Centennial-Scale Land-Cover Change on Babeldaob Island, Palau

Round 1

Reviewer 1 Report

The study compiles, through available maps and imagery data, the changes in land use in Palau. The authors can improve the manuscript by considering the following issues.

1. The authors indicate that maps are already available showing the land use of the area. Why is the need for satellite image analysis? What is the expected contribution? This should be emphasized in the introduction.
2. The authors need to cite more recent studies to justify their study and also discuss their results in the context of the existing literature. The international context section is expected to discuss the differences and similarities of the finding with the existing studies. However, it discussed mainly the threats facing forest ecosystems. 
3. The images were classified using supervised classification. Which algorithm did the authors use? The authors need to provide more details about the algorithm since they claim the high-resolution satellite images could not achieve a better level of accuracy than the existing maps.
4.  The thematic level of the analysis is very coarse. It is mainly forest and non-forest land. What about urban development?
5. The authors, in line 191, state that they cannot include mangroves in the temporal change analysis due to some problems. This should be part of their contribution. They should check the literature for how these problems are solved. They can state their limitation after doing the analysis. 

Author Response

Dear Reviewer 1, Thank you for your concise and clear list of comments and recommendations. 1) While there were available maps of the island, they did not show land use per se, other than agroforest, farmland, and urban cover in the official vegetation map.  The other datasets required various levels of digitizing, processing, and classification in order to summarize vegetation change on the island, which had not been done previously.  We added more details to the introduction to hopefully clarify that.  2) We added more references to the introduction to justify the study, and more references to the results to compare our outcomes to others internationally. Many of these are perhaps more relevant to the time period of our study ending in 2014, since we completed most of the work by 2012, and then spent years waiting for a good image to document changes after the major road completion.  3) We used maximum likelihood supervised classifications and added more details the classifications overall in the methodology and supplemental materials. However, we would prefer to mention the approach rather than including the actual equation for it, since several papers we found did not include it. We would prefer this to not be a highly technical remote sensing paper, and remote sensing practitioners will be familiar with this commonly used method.  4) The thematic level was course so that we could compare cover across the entire time period of the study, and we focused on forest and non-forest vegetation changes since that comprised the great majority of change on the island.  We had mentioned changes to non-vegetation but include more details in that section in the discussion mentioning urban cover specifically.  5) We analyzed the seaward mangrove fluctuations over the study period and completely reworked the mangrove discussion to incorporate that and compare with international mangrove studies. We also include a figure of change summary for mangroves in the supplemental materials and details about our methodological limitations in mangroves in that discussion section.  

Reviewer 2 Report

Comments are attached.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer 2, Thank you for your comments and recommendations. We were a bit surprised that you requested technical details about the methodology since as you mention, the study used standard approaches that are widely known. We were hoping to avoid writing a highly technical remote sensing paper, but your points are well taken, and we added most of the details that you requested to the methodology, and a flow chart of Landsat processing steps to the supplemental materials. We did not have a threshold for cloud cover or radiometric correction.  Our goal was to use totally cloud free imagery, but we ended up with near cloud free imagery for Landsat (1.6%) and minimal cloud cover for the Quickbird and WV-2 composites. We would also have preferred the composite images to be better balanced radiometrically, but the very high-resolution imagery mosaics came rather unbalanced, which we mention in the methodology caveats. There are references to the Landsat download site in the references, to the 2014 polygon-based USFS thematic map, to the official vegetation survey, and to all of the data we used in a public depository, which should become links in the final version, but as far as we know direct links to the other original image datasets do not exist.

We also added the ground truth points to a new figure and lat and long coordinates to Figures 1 and 2. We reworked the introduction as you recommended, and we feel that it is a more precise outline of the study now, including objectives and research gaps. We must apologize that we didn’t use your recommended citations or text related to runoff.  In line with your recommendation to streamline the introduction we added some brief text and more relevant citations (in this context) here related to soil retention and land cover change.  While it would be nice to generate some fine scale validation from other satellites, these are not easy to obtain, especially at the scale of resolutions that we used (the Himawari’s resolution is 500 m, and the HLS data was not available until 2021).  We waited years for newer satellite imagery to become available in order to document the post road completion changes and were hoping to use a newer and better composite image than the 2014 WV2 image, but after years of waiting in vain for that, we had no other options.  In addition, these classifications take a minimum of weeks, and we don’t have that time to submit that level of revision. We weren’t sure if the multi satellite  intercomparison you requested would be for the additional satellites you mentioned, or the three satellites that we used, but both are potentially problematic given the above-mentioned issues with new satellite imagery and the different resolutions and time periods of the satellites we used. The accuracy assessment table also provides a broad comparison of the three satellites we used and shows they were quite similar overall.  In any case, the main objective of the study was not to do an in-depth comparison of satellite classifications, but to utilize publicly available datasets to summarize land cover change and mention broad trends in addition to possibilities and limitations of a variety of methods. We did not look at topographic illumination specifically. The island is quite topographically complex, but the highest point is under 500 meters and there weren’t any obvious areas of hill shade to account for.  We did use a DEM to limit mangrove cover to low lying coastal areas, but it was probably too course to utilize for correcting whatever minimal topographic illumination effects there were.

Reviewer 3 Report

Centennial-scale Land Cover Change on Babeldaob Island, Palau (1921-2014)

Dear Author

The basic science of this paper has been conducted to a good and appropriate standard. The author and his team wrote this paper according to the journal scope. I mentioned some minor comments in the pdf file. Find to attached pdf is enclosed herewith.

Best Regards

 

Comments for author File: Comments.pdf

Author Response

Dear Reviewer 3, Thank you for your comments and suggestions. We removed the range of years from the title, which was requested by Reviewer 2, but it was not clear from your comment if this is the change you recommended.  We changed the wording in the abstract. We added the latitude and longitude of the island in the site description, and to figure 1.  The journal format for tables is not specifically mentioned in the author instructions, except to use the table option in Word to create tables.  We hope that these will be easy to format by the editors.   We broke up the sentences you highlighted. 

Round 2

Reviewer 1 Report

The authors have addressed the issues raised in the review. They should have highlighted the changes they made in the manuscript to facilitate the second review.

The authors should consider changing the color or the background of north arrows, scale bars, and the statement scales in some figures, especially Figure 8 to make them more readable. 

Author Response

Dear Reviewer 1,

Thank you so much for your time and attention in reviewing and helping us improve our work. Thanks as well for hanging in there to make another valuable suggestion for the minor revisions.  The lack of change highlights with the major revision was specifically my mistake, so please accept my personal apology for that.  We added updated figures for figures 4-8, with different colors or bold type to hopefully improve readability of north arrows, scale bars and statement scales.  However, the older versions are still in the folder, if the editors prefer to use any of those.  

Thanks and best regards,

Julian Dendy