Coral reefs are among the most biodiverse and vital ecosystems due to their ecosystem services. Effective coral reef management requires timely and detailed information for decision- making. Field-based research often covers limited areas and may not represent overall habitat composition. This study explores temporal and spatial trends in annual benthic composition on Heron Reef, Southern Great Barrier Reef, Australia, from 2002 to 2023 by integrating field data with satellite imagery. Annually, 3000-5000 photo quadrats were collected and analyzed using machine learning for benthic composition. Field-data were split into training and validation, with high-resolution multispectral imagery and Object-based image analysis producing annual benthic habitat maps. The overall accuracy was 67% over the 21-year period. Results indicated coral dominance on the reef slope and rock and rock/coral prevalence in the outer reef flat. Despite challenges of habitat heterogeneity and spectral similarities, remote sensing provided valuable insights into coral reef ecosystems with high spatial and thematic detail. The study emphasizes the importance of remote sensing in monitoring coral reef and highlights the need for further research to overcome classification inconsistencies and enhance monitoring capabilities. This work advances our understanding of coral reef dynamics and the role of remote sensing in marine ecosystem management.
promotor/supervisor feedback
nothing yet
