TO_Lifuka habitat map_November2012_Aserix

Download Report

Transcript TO_Lifuka habitat map_November2012_Aserix

Benthic Habitat Mapping
Lifuka Island
Aseri Baleilevuka
OCEANS & ISLANDS PROGRAM
SOPAC-SPC
What is habitat mapping?
• A habitat map is basically a map of the different features of the ocean
floor.
• Marine habitat mapping is the synthesis of physical and biological data
necessary delineate the distribution and extent of marine biota and
their habitats.
• Marine habitat maps helps us recognise the importance of marine
habitats as sources of beach sediments.
Purpose for the Lifuka project:
1. To use as base maps to determine the composition of the island
sediment system.
2. Adds to the foundation of essential baseline information that can
inform the development and design of appropriate adaptation options
for Lifuka.
Habitat Mapping Process
Field Data Collection and Analysis
1. Ground truthing - field data collection (Drop camera videos &
subsurface photographs)
2. Control point survey
3. Videos & Snorkeling Photos analysis
4. Satellite image data processing
Benthic Habitat Mapping Production
1. Expert knowledge, Manual delineation
2. Unsupervised classification (ArcGIS v9.3)
Ground truthing: Drop camera videos
echosounder
Video surface console
• SeaViewer under water video camera
Still photo
Ground truthing: Snorkel photos
UQ-BRSG
Drybag containing GPS
camera
Position of
photos
determined by
synchronizing
GPS & photo time
using GPicSync
software
Resulting kml file
Reference Image Points survey
•
To minimize mis-registration between field data and image,
control points are measured using Trimble R8 system.
Videos & Snorkeling Photos Analysis
• Screenshots of drop camera videos taken
(change in habitat/every 30s).
• Classification scheme developed.
• Used to label each snorkelling photo and video
screenshot with a mapping category.
• Lat and long for each photo is noted.
• 33 videos – 310 photos
snorkel - 303 photos
607 photos
• Local knowledge
Ü
csv file of photo log opened in QGIS and
saved as shapefile output
Lifuka fishermen
interpret habitats
Photo classification examples
Seagrass
Live coral
Rubble
Algae
Satellite Image processing
• Multi-spectral high resolution
World View 2 imagery (50cm x
50cm pixels) acquired on 10 June
2011.
• Reduced from 64 to 24 bit for
processing
• Global Mapper software used for
adjusting contrast and
geometrically correcting raster
image.
•Image rectified with reference
points from Trimble R8
• Output:
TO_Lifuka_WV2_10JUN2011_24bit_adjusted_contrast_rectified.tif
Habitat map production
• The National Oceanic and Atmospheric Administration (NOAA) habitat
classification scheme is used – based on 3 attributes: geographic “zones”,
geomorphological structure and biological cover.
Habitat Map of Zone Types
• Manually delineated using NOAA‘s
description of zones
• Digitized in ArcGIS v9.3 at a scale of
1:2,500
Habitat Map of Structure Types
• Manually delineated using NOAA’s
description of structure types
• Digitized in ArcGIS v9.3 at a scale of
1:5,000
Mapping of Biological cover
Unsupervised classification in ArcGIS v9.3
• Island and area of no data masked
• Iso Cluster tool used to create 20 spectral
classes
• Maximum likelihood tool to perform
EQUAL priori classification on image using
signature file produced above
0
0.5
Ü
1
2
3
Kilometers
Habitat Map of Biological cover
• classes produced used to inform and
direct definitions of biological cover as
described by the NOAA classification.
•Videos and snorkelling photographs as
guide, categories defined.
• Digitized in ArcGIS v9.3 at a scale of
1:2,500
Other methods and softwares tested
• Training shapefile of 200 polygons (digitized in ArcGIS v9.3)
• Segmentation using Berkeley Image Segmentation software in
conjunction with WEKA data mining software (Open source)
• NOAA’s Habitat Digitizer Extension to ArcView
• Supervised classification in ArcGIS v9.3
Acknowledgements
• Habitat mapping project conducted by the whole Marine Survey team
(Ocean & Islands Program, SOPAC - SPC)
• NOAA’s Centre for Coastal Monitoring and Assessment (CCMA)
• The University of Queensland (Biophysical Remote Sensing Group)
Poster Map