DEVELOPMENT OF A MODEL TO CORRECT MULTI-VIEW ANGLE ABOVE WATER MEASUREMENTS FOR THE ANALYSIS OF THE BIDIRECTIONAL REFLECTANCE OF CORAL AND OTHER REEF SUBSTRATES
Keywords: Correction, Algorithms, Model, Theory, Marine, Radiometry, Optical, Hyperspectral
Abstract. Spectral reflectance characteristics of substrates in a coral reef environment are often measured in the field by viewing a substrate at nadir. However, viewing a substrate from multiple angles would likely result in different spectral characteristics for most coral reef substrates and provide valuable information on structural properties. To understand the relationship between the morphology of a substrate and its spectral response it is necessary to correct the observed above-water radiance for the effects of atmosphere and water attenuation, at a number of view and azimuth angles. In this way the actual surface reflectance can be determined. This research examines the air-water surface interaction for two hypothetical atmospheric conditions (clear Rayleigh scattering and totally cloudcovered) and the global irradiance reaching the benthic surface. It accounts for both water scattering and absorption, with simplifications for shallow water conditions, as well as the additive effect of background reflectance being reflected at the water-air surface at angles greater than the critical refraction angle (~48°). A model was developed to correct measured above-water radiance along the refracted view angle for its decrease due to path attenuation and the "n squared law of radiance" and the additive surface reflectance. This allows bidirectional benthic surface reflectance and nadir-normalised reflectance to be determined. These theoretical models were adapted to incorporate above-water measures relative to a standard, diffuse, white reference panel. The derived spectral signatures of a number of coral and non-coral benthic surfaces compared well with other published results, and the signatures and nadir normalised reflectance of the corals and other benthic surface classes indicate good class separation.