The comparison of satellite image, UAV and geotagging photos for recognizing seagrass

Abstract

Abstract Seagrass beds are plant communities that have adapted to the depths of shallow seawater near the coast. These communities play a signi:icant role in the global carbon cycle, making it essential to monitor their existence. This research aims to visually compare seagrass recognition from three mapping technologies to determine their existence at the previous restoration site. We used uncrewed aerial vehicles (UAV) or drone photos taken in shallow water near Badi Island and processed them using photogrammetric software. This process created a cloud of pixels in virtual 3D space, providing a topographic map of the study area. Orthomosaic maps using stitching techniques were created, displaying shallow water landforms near the coast and various substrates. Many geotagged photos were collected at the study site. The elapse time technique, which involves taking photographs at regular intervals, was used to photograph benthic substrates at 5-second intervals following the bathymetric tracking lane, a predetermined path for underwater vehicles. The images were then geotagged using a time reference from a simultaneously operating mapsonder GNSS. This study result shows that drone orthomosaic maps and Sentinel 2 images have advantages in generating synoptic information, but the spatial resolution of drone orthomosaic photos surpasses Sentinel 2 images. Regarding detailed recognition of seagrass objects, photo geotagged techniques surpass orthomosaic drone maps and Sentinel 2 Natural Composite Color imagery. The appropriate altitude of drone photo for detail seagrass recognition was approximately 5 m or at GSD less than 10 mm/pixel. This study indicates that geotagging techniques should be used with drone mapping technology when seagrass restoration sites require preliminary information on existing species.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint