Spectral discrimination of coral reef environment using derivative analysis

Abstract

Coral reefs are often called the 'rainforests of the sea' because they are among the most diverse and complex of all ecosystems. Coral reefs play a vital role in the coastal and marine ecosystem. The Philippines has the second largest coral reef area in Southeast Asia which is estimated to be about 26,000 square kilometres. It lies within the Coral Triangle or the Indo-Malayan Triangle, a centre of marine biodiversity. However, reefs today are facing a wide and serious threats. Most of the coral reefs in the world are potentially endangered, mainly because of the unsustainable human exploitation exacerbated by global climate change. Mapping of coral reefs provides information to support the protection, conservation and monitoring of this vulnerable benthic habitat. However, typical maps derived using remote sensing data only includes classification of benthic communities due to the high level of complexity and spatial heterogeneity of the coral reef environment. Thus, there is a need to examine the reflectance properties or the spectral response of the different bottom types to serve as a reference for the characterization of the features. This study was conducted to identify spectral characteristics of various reef components by determining the wavelength bands in discriminating coral reef types which can aid in a high-resolution mapping of features. It also proposed a standardized methodology for field measurements of the different coastal habitats using a portable spectrometer in various sites of the Philippines. Spectral measurements of the different coral reefs were taken using the OceanOptics spectrometer unit with wavelength range of 200-1100 nanometres. The acquired spectral curves were divided into the following classes: live corals, dead corals, sand, dead coral with algae, coral rubbles, and bleached corals. Twenty-five in-situ spectral measurements for each sample were averaged and smoothed using the Savitzky-Golay algorithm. This method fits a least square polynomial curve in smoothing the data and is a function of filter size for the calculation of the curve equation, and the degree of polynomial to be used. After the smoothing process, the spectral properties of the measured benthic feature were evaluated using derivative analysis. This calculates the n-derivatives of the reflectance curves to identify the optimal wavelengths for spectral discrimination. Results showed that there are differences in the optical characteristics among the coral reef types which could be used to improve the accuracy of classification and an efficient aid in mapping corals in the country.