Software tool for tracking ocean color anomalies

We describe a software framework to integrate image processing algorithms and an expert system shell for automatic classification of ocean color anomalies on satellite images. Based on this framework, a prototype knowledgeguided system is developed to track harmful algal blooms and dark water events. In developing the prototype system, a fuzzy clustering algorithm is first applied to a set of SeaWiFS images taken over the coastal ocean off Florida's west coast. Heuristic rules are extracted from the clustered images to create a knowledge base. Later, the system is applied to a set of new images and identifies ocean color anomalies under the guidance of the knowledge base.