Environmental Data Initiative logo
EDI Staging Environment

Data Package Metadata   View Summary

MCR LTER: Coral Reef: A 2018 land cover map of Moorea, French Polynesia

General Information
Data Package:
Local Identifier:knb-lter-mcr.6005.3
Title:MCR LTER: Coral Reef: A 2018 land cover map of Moorea, French Polynesia
Alternate Identifier:DOI PLACE HOLDER
Abstract:

Using a collection of imagery from June - 2018 taken by the Worldview-3 satellite, land cover was classified for the island of Moorea, French Polynesia. A deep learning pixel classification model was trained for each of four separate dates of image collection when clouds were sparse over the island. The model was trained at the native resolution of the imagery (<2m pixels). Training data included the multispectral WV-3 bands in addition to derived bands that index vegetation productivity (NDVI), vegetation texture (NDVI IDM), and water cover (NDWI). A consensus land cover map was generated from model predictions across the four sets of imagery.

This material uses data collected by the U.S. National Science Foundation's (NSF) Moorea Coral Reef Long Term Ecological Research (MCR LTER) site under Grant No. OCE 2224354 (and earlier awards). Additional financial support to the MCR LTER site was provided through a generous gift from the Gordon and Betty Moore Foundation. Research was completed under permits issued by the French Polynesian Government (Délégation à la Recherche) and the Haut-commissariat de la République en Polynésie Francaise (DTRT) (Protocole d'Accueil 2005-2024).

Publication Date:2024-09-23
For more information:
Visit: DOI PLACE HOLDER

Time Period
Begin:
2018-06-11
End:
2018-09-13

People and Organizations
Contact:Information Manager (Moorea Coral Reef LTER) [  email ]
Organization:Moorea Coral Reef LTER
Creator:John, Christian (University of California, Santa Barbara, Postdoc)
Creator:Neumann, Kyle (University of California, Santa Barbara)
Creator:Maritorena, Stéphane (University of California, Santa Barbara)
Creator:Burkepile, Deron (Moorea Coral Reef LTER)

Data Entities
Spatial Raster Name:
Classified layer
Description:
Consensus classification layer of land cover on Moorea, French Polynesia
Detailed Metadata

Data Entities

Spatial Raster

Name:Classified layer
Description:Consensus classification layer of land cover on Moorea, French Polynesia
Additional Info:

Raster derived coordinate reference system

+proj=utm +zone=6 +south +datum=WGS84 +units=m +no_defs

Physical Structure Description:
Object Name:consensusLandUse.tif
Size:14045092 byte
Authentication:404b4625141163108487dfdc0d33b1d8 Calculated By MD5
Externally Defined Format:
Format Name:tif
Coverage Description:
Geographic Region:
Description:Moorea, French Polynesia
Bounding Coordinates:
Northern:  -17.45Southern:  -17.62
Western:  -150.0Eastern:  -149.67
Spatial Reference:
Name of Coordinate System:WGS_1984_UTM_Zone_6S
Horizontal Accuracy:
Report:{0.931,0.977}
Vertical Accuracy:
Report:Unknown
Cell Size(X):1.4
Cell Size(Y):1.4
Number of Bands:1
Origin:Lower Left
Max Raster Objects(X):13126
Max Raster Objects(Y):9912
Max Raster Objects(Z):1
Cell Geometry:pixel
Attribute(s) Info:
 
Column Name:classnm  
Definition:Indicates the predicted land cover class (1 = Water; 2 = Forest; 7 = Monoculture; 8 = Agriculture; 9 = Pasture; 10 = FallowBuffer; 12 = Dirt; 13 = Sand; 14 = Paved; 15 = Buildings)
Storage Type:integer  
Measurement Type:ratio
Measurement Values Domain:
Unitdimensionless
Typereal
Missing Value Code:
Code4294967295
ExplPixels outside prediction ROI but within extent of bounding box
Accuracy Report:  
Accuracy Assessment:  
Coverage:  
Methods:  
Data:https://pasta-s.lternet.edu/package/data/eml/knb-lter-mcr/6005/3/34a21202a05c4b073988f40f6d3783f6

Data Package Usage Rights

This data package is released under the Creative Commons license Attribution 4.0

International (CC BY 4.0 , see

The Data User has an ethical obligation to cite the data source appropriately in any

publication or product that results from its use, and notify the data contact or creator.

