Evaluation of Machine Learning Algorithms in the Classification of Multispectral Images from the Sentinel-2A/2B Orbital Sensor for Mapping the Environmental Dynamics of Ria Formosa (Algarve, Portugal)
Round 1
Reviewer 1 Report
Please, see the attached file.
Comments for author File: 
Comments.pdf
Author Response
"Please see the attachment."
Author Response File: 
Author Response.docx
Reviewer 2 Report
Reviewer’s comments
What are the contributions or novelty of your research?
In Line 16, you mention, "The images were submitted to GIS classification using groups of ML algorithms…." It is true that many machine learning algorithms, such as SVM, RF, KNN, and DT, are popularly employed in remote sensing classification. However, it is not clear about GIS classification. Can you give me some examples of GIS classification using machine learning algorithms?
Several papers have discussed the utilization of machine learning algorithms for land use classification using Sentinel-2A/2B data. However, I kindly request the authors to conduct an in-depth review of the existing literature concerning the application of machine learning algorithms for land use classification. This comprehensive review will ensure a well-rounded understanding of the current state-of-the-art techniques and potentially identify novel approaches or gaps in the field that can be explored in the research.
What is QGis V.3.22.16? it is QGIS V.3.22.16?
Why do we use "05" instead of "5" in the context of a quantitative number?
The equations in lines 242, 243, 244, 246, 250, and 253 can be represented in equation form using the equation function in Microsoft Word.
The results of this study demonstrate variations in the classification of land use data arising from differences in the utilization of training samples. However, the authors did not explicitly specify the study's purpose. How will the outcomes of this study significantly contribute to future research concerning machine learning-based land use classification?
A comprehensive discussion is required.
Your conclusion contains excessive content. It should be concise, short, and straight to the point.
I believe that moderate editing of the English language is necessary.
Author Response
"Please see the attachment."
Author Response File: Author Response.docx
Reviewer 3 Report
In this work, 4 widely-used classifiers, i.e. Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (KNN), and Decision Trees (DT) are utilized to address the classification problem of Sentinel 2A/2B images from an area of high environmental dynamics. The corresponding comparisons are taken place evaluating for each case some well-known evaluation metrics, i.e. Recall, the Global Kappa Index, and the General Accuracy Index.
The experimentation is extensive and the text is well-written. However, in my opinion, the technical innovations of the paper is limited. Regarding the classifiers used, there is no innovation, as they are well-known and widely used for many years. What are the general conclusions that someone can draw from this paper?
Also, in my opinion, The section titled "Image classifiers based on ML algorithms" can be written more concisely. In this section, the classifiers used are described, which is common knowledge found in many books and papers. For instance, a recent paper that provides a detailed presentation of works utilizing these and even more classifiers in the broader field of Remote Sensing is:
"Computer Vision and Pattern Recognition for the Analysis of 2D/3D Remote Sensing Data in Geoscience: A Survey".
Another, minor observation is that in Section 2 the subsections are referred as 3.x, where x=1,2,3,4.
Author Response
"Please see the attachment."
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Reviewer’s comments
After reviewing the author's revision, it remains unclear regarding the innovation, novelty, or contribution of this research. It appears that the author's primary objective is to test machine learning algorithms for plant classification (14 classes). In an article, authors are expected to introduce innovation, novelty, or specific problems. However, these aspects have not yet been clearly presented in this research.
The authors indicate that this research seeks to introduce machine learning algorithms for classifying high environmental dynamics encompassing 14 distinct classes. Numerous studies have previously explored the application of machine learning algorithms for land use and land cover classification. Hence, it is crucial to understand the specific challenges posed by classifying high environmental dynamics with the 14 classes that this research aims to address. Furthermore, it would be beneficial to discern how this study contributes innovation or a novel method for classifying all 14 classes, or alternatively, what complexities are involved in classifying data across these 14 classes.
Based on the study results, the authors show the results of data classification studies with training and validating using the intersected segments as samples and training and validating using entire segments as samples. What assumptions did the authors make in designing these experiments? How will the results of this study be useful for further application?
In the introduction section, the author should include more of the literature review or relevant research in order to illustrate the problems or gaps in the research we want to address. or present new innovations to increase efficiency.
The explanation provided in the Materials and Methods section does not align with the flowchart of the methodology presented in Figure 4.
In Table 2, the description of the color symbol should be presented.
In the Discussion section, the content appears to be redundantly similar to that of the Results section, lacking substantial real discussion.
The Conclusions section is not concise and lacks brevity. I disagree with the author's insistence on maintaining this writing style.
I believe that moderate editing of the English language is necessary.
Reviewer 3 Report
My suggested comments have been incorporated into the new version of the paper. Despite my reservations regarding the novelty of the paper, I recommend its acceptance.
