Several assumptions are made with the established atmospheric compensation algorithm for images from the SeaWiFS remote sensing platform. One of these assumptions, the existence of Case I (optically clear) ocean water, cannot be made for images of Lake Superior. A modification to the established atmospheric compensation algorithm is presented, where empirical information and external spatial data are utilized to compensate for the atmosphere in all regions of the lake. The established SeaWiFS atmospheric compensation algorithm uses a form of the Dark Object Subtraction (DOS) method. SeaWiFS has two Near-Infrared (NIR) bands used for atmospheric compensation. At these wavelengths, Case I water has no water leaving radiance. Therefore, radiance that reaches the sensor is due to atmospheric scattering alone. This NIR signal is used to determine the atmosphere type in that region of the image, which is used, in turn, to correct for the atmospheric effects in all bands. The alternative algorithm defines Lake Clear Water (LCW) as the inland analogy to Case I water. However, unlike Case I water, LCW has water leaving radiance in the SeaWiFS NER bands. Because of the oligotrophic (nutrient starved) nature of Lake Superior, it is reasonable to assume that this radiance is a constant determined by ground measurements. The atmospheric effect, then, is the difference between the expected water leaving radiance and that measured at the sensor. Like the established algorithm, this NIR signal is used to correct for the atmospheric effect in all bands in LCW regions. To implement the algorithm, an unsupervised classification method is used to map LCW and non-LCW regions in an image. Since the NIR signal in non-LCW regions is unusable, the NIR signal is extrapolated from neighboring LCW regions. This extrapolation is aided by meteorological data. Using look up tables created from the MODTRAN atmospheric model, an atmospheric type is calculated for each pixel in the image, and used for atmospheric effect subtraction in all bands. Results of this alternative atmospheric compensation algorithm were compared to optical water profile data gathered on several cruises in Lake Superior.
Library of Congress Subject Headings
Remote sensing--Data processing--Mathematical models; Superior, Lake--Remote sensing; Optical oceanography; Electrooptical devices--Atmospheric effects; Oceanography--Remote sensing
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Knobelspiesse, Kirk, "Atmospheric compensation for SeaWiFS images of Lake Superior utilizing spatial information" (2000). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus