Benthic habitats mapping in Pari Island, Kepulauan Seribu, Indonesia using drone and sentinel-2B imagery with object based image analysis method

Department of Marine Science and Technology, Faculty of Fisheries and Marine Science, IPB University, Bogor, 16880, Indonesia

^* Corresponding author: jppanjaitan@apps.ipb.ac.id

Abstract

Benthic habitats are living places and shelters for various types of aquatic organisms, including seagrass, seaweed, algae, dead corals, and living corals with different types of substrates, such as rubble, sand, and mud. This habitat has the potential to change, so it is necessary to monitor it regularly by mapping shallow-water habitats using remote sensing technology such as satellite imagery and drones. This study aims to classify and map benthic habitats in the shallow waters of Pari Island using drone and satellite images using the object-based image analysis (OBIA) classification method. A total of 1052 aerial photographs with a resolution of 3.6 cm / pixel and Sentinel-2B images with a spatial resolution of 10 m were processed using the support vector machine (SVM) algorithm. Both images produced six classes: live corals, macroalgae, seagrass, dead corals with algae, sand, and rubble. The overall accuracy test results for drone imagery and satellite imagery were 81.88% and 69.57%, respectively. Based on the overall accuracy, the drone images proved to be better than Sentinel-2 imagery.