A knowledge-guided approach to automatic classification of Coastal Zone Col
or images off the West Florida Shelf is described. The approach is used to
identify red tides on the West Florida Shelf, as well as areas with high co
ncentration of dissolved organic matter such as a river plume found seasona
lly along the West Florida coast over the middle of the shelf. The Coastal
Zone Color images are initially segmented by the unsupervised Multistage Ra
ndom Sampling Fuzzy c-Means algorithm. Then, a knowledge-guided system is a
pplied to the centroid values of resultant clusters to label case I, case I
I waters, a dilute river plume ("green river"), and red tide. The domain kn
owledge base contains information on cluster distribution in feature space,
as well as spatial information such as bathymetry data. Our knowledge base
consists of a rule-guided system and an embedded neural network. From 60 i
mages, after training the system, this procedure recognizes all 15 images w
hich contained a river plume and 45 images without. The system can correctl
y classify 74% of the pixels that belong to the river plume, which provides
a substantial advantage to users looking for offshore extensions of riveri
ne influence. Red tides are also successfully identified in a time series o
f images for which ground truth confirmed the presence of a harmful bloom.