Rapid Hard-bottom Identification Using Unsupervised Classification of Multibeam Bathymetry and Backscatter in Raja Ampat Waters, Indonesia

Hard bottom communities contribute significantly to the underwater marine-system. Thus, this has emphasized the importance of hard-bottom delineation and mapping. In Indonesia, these ecosystems spread in the entire country, specifically in eastern waters of Indonesia. Multibeam sonar survey is a proper technology to map the hard-bottom distribution. This technique can generate bathymetry Digital Elevation Model (DEM) and backscatter imagery. In this study, texture analysis of multibeam backscatter imagery and analysis of Bathymetric Position Index (BPI) of DEM bathymetry are presented. Following that, features selection is carried out to obtain the best features used in classification. Those second-order texture features are used to undertake a K-means unsupervised classification to distinguish between the hard-bottom and soft-bottom objects, which then is combined with the BPI feature. Results show that the combination of unsupervised classification and morphometric analysis can offer a favorable output in hard-bottom identification. This study could contribute as a supporting method to identify and quantify the hard-bottom distribution with limited in-situ sampling