Many recreational, military, and commercial activities take place in shallow coastal waters; therefore, interest is high in characterizing these areas. A variety of methods have been employed to determine water depths and classify the bottom using remote sensing. This research proposes to apply Philpot's principal components algorithm for bathymetric mapping to a MISI hyperspectral image, whereas previously this approach has been used on synthetic data. A description of the principal components algorithm is presented along with an outline of how it was applied to airborne hyperspectral images. The algorithm takes advantage of the ability to implement a deep-water correction, and in this linearized space, perform an eigenvector analysis to determine maximum variance in the data, which is related to depth. Unsupervised classification was performed on the first two principal component scores, resulting in a qualitative depth map and bottom type map.
An extensive water measurement campaign was conducted in Lake Ontario in order to characterize the optical properties of the water at the time the MISI images were taken. These properties were used as inputs to the HydroMod radiative transfer model in order to generate sensor-reaching radiance values for various depths and over different bottom types characteristic of a test site on the central New York shore of Lake Ontario. A principal components regression was performed using the algorithm-processed HydroMod model radiances and image data in an effort to determine the inputs to the image, i.e. depth and bottom type, without having a priori information. The limitations of the algorithm as well as the regression approach are discussed.
Library of Congress Subject Headings
Underwater imaging systems; Water levels--Ontario, Lake (N.Y. and Ont.); Lake sediments--Ontario, Lake (N.Y. and Ont.); Sedimentation and deposition--Ontario, Lake (N.Y. and Ont.); Ontario, Lake (N.Y. and Ont.)
Imaging Science (MS)
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
John R. Schott
Rolando V. Raqueño
Wilson, Nikole L., "Hyperspectral imaging for bottom type classification and water depth determination" (2000). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus