Ningaloo Marine Park: Coral reef and coastal zone maps achieved with hyperspectral, remotely-sensed data
Kobryn, H., Wouters, K., Beckley, L.E. and Heege, T. (2011) Ningaloo Marine Park: Coral reef and coastal zone maps achieved with hyperspectral, remotely-sensed data. In: 48th Annual Conference of the Australian Marine Science Association, 3 - 7 July, Fremantle, Western Australia.
Effective management and monitoring of large marine protected areas requires detailed baseline data about the distribution of marine habitats. Large areas with complex geomorphology and clear waters, such as Ningaloo Marine Park (NMP) in Western Australia, naturally lend themselves to the application of optical remote sensing (hyperspectral data) as a means of gathering information about benthic habitats, bathymetry and adjacent coastal environments. By analysis of HyVista hyperspectral data (3.5m pixels with 0.4-2.5 Um spectral range), we were able to retrieve very detailed information about Ningaloo Reef (down to 20m depth). Data products include hierarchical thematic classification with up to 46 benthic cover classes as well as perclass probability and objected-oriented classification which combined benthic classes, depth, slope and aspect. From these we established that:
• over 50% of the substrate cover is composed of macro- or turfing algae;
• hard and soft coral cover makes up only 7% of the study area;
• 52% (875ha) of the coral cover is located within the sanctuary zones;
• the majority of the corals are a mix of different densities or morphologies of Acropora;
• within pixels, the majority of the hard coral cover occurs as either very dense cover (>90%) or with a patchy distribution (20-45%).
Examination of the adjacent coastal areas showed that vegetation cover was highly variable with evidence of anthropogenic influence through 4WD tracks, erosion and weed invasion. Pastoral leases had the highest track densities (8.1 km per km2) whereas, in contrast, Cape Range National Park had the lowest track density (0.3 km per km2). Through the use of operational pre-processing of the data and a semi-automatic image classification approach we created seamless high resolution baseline maps for 3400km2 of the Ningaloo region which are particularly suitable for ongoing monitoring and management.
|Publication Type:||Conference Item|
|Murdoch Affiliation:||School of Environmental Science|
|Item Control Page|