Geographic Object-Based Image Analysis for Mangrove Species Distribution Mapping

Mangrove species inventory and mapping are essential to maintain and managing the biodiversity of a coastal forest ecosystem. However, due to the harsh environment of mangrove forests, using a conventional ground survey for mapping mangrove species is inefficient. This study used high-spatial-resolution remote sensing data to help us carry out the task more efficiently. The study aimed to depict the spatial distribution of mangrove species from a WorldView-2 image (2 m) based on geographic object-based image analysis (GEOBIA) in some areas of Clungup Mangrove Conservation (CMC), East Java, Indonesia. This site was selected to represent an intact and preserved mangrove forest where the mangrove species are naturally distributed. Field sampling was purposively employed to identify mangrove species and their physical characteristics. An image segmentation routine using multiresolution segmentation was applied to the WorldView-2 image and produced segments of primitive objects. To fill each segment with a mangrove species class, we implemented a nearest neighbor algorithm to classify each segment based on some selected mangrove species samples. We used all bands of WorldView-2 and derived some indices as the input variable for the classification process. This process successfully mapped three classes of mangrove species, Rhizophora apiculata, Rhizophora mucronata, and Ceriops tagal, with an accuracy of 81.