作者: Osmar Abílio de Carvalho Júnior , Potira Meirelles Hermuche , Renato Fontes Guimarães
DOI: 10.1590/S0102-261X2006000300002
关键词:
摘要: Parana river basin has one of the major fragments Decidual Seasonal Forest in Brazil. This vegetation is widely fragmented due to selective wood exploitation and growth pasture areas, what justifies development studies order understand its dynamics preserve diversity. Thus, present study aimed at defining a method for regional identification Deciduous basin. The deciduous forest typical phenological cycle comparison with other savanna physiognomies. Due these characteristics, temporal series normalized difference index (NDVI) images MODIS sensor was used detection. adopted methodology may be subdivided into following steps: (a) elaboration 3D cube NDVI images, where z profile corresponding signature or spectrum, (b) noise elimination using Minimum Noise Fraction (MNF) transformation, (c) variability examination vegetation, establishment best band applied differencing method. presents spectral behaviour, higher values raining season lower dry season, makes this kind different from others. employment changing detection algorithm between two images: enhances localization Forest. So, proved effective delimitation Forests considering sensor. Considering method, region characterized by presenting alteration values.