Multitemporal analysis of multispectral unmixing: a satellite remote sensing quantification of natural Mediterranean vegetation compositions

Environmental spatio-temporal heterogeneity characterizes major parts of Mediterranean regions, which extend across transition zones between humid and dry climates. Typical vegetation structure and patterns were evolved in these regions through the combined action of natural processes and heavy, continuous human intervention since the early days of human history. These formations and patterns are most sensitive to spatial variations in the habitat conditions and most responsive to their temporal changes. The study of these patterns and dynamics requires detailed information of vegetation distribution for wide regions. Satellite remote sensing is a unique data source for wide regional monitoring as a result of its information content and its frequent coverage. However, the magnitude of Mediterranean vegetation heterogeneity exceeds in detail the resolution of currently available Satellite data. This study is aimed at developing techniques which will allow the determination of life form compositions for each resolution cell. The methodology developed for this purpose is based on the processing of multi-temporal and multispectral Landsat TM data. A new technique of analysis seasonal variations in multi-spectral unmixing is presented