Landcover degradation analysis of Mediterranean forest by means of hyperplanes obtained from mixture linear algorithms (MLA)

Abstract

The percentage alteration of the Mediterranean forest landscape is one of the primary indicators for its degradation. In this sense, the land cover abundances change analysis by using mixture linear algorithms (MLA), is presented like a good alternative to study this degradation. This research analyzes the use of two information sources like Remote Sensing (Landsat-ETM+) and Field Radiometry (GER 1500) to obtain mixture hyperplanes. These are calculated by models based on least square estimations, assuming that each pure land cover (endmember) belonging to any geographic area, behaves as a random variable which distribution function is known. The mixture hyperplanes provide spectral signatures with a suitable correlation level with regard to the supplied from remote satellite sensors once corrected, for the same geographical zone. These established hyperplanes can be used in future researches about Mediterranean forest landscape changes, because they can represent the different levels of its degradation. In this sense, it is proposed that they will feed a land cover spectral library with free accessibility.