Advanced Detection and Classification of Kelp Habitats Using Multibeam Echosounder Water Column Point Cloud Data DOI Creative Commons
Amy W. Nau, VL Lucieer, Alexandre C. G. Schimel

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(3), P. 449 - 449

Published: Jan. 28, 2025

Kelps are important habitat-forming species in shallow marine environments, providing critical habitat, structure, and productivity for temperate reef ecosystems worldwide. Many kelp currently endangered by myriad pressures, including changing water temperatures, invasive species, anthropogenic threats. This situation necessitates advanced methods to detect density, which would allow tracking density changes, understanding ecosystem dynamics, informing evidence-based management strategies. study introduces an innovative approach with multibeam echosounder column data. First, these data filtered into a point cloud. Then, range of variables derived from cloud data, average acoustic energy, volume, density. Finally, used as input Random Forest model combination bathymetric classify sand, bare rock, sparse kelp, dense habitats. At 5 m resolution, we achieved overall accuracy 72.5% Area Under the Curve 0.874. Notably, our method high across entire swath, only 1 percent decrease falling within part impacted sidelobe artefact noise, significantly expands potential this type wide-scale monitoring threatened ecosystems.

Language: Английский

Advanced Detection and Classification of Kelp Habitats Using Multibeam Echosounder Water Column Point Cloud Data DOI Creative Commons
Amy W. Nau, VL Lucieer, Alexandre C. G. Schimel

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(3), P. 449 - 449

Published: Jan. 28, 2025

Kelps are important habitat-forming species in shallow marine environments, providing critical habitat, structure, and productivity for temperate reef ecosystems worldwide. Many kelp currently endangered by myriad pressures, including changing water temperatures, invasive species, anthropogenic threats. This situation necessitates advanced methods to detect density, which would allow tracking density changes, understanding ecosystem dynamics, informing evidence-based management strategies. study introduces an innovative approach with multibeam echosounder column data. First, these data filtered into a point cloud. Then, range of variables derived from cloud data, average acoustic energy, volume, density. Finally, used as input Random Forest model combination bathymetric classify sand, bare rock, sparse kelp, dense habitats. At 5 m resolution, we achieved overall accuracy 72.5% Area Under the Curve 0.874. Notably, our method high across entire swath, only 1 percent decrease falling within part impacted sidelobe artefact noise, significantly expands potential this type wide-scale monitoring threatened ecosystems.

Language: Английский

Citations

0