Bathymetric information pertaining to oceans, inland lakes, and rivers is crucial to the safety of nautical navigation, coastal management, and various scientific studies of aquatic environments. Optical remote sensing imagery offers a cost-effective alternative to echo sounding and bathymetric LiDAR surveys for deriving high density bottom depth estimates for coastal and inland water bodies. Most previous studies utilized a global log-linear regression model to invert multi-spectral images into bathymetric data for an entire image scene. The performance of conventional global models is limited when the bottom type and water quality vary spatially within the scene, or when bottom albedo is low. To address the inadequacy of conventional log-linear global inversion models, I proposed two methods to improve the accuracy of depth estimates.
The first method, which is based on the Levenberg-Marquardt optimization algorithm, can automated calibrate the parameters for a non-linear inversion model. This method has been successfully applied to an IKONOS multispectral image. Bathymetric data derived from the non-linear inversion model are slightly more accurate and stable, particularly for deeper benthic habitats, than those derived from a conventional log-linear model although their overall performances are very similar. The second method is geographically adaptive inversion model. Although the general mathematical form of the geographically adaptive model is the same, model parameters are optimally determined within a geographical region or a local area, in contrast to the entire scene in the global inversion model. By using high-resolution IKONOS and moderate-resolution Landsat satellite images, I demonstrated that regionally- and locally-calibrated inversion models can effectively address spatial heterogeneity problem of water quality and bottom type, and provide significantly improved bathymetric estimates for more complex coastal waters.
Besides bathymetry information, management of aquatic habitat in streams requires knowledge of conditions and processes both inside the stream channel and in the adjacent riparian zones. To build up our monitoring and modeling capability for the stream ecosystems, I developed an automated approach to the extraction of quantitative attributes about channel geomorphology and riparian vegetation for stream habitats assessment, by integrating airborne LiDAR and aerial photography. A bottom-up segmentation method is proposed to identify the critical morphological points on a channel cross-section and partition the cross-section into meaningful segments for geomorphologic attribute calculation. Numerical algorithms have been designed to derive the channel cross-section attributes, channel longitudinal attributes, and planform attributes by integrating LiDAR and aerial photographs. Attributes about the type, vertical structure and complexity of vegetation in riparian zone have been derived by synergistically combining LiDAR data and aerial photographs. The derived cross-section geomorphologic and vegetation attributes are associated with the corresponding segments of river reach along the channel for longitudinal analysis of habitat variation. A case study is presented to illustrate the computational procedure and utility of our method. I demonstrated that my automated method can generate dense measurements on various geomorphology and vegetation attributes at user-defined intervals along the stream for better quantifying the longitudinal variability of reach conditions.