INVESTIGATING THE EFFECTS OF RIVER DISCHARGES ON SUBMERGED AQUATIC VEGETATION USING UAV IMAGES AND GIS TECHNIQUES
- 1Department of Transdisciplinary Science and Engineering, Tokyo Institute of Technology, Tokyo, Japan
- 2Department of Geodetic Engineering, University of the Philippines, Quezon City, Philippines
- 3Marine Science Institute, University of the Philippines, Quezon City, Philippines
Keywords: SAV, UAV, DJI Phantom, seagrass, seaweeds, spatial interpolation, water quality
Abstract. One of the major factors controlling the distribution and abundance of marine submerged aquatic vegetation (SAV) is light availability. Reduced water clarity due to sediment loading from rivers greatly affects the health and coverage of seagrasses and seaweeds. Monitoring SAV using unmanned aerial vehicles (UAV) has been getting attention because of its cost-effectiveness and ease of use. In this research, a low-cost UAV was utilized to assess the impacts of river discharges on SAV in Busuanga Island, Philippines. Linear regression was performed to determine the effectivity and accuracy of UAV-based percent cover estimation compared to established field survey methods of monitoring SAV. Water quality was estimated in the study area by performing spatial interpolation methods of in situ measurement of turbidity, chlorophyll, temperature, salinity, and dissolved oxygen using a multi-parameter water quality sensor. Current velocity and tidal fluctuations were monitored using bottom-mounted sensors deployed near the river mouth and in seagrass and seaweed areas with relatively good water clarities. Four stations were surveyed using automated UAV missions which were flown simultaneously with field observations. Each station surveyed has varying distances from the river mouth. Results from the classification of the UAV data and field survey show that SAV is more abundant as the distance from the river mouth increases and the turbidity decreases. Classification overall accuracies of UAV orthophotos ranging from 87.91–93.41% were achieved using Maximum Likelihood (ML) Classification. Comparison of field-based and UAV-based survey of percent cover of seagrasses show an overestimation of 1.75 times from the UAV compared to field observations.