Satellite-derived bathymetry for nearshore benthic habitats in Timor-Leste

Bathymetric data derived from multispectral, high-resolution (2 m) satellite imagery from DigitalGlobe’s WorldView-2 satellite to provide near complete coverage of nearshore terrain along the northern coastline of Timor-Leste, including the district of Oecusse to the west of mainland Timor-Leste and Atauro Island. Satellite images acquired by the NOAA Coral Reef Ecosystem Program (CREP) for the region are from Feb 9, 2010 to October 2, 2013. Methods used by CREP to estimate depths from the WorldView imagery were adapted from instructions developed by Kyle Hogrefe for using IKONOS imagery to derive seafloor elevations in optically clear water (Ehses and Rooney 2015).

Bathymetry values shallower than ~20 m were derived by gauging the relative attenuation of coastal, blue, green and yellow spectral radiance as a function of depth. A multiple linear regression analysis of coastal, blue, green and yellow band spectral values against in-situ depth determined the variables of y-intercept, coastal, blue, green and yellow slope values. The variables for each band were then used in a multivariate slope intercept equation to derive depth over the imagery. Variables and combinations of the bands were adjusted to improve the statistical accuracy and spatial coverage of the final derived bathymetry product. Digital image processing to derive depths was conducted with ENVI, and editing and integration was performed using Esri's ArcGIS.

The method assumes uniform water clarity but deviations from that condition made extraction difficult in water depths greater than 20 m. Results show that biotic material, or sediment in the water column skewed results shallower if the material has a high albedo and deeper if the material has a low albedo.