Primary supervisor: Dr Myriam Lacharité
Intended location(s) of student: Taroona
Background and Research Aims:
In Tasmania, the health and extent of seagrass beds are monitored for potential impacts due to increased nutrient load in the water from anthropogenic sources (e.g. storm water drain, sewage outflow and aquaculture sites). This can result in changes in the spatial extent and configuration of seagrass and increases in epiphytic cover. Underwater cameras are used to monitor these changes, but their narrow spatial footprint limits their use to estimate the extent and seascape configuration of seagrass beds. Unoccupied aerial systems (UAS; e.g. drones) provide high-resolution imagery in coastal systems at relatively low costs. UAS imagery has mostly been collected in exposed intertidal systems or shallow tropical waters; however, there is growing interest for their use in temperate turbid waters subjected to sub-optimal atmospheric conditions (e.g. high cloud cover, rain, swift wind).1 When operated under optimal conditions, UAS imagery could allow a complete characterisation of bed extent in temperate shallow waters (e.g. < 5 m).
Monitoring requires robust detection limits and objective, repeatable methodology. In this project, the student will determine the feasibility of employing automated classification methods to identify seagrass beds (against non-seagrass epiphytes and macrophytes, and sediment) and delineate their spatial configuration in UAS imagery.
The student will use UAS imagery collected with a DJI Phantom 4 coupled with in-situ ground-validation data from an underwater drop-camera. Imagery will be collected under similar environmental conditions (e.g. wind speed, cloud cover, sun angle, tidal height), and at different times (twice for each site). Imagery will be collected in the vicinity of finfish aquaculture leases in southeast Tasmania: two sites near finfish leases and two sites further away from leases.