Benthic Habitat Mapping and Bathymetry Retrieval in the Shallow Water of Cham Island, Vietnam

Abstract

Coral reef ecosystems are under enormous pressure from human use and global climate change, particularly in the nearshore region. Cu Lao Cham Island and its surrounding water is a highly biodiverse marine area, it is home to 290 coral species together 370 species of underwater plant life, and nearly 280 reef fish species. Human activities like indiscriminate fishery and excessive tourism activities have had a great impact on these coral reefs. Thereby breaking the structure of marine ecosystems, leading to a decline in the number and types of aquatic and marine products. Therefore, high spatial resolution benthic habitat information is essential for coral reef protection and management in Cham Island. Planet Scope (PS) with very high spatial resolution (3 m) and temporal resolution (almost daily) can provide an ideal system for benthic habitat and seagrass mapping. In our study, several algorithms were applied with PS to provide bathymetry estimation, bottom reflectance retrieval, and per-pixel classification algorithms (Vector Machine - SVM, and Random Forest - RF) to identify different benthic compositions in the shallow water of Cham Island. The field measurements were used to validate classification results. Our results illustrate the benefit of using the bottom reflectance to discriminate the benthic features, with overall accuracy of 90.95% with SVM, and 90.55% with RF.