literature. One of them provided the best results along six analyzed transects. The study is interesting, even if it does not present any innovation, but just an application in an area which presents erosion trends. It will probable be suited for publication, but there are a few things that should be fixed.

Major comments:

• After figure 4, when you talk about the orthophotos, you should explain better how they were surveyed. I understand later that they are taken from aerial photographs, but the concept is really confusing. I should read how they were acquired, which sensors were used (and which characteristics: RGB? Multispectral?), how they were processed (photogrammetry?), what is the spatial resolution. Moreover, it is not clear how have you extracted the reference shorelines. Were they manually vectorized? 
• I think there is a different notation in figure 5. I read profiles 6,7, 8 and 9 but you cited 1, 1A, 2 and so on. Please correct (also the caption). It is very difficult for me to read and follow the graph in this way. In the same figure, I think there are some words not in english (ene-03, abr-13...). In addition, how can I understand north or south from this picture, as you state when you write "with greater widths towards the south (profile 1)..."?
• I think that a separate discussion paragraph would help the robustness of your work. The manuscript lacks a comparison with other published studies that would help demonstrate the scientific contribution and novelty of your work. You presented all the results, but are they consistent with the existing literature? Which advancement or innovation have you found that is not present in the state of art? What can your work be used for? By answering these questions, I think the paper would enhance a lot.

Minor comments:

• I would write something more specific in the abstract, for example when you write "Changes in shoreline position.... were evaluated" you can add the used method/methodology.
• I was expecting to find in the introduction also a little state of art of the methods and machine learning algorithms related to shoreline extraction. This would help the discussion and highlight the advancement in the scientific field. So, you should expand this section.
• In line 11, page 3, "according to the concept by Reference [13]", I think in this case you should write "according to the concept by Silvester and Hsu [13]".
• At the end of the introduction, a dot is missing.
• Caption Figure 1: despite a typo error in the word "imagens", I would change the caption, as it mainly represents the location of the same area but in different times. An example could be "Location of the case study (a) in 2020, before the works in the Coastal Protection Project (CPP) and (b) in 2025, after the CCP."
• In the first row after figure 1, what do you mean by "Coutier 1.53"?
• Figure 2: the upper left figure should be "DJF" and not "DEF"
• Section 2.2: Is there a reason why you selected that specific periods? A clarification would help to understand.
• Figure 3: there is a typo error in the white box with orange boundary (trainning instead of training) and you should centre the words in the grey box.
• You should explain why in figure 4 the profiles are named 1, 1A, 2.. and not in progressive numbers. For instance, do profiles 1 and 1A share the same characteristics?
• page 7: when you describe the Dean and Bodge formulation, you should cite the relative bibliography.
• in the last rows of page 10, "as indicated in figure 5a-5d" I think this is figure 6.
• figure 6: using white and gray colours for the shorelines makes it not easily readable. You could put colours in the images or change the type of lines (example: continuous line, or dotted line...)
• page 13: "as it can be seen from the graphs", are they figure 7 and 8? Please write.
• Check also captions of figures 7 and 8, where profile 6 and 7 are mentioned.
• The references are adequate, yet they could be further expanded given the algorithms employed. Since these algorithms have already been applied in previous works (though not directly compared), the literature review would be more robust if it incorporated studies discussing you prior use.

Author Response

Dear Reviewer:

Please see the enclosed file.

Regards

The authors

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript analyzes shoreline and beach-profile change at Bocagrande from 2001 to 2016. The study integrates topographic/bathymetric surveys, grain-size data, orthophotos/Google Earth mapping, and equilibrium-profile fitting. The dataset is valuable, but methodological and editorial issues must be addressed before publication.

1. The satellite images displayed in the paper are too blurry, and white or gray lines on grayscale images make it difficult for readers to locate them. Meanwhile, how authors delineate coastlines from satellite images is unknown. Do they just draw them directly from the images? And the tide level should be provided to prove they are acquired during the high-tide days.
2. The introduction is well organized, but the research gap between this paper and previous studies is not identified clearly. Please explain the gap this paper fills for the previous research.
3. The extremely high accuracy of ML brings me a big concern about the possible overfitting. I am concerned about the data dependence. The authors did not mention how they chose the 15% test data. The test data may have high correlation with the validation data, causing a fake high R-square. Add robustness diagnostics, such as MAE, bias, and performance stratified by season/year. Consider a leave-one-profile-out and leave-one-year-out validation for assessing generalization.
4. In Figure 2, “DEF” should be “DJF”. How are authors so careless about this minor error?
5. In the second paragraph of 2.2 Methodology, the authors mentioned “GPSD” and “HYAPACK MAX”. Should they be “DGPS” and “HYPACK MAX”? Double-check it
6. Profiles 6–9 are referenced in Figure 5, but where are they located?
7. Like I mentioned before, Figure 6 is a disaster for readers. The authors should ask a GIS specialist to improve this figure. Also, why not draw more coastlines for comparison? This is the journal “Geomatics,” and the quality of the figures is unacceptable.
8. How can the authors have a typo in the title? Imagen?

Author Response

Dear Reviewer:

Please see the enclosed file.

Regards,

The authors

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Analysis of Shoreline Dynamics and Beach Profile Evolution Over More Than a Decade: Satellite Imagen Characterisation and Machine Learning Modelling.

Corrections

The title of the article needs to be rewritten by removing “imagen” and adding “imagery”.

Introduction:

1. The first statement in the introduction needs an in-text citation(s).
2. There are also coastal change processes that are manmade, such as sand mining activities, etc. This context needs to be included in the introduction.
3. In the second paragraph, the first statement, including other statements made, also needs in-text citation(s).
4. A similar observation is noticeable in the third paragraph of the introduction, especially the first and the second statements made, which need in-text citation(s).
5. A similar observation is noticeable in the fourth paragraph of the introduction, especially the first statements made, which need in-text citation(s).

Materials and Methods

1. The second, third and fourth paragraphs in the case study lack sufficient in-text citations for all the statements made.

Results and Discussions

There is a lack of previous studies, which can be used to accept or reject the current study's outcome in the discussion statements.

Reference:

The DOI link on the list of references is not working; it needs to be corrected for easy access to the materials cited in the paper.

 

Author Response

Dear Reviewer:

Please see the enclosed file.

Regards,

The authors

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have included and modified what was requested. I do not have any other issue, so I would accept the paper.

Author Response

Thank you so much.

Reviewer 2 Report

Comments and Suggestions for Authors

Thanks for providing the revised version. The new manuscript has been improved in terms of writing, but the figures still require modification. The current figures are in low resolution, and you should update almost all figures with higher resolution. For example, how can readers figure out those shorelines? You should provide a zoom-in figure covering only your study area to show the locations of multi-shorelines. In Figure 3, it is GDPS, not GPSD. In Figure 6, you can use different colors for your "comparison" to separate accretion and erosion. Overall, update those figures with higher resolution.

Author Response

Dear Reviewer: 

Please see the enclosed file.

Kind regards,

The authors

Author Response File: Author Response.pdf