REMOTE SENSING ANALYSIS OF BELMIRA'S PARAMO VEGEATATION WITH LANDSAT IMAGERY

The purpose of this study is to distinguish the forest of Belmira's Paramo from other land cover classes. Three LANDSAT images are available (1996, 2002 and 2003). Remote sensing analysis of the vegetation coverage includes image correction and classification and validation process. The COS(t) model and the quadratic interpolation function were used for image correction. The iterative self-organizing cluster analysis is considered for image non supervised classification and the maximum likelihood classifier is taken into account for image supervised classification. 70 GPS land observations and the error matrix analysis, were used for validation process. The Result is a map for each image, with two land cover categories: forest & non-forest. Classification error is 2% and map-land observations correspondence is 80%. However, the presence of clouds and shadows affect the remote sensing accuracy.