CHANGE DETECTION USING THE GRAMM-SCHMIDT TRANSFORMATION APPLIED TO MAPPING FOREST MORTALITY

Coordinate scores of appropriately rotated multitemporal TM pixel vect ors directly measure scene change. A successful transformation require s initial identification of a subset of multispectral space containing all unchanged pixel vectors, and subsequent definition of axes which measure the displacement of pixel vectors from this space. Carefully s elected pixel vectors guide the Gramm-Schmidt orthogonalization proces s to produce such a transformation. A Gramm-Schmidt analysis of a mult itemporal Thematic Mapper image of the Lake Tahoe Basin produces compo nents which correspond to identifiable scene characteristics, are easi ly interpretable, and correlate highly with field measurements of coni fer mortality.