Proceedings Article | 19 October 2016
Proc. SPIE. 9999, Remote Sensing of the Ocean, Sea Ice, Coastal Waters, and Large Water Regions 2016
KEYWORDS: Data modeling, Scattering, Signal attenuation, Water, Remote sensing, Error analysis, Reflectivity, Computer simulations, Multispectral imaging, Image quality
Until now, Indonesia only had a single large-scale bathymetry map (1:250.000) for the entire nation and detailed maps (1:50.000 or 1:25.000) of a few locations. A straightforward and cost-efficient bathymetry mapping of the Indonesian coral reef is urgently needed. To address this problem, we present a generalized multispectral bathymetry estimation formula that requires few field measurement data to adjust coefficients. The simulation dataset corresponding to the Indonesian coral reef and Worldview-2 Imagery spectral response was built to extract the coefficient. Then, the formula was validated using real Worldview- 2 images from three shallow coral reef sites: the Gili Mantra islands, Menjangan Island, and Gondol Beach in Indonesia. During the evaluation of the dataset simulation, the formulated equation was tested; it achieved an adjusted R2 of 0.93 and an RMSR (Root Mean Square Residual) of 0.9 m. Such results show the good quality of our generalized bathymetry formula. Moreover, we tested the described formula using the imagery. The Gili islands shows that the best estimations of the depth corresponded to 0.62 and 1.5 m for R2 and RMSR, respectively. In the shallow areas, the depth estimation was accurate, but the error gradually increased with depth. The depth estimation for the Menjangan and Gondol sites failed (R2 ≤ 0.1), mostly because of a high image noise. In conclusion, the formula was able to estimate the depth accurately in a shallow water area using the least noisy image, but its error increased with depth.
