A Species-level Automated Monitoring Tool for Coral Reef Ecosystem Management

We propose to develop a new fluorescence-enhanced, AUV mounted, high resolution 3D laser imaging system for coral reef ecosystem management and research. Advantages of this system will include: enhanced automated discrimination of bottom cover type (e.g., live standing coral, dead standing coral, coral rubble, algae); detection and counting of coral recruits, direct determination and mensuration of 3D reef structure; autonomously repeatable survey tracks for long term recovery, resiliency and growth studies. Several studies have demonstrated that fluorescence imaging of reefs can provide enhanced automated discrimination between dead and live coral and algae cover, and potentially also of other seafloor organisms based on distinctive fluorescence properties. The combination of the fluorescence spectral signal with shape and structure information from the 3D imaging LiDAR data collection and the use of machine learning will provide a powerful multidimensional tool for bottom cover classification. The system will build upon our existing AUV compatible single-channel 3D laser imager. Since classification is done to the pixel level, area coverage can be assessed with many more data points than can be achieved by conventional diver-based survey techniques, and with more automated classification than existing color video analysis approaches. Making repeated surveys at user-selected time intervals will enhance the ability to monitor changes in reef community and physical structure.