Communication, collaboration, or co-authorship (as appropriate) with the creators of this

data package is encouraged to prevent duplicate research or publication. The Data User is

urged to contact the authors of these data if any questions about methodology or results

occur. The Data User should realize that these data may be actively used by others for

ongoing research and that coordination may be necessary to prevent duplication or

inappropriate use. The Data User should realize that misinterpretation may occur if data are

used outside of the context of the original study. The Data User should be aware that data

are updated periodically and it is the responsibility of the Data User to check for new

versions of the data.

While substantial efforts are made to ensure the accuracy of data and associated

documentation, complete accuracy of data sets cannot be guaranteed. This data package (with

its components) is made available "as is" and with no warranty of accuracy or fitness for

use. The creators of this data package and the repository where these data were obtained

shall not be liable for any damages resulting from misinterpretation, use or misuse of the

data package or its components.

Keywords

By Thesaurus:
(No thesaurus)LTER, MCR, Moorea Coral Reef LTER, machine learning, Moorea Coral Reef
Institute, Organization, or Funding AgencyCalifornia State University Northridge, Gordon and Betty Moore Foundation, NSF, Richard B. Gump South Pacific Research Station, University of California Santa Barbara
LTER Controlled Vocabularycoral reefs, corals, land use, land cover, remote sensing, satellite imagery, spatial variability

Methods and Protocols

These methods, instrumentation and/or protocols apply to all data in this dataset:

Methods and protocols used in the collection of this data package
Description:

A land cover map of Moorea, French Polynesia was generated using predictions from a pixel classification machine learning model.

Image preparation

Imagery from the Worldview-3 satellite was collected on June 11, June 30, July 12, and September 13, 2018. Images were aligned, georeferenced, and orthorectified using the Ortho Mapping workspace in ArcGIS Pro v3.1.2. Ground control points for georeferencing were generated using building data from the Direction des Affaires Foncières (DAF) de la Polynésie-Française (the Directorate of Land Affairs, French Polynesia). Tie points among images were drawn only on land and in lagoons within the extent of Mo’orea’s outer reef to prevent spurious points from being detected over open ocean. Imagery was terrain corrected (orthorectified) using the Moorea Island Digital Ecosystem Avatar (Moorea IDEA) digital terrain model (Gruen et al. 2017).

Description:

Model training

A cloud mask was generated manually by tracing the perimeters of clouds in each image. Training data were generated using a combination of land use data from DAF and photointerpretation guided by knowledge of the area. Classes included Water, Buildings, Paved, Sand, Dirt, Forest, Monoculture, Agriculture, Pasture, and Fallow/Buffer. Cloud-free portions of the aligned, terrain-corrected imagery were broken into 64x64 pixel chips with a 32x32 pixel stride and stored as classified tiles with classes defined by the training data. A U-Net pixel classifier was trained on a ResNet-50 architecture using transfer learning with the classified tiles for 20 epochs, with 10% of data withheld for validation. Model accuracy was 0.86 ± 0.04 (mean ± std. dev. across date-wise models).

Description:

Consensus map generation and validation

Predicted land cover for each date was projected to a common resolution and stacked. A consensus map was generated by taking the modal class for each pixel of the stacked layers. Pixels classified as Agriculture, Pasture, and Fallow/Buffer were aggregated into an umbrella “BufferAg” class to capture the high degree of conflation by models among these classes. The consensus map was validated using two sets of validation data. First, a spatially random reference sample of 100 points over the island of Moorea was classified using photointerpretation and compared with consensus model predictions at those locations. This validation approach yielded 98% classification accuracy. A second set of reference points was drawn by taking spatially random samples of equal sizes from each of the training data classes and manually reclassifying those points using photointerpretation. This approach yielded a more balanced collection of samples from less-represented land cover classes but was less representative of mean classification accuracy over the whole of the island. The second validation approach yielded 82% classification accuracy among the classes of interest, BufferAg, Buildings, Dirt, Forest, and Monoculture. Most of the discrepancies in classification came from errors of omission of exposed soil which was often co-occurring with intensive monoculture and captured by that class, and errors of omission of small buildings which were lost as noise to their surrounding vegetation buffer.

