Mapping the Coral Reef Habitats of Navassa Island I: Single Beam Acoustics

Gleason, Miller, McClellan, Reid

Navassa Island is a unique element of the Caribbean Islands National Wildlife Refuge because it is the only one of the Refuge's 14 components that includes jurisdiction over marine habitats. Development of management plans for Navassa have been hampered by a lack of information on the extent and character of the reef habitats surrounding the island. Expeditions by NOAA Fisheries and U.S. Fish and Wildlife personnel in 2002 and 2004 began an inventory of Navassa's marine resources, but a systematic exploration of the shelf around the island and a statistically rigorous census of ecologically and economically important organisms such as corals, fishes, conch, and lobster were impossible on those trips because no map of the benthic habitats existed for the island.

Single beam acoustic data were acquired around Navassa in 2004 to provide the basis for a habitat map of the shelf for depths between 15 and 60 m. These data were then processed to create a map that is compatible with the habitat classification and stratified random sampling scheme used by NOAA Fisheries for fish census surveys in the Florida Keys. The classification method involved three steps. First, each acoustic echo was labeled as hard bottom or bare sediment using the commercial clustering software QTC Impact. Second, each acoustic echo was compared with its neighbors to classify it based on the vertical relief and the relative proportion of bare sediment to hard bottom in the surrounding neighborhood. Finally, the individual echoes were aggregated to a 200 m x 200 m grid to convert point samples to a continuous habitat map.

The acoustic habitat map of Navassa agrees in most respects with "ground truth" provided by divers and with qualitative comparisons with satellite imagery. The major discrepancies are a) in areas of hard bottom with a thin veneer of sediment cover, and b) related to the choice of neighborhood and aggregated grid sizes, which are the tunable parameters of the classification. Future work to improve the technique should therefore focus on a) use of alternate / additional acoustic frequencies to penetrate shallow layers of sediment, and b) incorporating multiple scales within the definitions of habitat classes.