Evaluation of WorldView-2 and acoustic remote sensing for mapping benthic habitats in temperate coastal Pacific waters

Maps of nearshore marine habitat are fundamental tools for the management and conservation of coastal ecosystems. While traditional field mapping techniques, such as towed video and diver surveys, are still commonly employed for generating highly detailed maps of seafloor habitat, acoustic and satellite remote sensing have proven to be efficient alternatives for generating habitat maps. To date remote sensing using satellite imagery has dominated mapping efforts over coral reefs and other tropical ecosystems with fewer studies applied in temperate marine regions where acoustic studies are more common. Few studies exist that have assessed the performance of high resolution satellite imagery in mapping seafloor habitat in temperate regions. This paper compares the efficacy of high resolution satellite imagery (WorldView-2) and a single-beam acoustic ground discrimination system (QTC View V) for mapping the distribution of submerged aquatic vegetation at a site within the Gwaii Haanas National Marine Conservation Area (GHNMCA) off the north coast of British Columbia, Canada. Ground-truth data for training and validation were collected using a towed underwater video camera. Prior to classification the WorldView-2 image (8 bands, 2 m resolution) was processed following orthorectification, atmospheric correction, glint correction, land and optically deep water masking. An acoustic survey was conducted using a 200 kHz echosounder and data were processed and interpolated using QTC IMPACT software. The WorldView-2 imagery performed best in mapping habitat in regions shallower than 3 m, obtaining a total accuracy 75%, where it could identify the distribution of green algae (Ulva spp.), brown algae (Fucus spp.) and eelgrass (Zostera marina). The 200 kHz data were unable to detect the distribution of brown and green algae but were able to map the distribution of eelgrass as well as a subtidal red algae (Chondrocanthus exasperatus) (total accuracy 80%). A final habitat map containing all habitat types present at the study site was produced using the output from both datasets. The study resulted in recommendations for remote mapping of submerged vegetation in temperate coastal area.