People and Organizations

Publishers:
Organization:Environmental Data Initiative
Email Address:
info@edirepository.org
Web Address:
https://edirepository.org
Id:https://ror.org/0330j0z60
Creators:
Organization:Moorea Coral Reef LTER
Address:
Marine Science Institute,
University of California,
Santa Barbara, CA 93106-6150 USA
Web Address:
http://mcr.lternet.edu/
Individual: Christian John
Organization:University of California, Santa Barbara
Position:Postdoc
Address:
Marine Science Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
cjohn@ucsb.edu
Id:https://orcid.org/0000-0002-6249-040X
Individual: Kyle Neumann
Organization:University of California, Santa Barbara
Address:
Marine Science Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
kyle.neumann@ostreaengineering.com
Individual: Stéphane Maritorena
Organization:University of California, Santa Barbara
Address:
Earth Research Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
stephane.maritorena@ucsb.edu
Id:https://orcid.org/0000-0002-5508-8391
Individual: Deron Burkepile
Organization:Moorea Coral Reef LTER
Address:
Marine Science Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
dburkepile@ucsb.edu
Id:https://orcid.org/0000-0002-0427-0484
Contacts:
Organization:Moorea Coral Reef LTER
Position:Information Manager
Address:
Marine Science Institute,
University of California,
Santa Barbara, CA 93106-6150 USA
Phone:
805.893.2071 (voice)
Email Address:
mcrlter@msi.ucsb.edu
Web Address:
http://mcr.lternet.edu/
Id:https://ror.org/017vyhs10
Metadata Providers:
Individual: Information Manager
Organization:Moorea Coral Reef LTER
Position:Information Manager
Address:
Marine Science Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
mcrlter@msi.ucsb.edu
Id:https://ror.org/017vyhs10

Temporal, Geographic and Taxonomic Coverage

Temporal, Geographic and/or Taxonomic information that applies to all data in this dataset:

Time Period
Begin:
2018-06-11
End:
2018-09-13
Geographic Region:
Description:Moorea, French Polynesia
Bounding Coordinates:
Northern:  -17.45Southern:  -17.62
Western:  -150.0Eastern:  -149.67

Project

Parent Project Information:

Title:The Moorea Coral Reef (MCR) LTER
Personnel:
Individual: Russell Schmitt
Organization:Moorea Coral Reef LTER
Position:Lead Principal Investigator
Address:
Marine Science Institute,
University of California,
Santa Barbara, CA 93106-6150 USA
Phone:
805.893.3777 (facsimile)
Email Address:
schmitt@lifesci.ucsb.edu
Web Address:
http://mcr.lternet.edu/
Id:https://orcid.org/0000-0002-9674-4244
Role:Principal Investigator
Individual: Sally Holbrook
Organization:Moorea Coral Reef LTER
Position:Principal Investigator
Address:
Marine Science Institute,
University of California,
Santa Barbara, CA 93106-6150 USA
Phone:
805.893.3777 (facsimile)
Email Address:
holbrook@lifesci.ucsb.edu
Web Address:
http://mcr.lternet.edu/
Id:https://orcid.org/0000-0002-3015-3484
Role:Principal Investigator
Individual: Peter Edmunds
Organization:Moorea Coral Reef LTER
Position:Principal Investigator
Address:
Department of Biological Sciences, California State University,
18111 Nordhoff Street,
Northridge, CA 91330-8303 USA
Phone:
818.677.2502 (voice)
Email Address:
peter.edmunds@csun.edu
Web Address:
http://mcr.lternet.edu/
Id:https://orcid.org/0000-0002-9039-9347
Role:Principal Investigator
Individual: Robert Carpenter
Organization:Moorea Coral Reef LTER
Position:Principal Investigator
Address:
Department of Biological Sciences, California State University,
18111 Nordhoff Street,
Northridge, CA 91330-8303 USA
Phone:
818-677-3256 (voice)
Phone:
818-677-2034 (facsimile)
Email Address:
robert.carpenter@csun.edu
Web Address:
http://mcr.lternet.edu/
Id:https://orcid.org/0000-0003-4232-1868
Role:Principal Investigator
Individual: Deron Burkepile
Organization:Moorea Coral Reef LTER
Position:Principle Investigator
Address:
Marine Science Institute, University of California,,
Santa Barbara, CA 93106-6150 United States
Email Address:
dburkepile@ucsb.edu
Id:https://orcid.org/0000-0002-0427-0484
Role:Principle Investigator
Funding:

