SS3.03 Recent Advances in Coastal and River Plume Remote Sensing
Date: Thursday, June 13, 2002
Location: Poster Session - VCC
 
ZhangM, Computer Science Department, Winona State University, Winona, USA, MZhang@winona.msus.edu
Hu, C, , College of Marine Science, University of South Florida, Saint Petersburg, USA, hu@seas.marine.usf.edu
Kirkpatrick, G, J, Mote Marine Laboratory, Sarasota, USA, gkirkpat@mote.org
Muller-Karger, F, E, College of Marine Science, University of South Florida, Saint Petersburg, USA, carib@seas.marine.usf.edu
 
TRACKING HARMFUL ALGAL BLOOMS ON SEAWIFS IMAGERY USING A FUZZY CLUSTERING ALGORITHM
image
An approach using an unsupervised fuzzy c-means clustering algorithm to process SeaWiFS images for the identification of recent outbreaks of Harmful Algal Blooms (HABs) is described. Rayleigh and a small aerosol component were first removed from the SeaWiFS imagery, and then the clustering algorithm was applied. The fuzzy clustering algorithm was used as an image segmentation tool in this approach. Cluster labeling rules were extracted from a time series of SeaWiFS images and were used to label clusters belonging to HABs. The HABs were tracked on a sequence of SeaWiFS images acquired in late 2001. Results are validated using field data collected on site.