Multi-spectral mapping of reef bathymetry and coral cover; Kailua Bay, Hawaii

We used high-resolution, airborne, digital, multi-spectral imagery to map bathymetry and the percent of living coral in the nearshore marine environment of Kailua Bay, Oahu, Hawai'i. Three spectral bands, with centers at 488, 551, and 577 nm (each with a full-width half maximum of 10 nm), were selected for good water transmission and good coral/sand/algae discrimination. However, the third band (577 nm) was not used in the depth and bottom-type solutions. The spatial resolution of 1 m per pixel was selected to balance resolution with the size of the total data set. A radiative transfer model accounting for the optical effects of the atmosphere, ocean surface, water, and reflection off the ocean bottom substrates was applied to the multi-spectral images, normalizing multiple images to one another for a mosaic that spans the bay. Atmospheric parameters in the radiative transfer model were estimated from published values measured for similar environments. Water-attenuation coefficients for the model were determined from the observed spectral data values over the sand bottom type in the bay. Relative depth and bottom-type coefficients were derived by a method most simply described as the "differencing" of two spectral bands. Accuracy exceeding 85% in predicted depth was achieved to a depth of 25 in. Depth prediction errors were assessed with comparison to hydrographic survey data. Classification of bottom-type coefficients into seven "percent living coral" categories results in 77% overall accuracy tested by diver-obtained line-intercept transect data (ground truth). Bottom-type coefficients derived by the model were corrected for atmospheric and ocean conditions on the date of collection, so spatial changes in bathymetry and "percent living coral" through time can be analyzed and related to environmental factors. The radiative transfer model and the "differencing" method used to solve for depth and "percent living coral" can be applied to any airborne, passive remote sensing digital data with appropriate spectral bands.