coastTrain: A Global Reference Library for Coastal Ecosystems

Round 1

Reviewer 1 Report

The paper developed a global reference dataset of 193,105 occurrence records of seven coastal ecosystem types—muddy shorelines, mangroves, coral reefs, coastal saltmarshes, seagrass meadows, rocky shoreline, and kelp forests, which is important value to support novel global mapping initiatives, promote validations of independently developed data products, and enable improved monitoring of rapidly changing coastal environments worldwide.

Author Response

RESPONSE: We thank the reviewer for highlighting the value of our contribution.

Reviewer 2 Report

This is a description of a new reference library, called coastTrain, which builds up upon a number of existing sources, and which aims at providing an open-access database for coastal ecosystems.

I enjoyed reading this manuscript, and I am sure that this new library will be of benefit to coastal managers, scientists, modelers, even the general public. I am, however, not convinced that the authors chose the right journal, given that remote sensing is playing only a minor role therein.

I did not find scientific flaws, or even mistakes, in this manuscript, as scientific aspects are not in its focus. Hence, I recommend publication, either in this or in another journal (focussing on coastal aspects rather than remote sensing). I further recommend adding a section, in which the authors elaborate on the benefit of coastTrain with respect to other, existing data bases, plus another section with a summary. The manuscript would certainly benefit from a case study, in which the authors demonstrate the use of coastTrain, and its advantage over other systems.

Author Response

This is a description of a new reference library, called coastTrain, which builds up upon a number of existing sources, and which aims at providing an open-access database for coastal ecosystems.

2.1       I enjoyed reading this manuscript, and I am sure that this new library will be of benefit to coastal managers, scientists, modellers, even the general public. I am, however, not convinced that the authors chose the right journal, given that remote sensing is playing only a minor role therein.

RESPONSE: Our dataset was primarily designed to support remote sensing analyses of coastal ecosystems. The authorship team has regularly published manuscripts in this journal, including several that have utilised these data (e.g. Bunting et al 2022, Global Mangrove Watch). We therefore we consider Remote Sensing the ideal journal to publish data critical for developing remote sensing analyses of coastal ecosystems.

2.2       I did not find scientific flaws, or even mistakes, in this manuscript, as scientific aspects are not in its focus. Hence, I recommend publication, either in this or in another journal (focussing on coastal aspects rather than remote sensing). I further recommend adding a section, in which the authors elaborate on the benefit of coastTrain with respect to other, existing data bases, plus another section with a summary.

RESPONSE: We thank the reviewer for the recommendation to publish our manuscript. Our manuscript includes a section describing the utility of databases such as coastTrain for environmental research (lines 60-68) and summarise the benefits of a coastal training library such as coastTrain for ecosystem remote sensing (lines 122-133)

2.3       The manuscript would certainly benefit from a case study, in which the authors demonstrate the use of coastTrain, and its advantage over other systems.

RESPONSE: Our manuscript is based around several case studies demonstrating the use of coastTrain, citing >10 publications where the data have been used for coastal ecosystem remote sensing (e.g. Bunting et al., 2022; Murray et al., 2019; Lyons et al., 2020).

Reviewer 3 Report

 

Thus review is conducted in accordance with the specifications of  the Remote Sensing Earth Observation Data (https://www.mdpi.com/journal/remotesensing/sections/earth_observation_data).

1.     This manuscript represents the production of a global reference library for 7 types of  coastal ecosystems, derived from occurrence records of 4 already-developed mapping initiatives, harmonized to consistent standards represented by IUCN. As such, this would be a valuable resource for researchers examining such systems, so I was excited to see the work (and reminds me of a project I was involved some years ago).

2. By the standards of the specifications, this work would presumably classify as “Data Descriptors.” As such, “Described datasets must to be publicly deposited prior to publication under an open license,… The link to the publicly hosted version of the dataset must be given in the paper.

3. The manuscript describes the class definitions and 4 source data sets in some detail.  But what the project itself did to actually produce its product (coastTrain v 1.0) by IUCN standards for class definition, harmonization, training and  validation is quite sparsely described.

4. For the all-important standard of  “ ….publicly deposited…” the manuscript is obtuse. Reference is made to versioned controlled on Github. Under the Data Availability Statement (small print, at the end) it states that version 1.0 is available from zenado, with updates on the Github site.

