Satellite remote bathymetry : a study of quantitative models

An innovative approach to construct mathematical models for mapping bottom features in coastal shallow waters with satellite data (remote bathy- methy) is presented. Most remote bathymetric models have been on the basis of the simple relationship that the sensor-recorded energy reflected from water, after eliminating atmospheric and water column effects by deducting the contribution from deep water, is inversely proportional to the depth of water. This elementary bottom reflectance-based theory depends on a high bottom reflectance, a high water clarity, and/or shallow depths. Research has shown that the bottom-reflected energy levels are too low to be sensed by satellite sensors if these conditions are not met. This has greatly limited the application of remote bathymetric techniques because most coastal waters are somewhat turbid and have poorly-reflecting bottoms, and bathymetric charting usually requires water depth measurements beyond merely a few feet and in less than optically ideal waters. Further study also reveals that the approach of subtracting deep water radiances, a mainstay of the bottom reflectance- based theory, can introduce error in remote bathymetric modeling due to the uniform deduction of a constant from pixels with lower water column-radiant energy contribution.
This study presents a new mechanism for remote bathymetric modeling that overcomes these inherent deficiencies of previous remote bathymetric techniques. Spectral observations show that the emergent radiance from water is dominated by water column scattering rather than by the bottom reflection, unless the water is very shallow and transparent or overlies a highly reflecting bottom. Observation of this phenomenon has made it possible to develop a water column scattering-based remote bathymetric model which can be applied to turbid and somewhat deep coastal waters. The singly scattered irradiance model was adopted in this modeling. Analysis of both satellite (Landsat TM and SPOT-XS) and in .rirw-observcd data indicates that the water column scattering-based model can describe the relationship between the reflected radiance from water and water column depth much better than the bottom reflectance-based model. The model has been applied to Landsat TM image to calculate predicted water depths of the study area. The resulting depth map shows a significant agreement with the sea truth data of the study area.