Application of soft classification techniques for coral habitat mapping using satellite imagery at Lang Tengah Island

Abstract

Coral reefs in islands around Malaysia are astonishingly rich in marine diversity. Remote sensing is a potential tool for the mapping, monitoring and management of the coral reef ecosystems. In this study, the remote sensing satellite imagery were used to determine coral reef substrates at Lang Tengah Island. Commonly, most marine scientists adopt simple techniques of remote sensing, i.e. hard classification method in their research. However, this study was conducted to examine the potential of soft classification technique to classify coral reef substrates and to compare their accuracy with hard classification technique. The hard classification technique used is the Maximum-Likelihood, whilst the two soft classifications techniques chosen for this study were Bayesian probability theory and Dempster-Shafer theory.