Monitoring burnt areas by principal components analysis of multi-temporal TM data

Abstract

The principal components transformation is used to highlight areas that show a change in time, normally associated with fire damage and vegetation regrowth. These variations are enhanced in the high order components. The individual scenes, as well as the components, are classified by unsupervised techniques to yield thematic maps on which dynamic cover type classes are described. These maps are later compared with the resulting images of arithmetic operations, such as subtraction and ratioing.