Data Management Plan

Benthic habitat classes were derived for nearshore waters (< 20 m depths) around Timor-Leste from DigitalGlobe WorldView-2 satellite imagery, acquired from February 9, 2010 to December 14, 2013, by the NOAA Coral Reef Ecosystem Program (CREP). Habitat classes include soft (sand or mud) and hard (rubble, boulders, etc.) substrates by depth (shallow, mid and deep); seagrass, mangrove, and macroalgae dominated areas; intertidal zones; lagoons; and emergent rocks. Supervised classifications were performed using several different methods (maximum likelihood, and minimum distance) with imagery at several different stages of processing (deglinted, radiance, reflectance and deglinted, depth invariant, sieve, and clump), with final classification based on the classification method and imagery variant that provided the best overall results for each WorldView-2 image.

Timor-Leste

Lineage Statement:
Benthic habitat classes were derived for nearshore waters around Timor-Leste from WorldView-2 satellite imagery. Habitat classes include different combinations of shallow, mid and deep depths, soft (sand or mud) or hard (rubble, boulders, etc.) substrates, macroalgae, seagrass or mangrove dominated areas, intertidal areas, lagoons, and emergent rocks. Supervised classifications were performed using several different methods and with imagery at several different stages of processing, with final classification based on the classification method and imagery variant that provided the best overall results for each WorldView-2 image.

Process Steps:

• Satellite imagery was collected by the WorldView-2 commercial Earth observation satellite owned by DigitalGlobe. WorldView-2 provides panchromatic imagery of .46 meter resolution, and eight-band multispectral imagery with 1.84 meter (6 foot 0 inch) resolution.
• Image pre-processing steps are described in the NOAA Technical Memorandum "Depth Derivation Using Multispectral WorldView-2 Satellite Imagery". Pre-processing of all WorldView-2 satellite images was performed using the same methods, to produce a deglinted image, radiance and depth invariant data layers. (Citation: Depth Derivation Using Multispectral WorldView-2 Satellite Imagery, NOAA Technical Memorandum NMFS-PIFSC-46)
• Each image data layer was loaded into ENVI and 13 different regions of interest (ROIs) were created. The ROIs are classified as followed: hard shallow, soft/sand shallow, hard mid depth, soft/sand mid depth, hard deep, soft/sand deep, seagrass, intertidal, mangroves, emergent rocks, macroalgae, and lagoon. Areas within the imagery unable to be classified were annotated as unknown. The Image classification was performed separately on the multi-band image using three different methods: 1) The Mahalanobis Distance Classification, which is a direction sensitive distance classifier. J.A.Richards, 1999, Remote Sensing Digital Image Analysis, Springer-Verlag, Berlin, p. 240. 2) The Maximum Likelihood Distance, which assumes that the statistics for each class in each band are normally distributed. 3) Minimum Distance, which uses the mean vectors for each class and calculates the distance from each unknown pixel the the mean vector for each class. More information about the Mahalanobis Distance, Maximum Liklihood Distance and the Minimum Distance algorithm can be found in the ENVI documentation. The results of each of the three classifications were compiled into an ENVI standard image. The outputs were exported in a geotiff format. (Citation: ENVI)
• Geotiffs were loaded in ArcGIS and all land and NoData values were removed using the Arc toolbox "extract by mask" tool. The outputs were the same eleven classes as the ROI's. Deep sea areas were manually digitized from classified images as a shapefile with reference to the satellite image as needed, then converted to raster files and mosaicked back onto classified images. Shallow lagoons were also manually digitized, converted to raster files , mosaicked back onto the classified imagery. This is required because the habitat derivation method does not work well on very shallow waters in Timor-Leste as these areas often have high turbidity levels. (Citation: ESRI ArcGIS Desktop 10.3, 'Extract by Mask’ tool)
• The complete methodology to classify benthic features using WorldView-2 satellite imagery. (Citation: Watkins, Russell, L., 2015, A Methodology for Classification of Benthic Features using WorldView-2 Imagery, Report prepared for the Ecospatial Information Team, Coral Reef Ecosystem Division, Pacific Islands Fisheries Science Center, Honolulu, HI, under NOAA contract number WE-133F-15-SE-0518, 29pp.)
• The methodology along with the results are described in the final report produced by CREP for the Government of Timor-Leste, Interdisciplinary baseline ecosystem assessment surveys to inform ecosystem-based management planning in Timor-Leste: Final Report. (Citation: PIFSC. 2017. Interdisciplinary baseline ecosystem assessment surveys to inform ecosystem-based management planning in Timor-Leste: Final Report. NOAA Pacific Islands Fisheries Science Center, PIFSC Special Publication, SP-17-02, 234p.)

An experienced outside contractor was hired to review and suggest modifications to improve the the initial methods for deriving benthic habitat data from WorldView-2 satellite imagery.

Three different modifications for both types of processing were performed on each image. A visual comparison of the 6 resulting geotiffs was made to select the one that best reflected the distribution of benthic habitats for each WorldView-2 satellite image.

Data can be accessed online via the NOAA National Centers for Environmental Information (NCEI) Ocean Archive

NOAA IRC and NOAA Fisheries ITS resources and assets.