Ensemble Machine Learning on the Fusion of Sentinel Time Series Imagery with High-Resolution Orthoimagery for Improved Land Use/Land Cover Mapping

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Major concerns:

 

From Table 1, in my opinion, the stacking model has no significant improvement on overall accuracy over other three base learners, and it has to do more work to produce a model (stack the three base models like Figure 2 showed, which means more time consume). Besides, the stacking model did not perform better at the class level either. So I actually did not see some irreplaceable key advantages of the stacking models. Why authors insist to use it?

 

The authors have improved the LULC mapping by combining Sentinel and NAIP imageries. I think authors should compare their results with similar work in Discussion, to better draw a solid conclusion that the method was effective. Moreover, from the perspective of application, is the method applicable worldwide? It seems the NAIP is a program conducted in the United States, so does this mean this method has regional limitations?

 

Minor comments:

Line 38- ‘remote seeing’ -> ‘remote sensing’

Figure 6- Is it necessary to show and compare the classification results of the other three base learners?

Author Response

We have provided response to each questions/comments raised by reviewer in the attahced document"remotesensing-3071806_ResponseTo_Reviewer_07_2024_#1"

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper is innovative in its exploration of the latest machine learning methods and their applications to classification problems and other scientific fields. However, it poses a challenge for non-experts due to the extensive use of specific ML terminology without sufficient explanation, justification, or references. Providing these could make the paper more accessible to the journal's diverse readership, which spans various aspects of remote sensing, including environmental applications, and not solely the technicalities of classification and machine learning methods.

Therefore, in my opinion, the paper can be accepted for publication, provided the following improvements are considered:

1)       Lines 89-91

Terms such as bagging, boosting, and stacking were mentioned without more explanation on the meaning for ML algorithms. More info on these, were introduced at lines 269-271. I would explain better this concept at introduction with appropriate additional references.

2)       Geobia has been cited more times in introduction but then is not recalled when it was used on mapping classified image. For example, it is unclear to an unobservant reader how the map in Figure 5 was retrieved, from which tool, what its spatial resolution is, or what about other spatial metrics, such as the Minimum Mappable Unit if produced with GEOBIA. GEOBIA is mentioned only in the introduction and not further elaborated upon in the text.

3)       LLO-CV (Leave-Location-Out Cross-Validation) acronym has been never explicated.

Overall, the text should be made more understandable for a broader audience, particularly for readers interested in mapping essential elements of agricultural land using remote sensing.

Additionally, please review the formatting of lines 2014-2019 and ensure the caption style of tables and figures is consistent.

Comments on the Quality of English Language

The level of English is quite good, although I am not qualified for this as I am not a native speaker. The changes needed are mostly stylistic. Digressions and more simple sentences containing terminology definitions that better clarify the technical context would be desirable.

Author Response

We have addressed reviewer's questions and comments in the attached response to reviewer document "remotesensing-3071806_ResponseTo_Reviewer_07_2024_#2"

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed all my concerns clearly. Thanks. I believe this manuscript is ready to be published.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have responded positively to the suggestions made. As a result, the paper has been significantly improved. Therefore, in my opinion, it can be accepted for publication in the journal "Remote Sensing".