A quick web search to see if it comes up found only a different Github CoastTrain, which looks like a USGS effort, and independent of this one.  The 4 reference datasets all have nice websites, with clear pointers to content. 

5. Overall I found this work to be a very nice start, but pre-mature as a final product.

Author Response

Thus review is conducted in accordance with the specifications of  the Remote Sensing Earth Observation Data (https://www.mdpi.com/journal/remotesensing/sections/earth_observation_data).

3.1       This manuscript represents the production of a global reference library for 7 types of  coastal ecosystems, derived from occurrence records of 4 already-developed mapping initiatives, harmonized to consistent standards represented by IUCN. As such, this would be a valuable resource for researchers examining such systems, so I was excited to see the work (and reminds me of a project I was involved some years ago).

RESPONSE: Thank you for this positive assessment of our manuscript.

3.2       By the standards of the specifications, this work would presumably classify as “Data Descriptors.” As such, “Described datasets must to be publicly deposited prior to publication under an open license,… The link to the publicly hosted version of the dataset must be given in the paper.

RESPONSE: The DOI link to the dataset is included in the data availability section as per author guidelines (line 311). We also reference this in the main text (line 276). To clarify where the dataset link is in the manuscript we have added “see Data Availability Statement” to sections that describe the data archive (line 276).

3.3       The manuscript describes the class definitions and 4 source data sets in some detail.  But what the project itself did to actually produce its product (coastTrain v 1.0) by IUCN standards for class definition, harmonization, training and  validation is quite sparsely described.

RESPONSE: We have added a new sentence to indicate how the crosswalk procedure was conducted, which involved:

Line 154: Each contributor to coastTrain used the published ecosystem descriptions, conceptual models and considerable theoretical background provided by the IUCN Global Eco-system Typology (www.global-ecosystems.org) to ensure the crosswalked Ecosystem Functional Groups were appropriate for their data.

3.4       For the all-important standard of  “ ….publicly deposited…” the manuscript is obtuse. Reference is made to versioned controlled on Github. Under the Data Availability Statement (small print, at the end) it states that version 1.0 is available from zenado, with updates on the Github site.

A quick web search to see if it comes up found only a different Github CoastTrain, which looks like a USGS effort, and independent of this one.  The 4 reference datasets all have nice websites, with clear pointers to content.

RESPONSE: Searching for archived datasets via internet search engines is unlikely to return results for archived datasets. We advocate clicking the link to the DOI version of the dataset, which is provided in line 311.

3.5       Overall I found this work to be a very nice start, but pre-mature as a final product.

RESPONSE: We thank the reviewer for their assessment of our manuscript.

Reviewer 4 Report

This manuscript presents the estimation of the distribution, extent and change of coastal ecosystems, which is essential for monitoring global change. The spatial models developed to estimate the distribution of land cover types require accurate and up-to-date reference data to support model development, model training and data validations. But due to the labor-intensive tasks required to develop reference datasets, often requiring intensive campaigns of image interpretation and/or field work, the availability of sufficiently large quality and well distributed reference datasets has emerged as a major bottleneck hindering advances in the field of continental to global-scale ecosystem mapping. In order to enhance the ability to model coastal ecosystem distributions globally, the authors developed a global reference dataset of 193,105 occurrence records of seven coastal ecosystem types—muddy shorelines, mangroves, coral reefs, coastal saltmarshes, seagrass meadows, rocky shoreline, and kelp forests—suitable for supporting current and next-generation remote sensing classification models. The coastTrain version 1.0 contains curated occurrence records collected by several global mapping initiatives, including the Allen Coral Atlas (https://allencoralatlas.org), Global Tidal Flats (www.intertidal.app), Global Mangrove Watch (https://www.globalmangrovewatch.org) and Global Tidal Wetlands Change (www.globalintertidalchange.org). To facilitate use and support consistency across studies, coastTrain has been harmonized to the International Union for the Conservation of Nature’s (IUCN) Global Ecosystem Typology. CoastTrain is an ongoing collaborative initiative designed to support sharing of reference data for coastal ecosystems, and is expected to support novel global mapping initiatives, promote validations of independently developed data products, and enable improved monitoring of rapidly changing coastal environments worldwide. 

This research is very useful to provide a reference for monitoring coastal ecosystem for global mapping of coastal environmental changes.

Figures and tables are fine, but the similarity is not checked.

So it is suggested to make a checking of its similarity before accepted. 

Author Response

RESPONSE: We have thoroughly checked the revised manuscript during the revision process.

