Secondary forest age mapping in Amazonia using multi-temporal Landsat/TM imagery

Satellite imagery can be a powerful tool for identification of deforested areas, often providing reliable classification of the Earth surface. The aim of this paper is to identify and to characterize the age of tropical secondary forests in a test site near Tapajos National Forest, located on the south of Santarem city, Para State, Brazil, using a time series of Landsat/TM images. The Landsat/TM images from 1986 to 1992 were co-registered, segmented and classified into several classes of land cover. These thematic images were superposed in a GIS and, using logical (Boolean) operations, a stages of regeneration map was generated. The stages of regeneration classes were defined as areas with recent activities, areas with different stages of secondary vegetation (varying from one to six years of regeneration), old secondary forest (more than six years old), and primary forest. It was observed that there is a functional relationship between the stages of the classified secondary vegetation and the normalized difference vegetation index (NDVI) using the bands 3 and 4 of the Landsat/TM images. It was concluded that the use of time series of optical images can be a valuable tool for determining the evolution of secondary tropical forests, providing that image acquisition is possible at least once a year.