Please use this identifier to cite or link to this item:
Title: Linear spectral mixture model as a tool for monitoring deforestation and timber exploitation in the Brazilian Amazon
Authors: Santos, João Roberto dos
Shimabukuro, Yosio Edemir
Duarte, Valdete
Graça, Paulo Maurício Lima Alencastro de
Silva, Patrícia Guedes da
Keywords: Deforestation
Image Analysis
Land Use
Logging (forestry)
Radar Imaging
Linear Spectral Mixture Model
Tropical Rain Forest
Remote Sensing
Issue Date: 2002
Publisher: Proceedings of SPIE - The International Society for Optical Engineering
metadata.dc.relation.ispartof: Volume 4879, Pags. 320-325
Abstract: The objective of this study is to show the operational capacity of a "linear spectral mixture model" using TM/Landsat data for the characterization/monitoring of the annual deforestation and the timber logging exploitation process in the Amazon. In the methodological procedure, the original TM bands were initially converted to "vegetation", "shade" and "soil" fraction images, derived from the linear spectral mixture model. After the selection of fraction images, the scene segmentation was made using a region growing algorithm, and then an unsupervised classifier (per region) was applied. Afterwards, the thematic polygons were manually edited to generate the final maps. An analysis was made on the proportion of "vegetation", "shade" and "soil" components, for primary forest, selective logging, regrowth, and deforestation areas, for the timeframe 1997-2001. This analysis demonstrates, through the ternary diagram, that the variations in the spatial attributes of these component fractions were caused by a land cover/land use change process. A set of images and maps, showing the temporal identification of deforested and timber logging exploitation areas is shown, as a result of the operational use of this technique. The spatial distribution of these landscape changes provides subsidies to environmental agencies for the control and enforcement of specific conservation policies referring to the Amazon forest resources.
metadata.dc.identifier.doi: 10.1117/12.462387
Appears in Collections:Trabalhos Apresentados em Evento

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.