Reviewer 5 Report

The paper  coastTrain: a global reference library for coastal ecosystems Contain suitable data description, with sources well defined with data quality appropriate other researchers for coastal ecosystems can reuse wich. Therefore, I recommend publishing the manuscript in its current state.

Figure 1 has low quality. It isn't easy to recognize the coverage of ecosystems and the colours that correspond to each one. This figure could be improved considerably.
The work indicates the following: “In the coming years, coastTrain will continue to expand through continued outreach to the global coastal research community and international initiatives related to coastal conservation. “But it poorly describes what steps the methodology will follow to achieve these continued outreaches to the global coastal research community and international initiatives.
The methodology described in the Workflow (Figure 2) regarding constrain compilation could be further specified, e.g. in section 3.2 to elaborate on what precisely this is:​
“All data processing operations were 250 recorded in electronic lab notebooks which accompany the dataset in the versioned data 251 repositories.”

What about the Constrain update being a reliable tool in the context of climate change? Moreover, what about the considerations to be considered in some latitudes, such as macrotidal zones, eustatic variations, or sea level rise? 

Author Response

The paper  coastTrain: a global reference library for coastal ecosystems Contain suitable data description, with sources well defined with data quality appropriate other researchers for coastal ecosystems can reuse wich. Therefore, I recommend publishing the manuscript in its current state.

5.1       Figure 1 has low quality. It isn't easy to recognize the coverage of ecosystems and the colours that correspond to each one. This figure could be improved considerably.

RESPONSE: It is challenging to represent >190,000 globally distributed points in a single 5 panel figure. However, the purpose of this figure is to show the wide global coverage of the dataset, and we believe it is sufficient to meet this purpose. Data users are able to download the coastTrain dataset for thorough inspection in GIS if required. We would, however, welcome suggestions by the editor to ensure a high resolution version of this figure is made available on publication.

5.2       The work indicates the following: “In the coming years, coastTrain will continue to expand through continued outreach to the global coastal research community and international initiatives related to coastal conservation. “But it poorly describes what steps the methodology will follow to achieve these continued outreaches to the global coastal research community and international initiatives.

RESPONSE: We have changed this sentence to ‘targeted outreach’ to reflect that growing coastTrain will be a targeted effort focused on researchers and mapping initiatives where it is known that suitable data has been developed and used for the purpose of coastal ecosystem mapping.

5.3       The methodology described in the Workflow (Figure 2) regarding constrain compilation could be further specified, e.g. in section 3.2 to elaborate on what precisely this is:“All data processing operations were 250 recorded in electronic lab notebooks which accompany the dataset in the versioned data 251 repositories.”

RESPONSE: We added a new sentence here to describe the purpose and methods used for harmonising the coastTrain dataset.

5.4       What about the Constrain update being a reliable tool in the context of climate change? Moreover, what about the considerations to be considered in some latitudes, such as macrotidal zones, eustatic variations, or sea level rise?

RESPONSE: As stated in 298, we recommend users use the reference date information against each data point to ensure that models are trained from data corresponding to when the observations are made. This enables the dataset to be used in highly dynamic coastal areas that are subject to wide variation and change.

Round 2

Reviewer 2 Report

Still, I believe that a paper published in MDPI Remote Sensing should contain Summary and Conclusions. Please do add.

Author Response

Response: Thank you again for your helpful comments on our manuscript. We now include a conclusions section in our paper.

 

Reviewer 3 Report

Having such a global reference set that can be used for modeling/remote sensing would certainly be a  valuable resource. To be valuable, potential users must be able to find it. I take the authors' point that clicking on the DOI link takes you to the site, and good to have the more clear statement of direction. Good.

That said, I would *still* prefer that it could be found without needing to find the article first. This work is based on the extraction of information from 4 major data datasets. Each of these datasets are well-done, have very nice and informative websites. Each can be readily be found web searching.

Not the case here. Certainly creating such websites is a lot of work and likely beyond the scope covered here. Worth thinking about down the line.

 

Author Response

Response: Thank you for this helpful suggestion. On your advice, we have now developed a new website for this project (www.coasttrain.org). The website is currently being indexed by Google and should go live in approximately 48-72 hours. The site includes mirrored information from the Zenodo archive and the live Github repository, as well as the following data viewer app (https://murrnick.users.earthengine.app/view/coast-train). Thank you again for this suggestion.