Sentinel-2A multispectral image analysis for seagrass mapping in Bintan’s shallow water ecosystem: A case study of Teluk Bakau, Malang Rapat, and Berakit villages

Abstract

Seagrasses are in danger due to anthropogenic activities causing a reduction in seagrass area and global damage. It is crucial to map seagrass distribution, as there is limited information on its existence. Sentinel-2A satellite provides high-resolution multispectral data with a 10-m resolution. The study aimed to evaluate the ability of Sentinel-2A imagery from the Google Earth Engine (GEE) platform to classify with the random forest (RF) algorithm in mapping seagrass in shallow water and various other objects in the study area. The results showed the area's presence of seagrass, coral, sand, sand seagrass, and rubble. Also, the photo transect method was used for collecting field data. The random forest algorithm had an accuracy of 76% in classifying each of the five classes. The combination of Sentinel-2A imagery and random forest algorithms can provide insight into the status and distribution of seagrass.