Change detection in the semi-arid landscape using radiometric rotation applied to remote sensing data

The main task is to apply in the semi arid landscape of Maule region (Chile), the multi-temporal change detection algorithm ¿RCNA multi-spectral¿, using TM/Landsat images (2004 and 2008). The RCNA is based in angular parameters; those angles are calculated in function of the axis formed by the straight line of regression of those points labeled in the field survey as no-change. The image for detection was transformed from a continuous (floating-point) to thematic image, through a slicing and labeling process. Hence it is possible to discriminate: two classes related to degradation, two referring to regeneration and one of no-change. In the timeframe studied 4.6% of area under study presents degradation patterns, derived from the complete removal of the matorral type, followed by changes of land use. We found out that the application of multi-spectral RCNA technique in semi-arid landscape is sufficiently robust and the complex radiometric correction is not necessary.