NSF (LTER) MCR I #OCE-0417412, MCR II #OCE-1026851, MCR IIb #OCE-1236905, MCR III #OCE-1637396, and MCR IV #OCE-2224354

Additional Award Information:
Funder:National Science Foundation
Funder ID:http://dx.doi.org/10.13039/100000001
Number:0417412
Title:LTER: Long-Term Dynamics of a Coral Reef Ecosystem
URL:https://www.nsf.gov/awardsearch/showAward?AWD_ID=0417412
Additional Award Information:
Funder:National Science Foundation
Funder ID:http://dx.doi.org/10.13039/100000001
Number:1026851
Title:LTER: MCR II - Long-Term Dynamics of a Coral Reef Ecosystem
URL:https://www.nsf.gov/awardsearch/showAward?AWD_ID=1026851
Additional Award Information:
Funder:National Science Foundation
Funder ID:http://dx.doi.org/10.13039/100000001
Number:1236905
Title:LTER: MCR IIB: Long-Term Dynamics of a Coral Reef Ecosystem
URL:https://www.nsf.gov/awardsearch/showAward?AWD_ID=1236905
Additional Award Information:
Funder:National Science Foundation
Number:1637396
Title:LTER: MCR III: Long-Term Dynamics of a Coral Reef Ecosystem
URL:https://www.nsf.gov/awardsearch/showAward?AWD_ID=1637396&HistoricalAwards=false
Additional Award Information:
Funder:National Science Foundation
Number:2224354
Title:LTER: MCR IV: Long-Term Dynamics of a Coral Reef Ecosystem
URL:https://www.nsf.gov/awardsearch/showAward?AWD_ID=2224354&HistoricalAwards=false

Maintenance

Maintenance:
Description:

Data in support of manuscript

Frequency:notPlanned
Other Metadata

Additional Metadata

additionalMetadata
        |___text '\n      '
        |___element 'metadata'
        |     |___text '\n         '
        |     |___element 'importedFromXML'
        |     |        \___attribute 'dateImported' = '2023-10-27'
        |     |        \___attribute 'filename' = 'MCR_LTER_template_v2023_09_22.xml'
        |     |        \___attribute 'taxonomicCoverageExempt' = 'True'
        |     |___text '\n      '
        |___text '\n   '

Additional Metadata

additionalMetadata
        |___text '\n      '
        |___element 'metadata'
        |     |___text '\n         '
        |     |___element 'emlEditor'
        |     |        \___attribute 'app' = 'ezEML'
        |     |        \___attribute 'release' = '2024.07.17'
        |     |___text '\n      '
        |___text '\n   '

Copyright 2024 Environmental Data Initiative. This material is based upon work supported by the National Science Foundation under grants #2223103 and #2223104. Any opinions, findings, conclusions, or recommendations expressed in the material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. Please contact us with questions, comments, or for technical assistance regarding this web site or the Environmental Data Initiative. Please read our privacy policy to know what information we collect about you and to understand your privacy rights.

EDI is a collaboration between the University of New Mexico and the University of Wisconsin – Madison, Center for Limnology:

UNM logo UW-M logo