Use of self-organizing maps in the identification of different groups of reclamation sites in the Amazon forest-Brazil

Abstract

Brazil has the third largest reserve of contained tin, around 12.3% of the amount produced in world, being a large part of these reserves located in the Amazon region. As a result of this mineral wealth, the Amazonian ecosystem has been suffering a rapid process of environmental degradation since the sixties. In this sense, given the mining adverse consequences to the environment, Brazilian Constitution obligate the land reclamation of degraded areas by mining and it has been performed by the majority of the mining companies. However, given the environment complexity and its relationship with the biological diversity, there is a great necessity of better understanding in assessment of evolution of these degraded areas in recovery. Thus, the present work had as objective identifying different groups of degraded areas in reclamation process by means of soil texture, biochemistry and vegetation indicators. The data was analyzed through Artificial Neural Networks (ANN) Self Organizing Maps (SOMs). The results showed four different groups and it was identified a relationship between the different textures soils as a result to the recovery method applied.