Hyperspectral Discrimination of Healthy Versus Stressed Corals Using In Situ Reflectance

Abstract

Remote sensing technology has many attributes that would be beneficial to monitoring submerged coral reef ecosystems such as the ability to revisit a site repetitively and consistently. One of the limiting factors, however, is the lack of a quantitative means of discriminating optically similar features such as healthy coral and algae-covered surfaces. We have shown previously that common coral reef features are optically similar to the naked eye, but can be discriminated using hyperspectral technology, thus reducing misclassification. The greatest source of classification error is a result of optical confusion between healthy coral and algae-covered surfaces even when high spectral resolution data is considered. This is an important distinction to make since an unhealthy coral is susceptible to rapid colonization by macroalgae and remote monitoring missions must be able to distinguish between these ecosystem phases. A high spectral resolution in situ library is presented and evaluated for potential use as a classification system for hyperspectral remotely sensed images of coral reef environments.