Date of Award
7-2021
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Ocean Engineering and Marine Sciences
First Advisor
Charles R. Bostater
Second Advisor
Donald Platt
Third Advisor
Ronnal Reichard
Fourth Advisor
Richard Aronson
Abstract
Glint in satellite reflectance images contaminates pixels making it difficult to ascertain what lies beneath the water surface. An algorithm was created to correct glint contaminated pixels using a nearest neighbor non-glint pixel search algorithm written in Python. The results of the glint corrected pixels were then analyzed using an algorithm to estimate water visibility using light attenuation depths and second derivative estimators based upon glint corrected satellite reflectance using channels from a WorldView-3 (WV-3) image. The above was accomplished for sand, mud, and seagrass bottom types. The results suggest that estimated attenuation depths (a measure of water visibility) is feasible. To estimate the attenuation coefficient using a second derivative estimator, calculation of any three of the eight multispectral bands can be used depending on what band the user wants to enhance. The WorldView-3 satellite and the DLR Earth Sensing Imaging Spectrometer (DESIS) hyperspectral sensor located on the International Space Station (ISS) was also evaluated to determine if these space-based sensors would simultaneously overpass a water area for applying image fusion and multi sensor water glint corrections. Due to the variable orbit of the ISS and locations near the Florida Space Coast, no near simultaneous over passes occurred for use during this research. Two areas were found in the Atlantic Ocean coastal waters in the Bahamas, and these overpass locations and dates are presented.
Recommended Citation
Taggart, Andrea, "A Glint Contaminated Pixel Correction Algorithm and Potential Application for Estimating Water Visibility" (2021). Theses and Dissertations. 1219.
https://repository.fit.edu/etd/1219