@InProceedings{MelloMSSDSBSP:2002:ExBrMa,
author = "Mello, Eliana Maria Kalil and Moreira, Jos{\'e} Carlos and
Santos, Jo{\~a}o roberto dos and Shimabukuro, Yosio Edemir and
Duarte, Valdete and Souza, Iris de Marcelhas e and Barbosa,
Claudio Clemente and Souza, Ricardo Cartaxo Modesto de and Paiva,
Jo{\~a}o Argemiro de Carvalho",
title = "Prodes digital: experiencia brasileira no mapeamento automatizado
do desflorestamento da Amaz{\^o}nia",
booktitle = "Proceedings...",
year = "2002",
organization = "Simp{\'o}sio Latino-Americano de Percepcion Remota Y Sistemas de
Information Espacial, 10; Reunion Plenaria de SELPER, 21.",
keywords = "ESTUDOS INTEGRADOS DO MEIO AMBIENTE, deforestation, Amazon, land
use, land cover, dynamics, Geographic Information Systems, linear
spectral mixing model, image segmentation, image classification.",
abstract = "O INPE - Instituto Nacional de Pesquisas Espaciais, promove,
atrav{\'e}s do projeto PRODES, a interpreta{\c{c}}{\~a}o de
imagens do Sensor TM (Thematic Mapper)do sat{\'e}lite Americano
Landsat, tendo como objetivo acompanhar a evolu{\c{c}}{\~a}o do
desflorestamento bruto da Amaz{\^o}nia Legal, inicialmente
utilizando o software SGI - Sistema Geogr{\'a}fico de
Informa{\c{c}}{\~o}es e posteriormente o SPRING - Sistema de
Processamento de Informa{\c{c}}{\~o}es Georrefer{\^e}nciadas
(INPE-DPI, 2002), desenvolvidos pelo INPE. A partir de 1994 os
pesquisadores e engenheiros da OBT- Coordenadoria de
Observa{\c{c}}{\~a}o da Terra, das Divis{\~o}es de
Sensoriamento Remoto e de Processamento de Imagens est{\~a}o
desenvolvendo uma alternativa tem{\'a}tico realizado pelo INPE,
com o intuito de introduzir automa{\c{c}}{\~a}o em algumas
etapas do projeto de Monitoramento da Floresta Amaz{\^o}nica.
Este projeto {\'e} conhecido como Prodes Digital. O presente
trabalho estabelece o procedimento aplicativo, atualmente em uso
pelo INPE, para identificar e mapear {\'a}reas desflorestadas na
Amaz{\^o}nia Brasileira, atrav{\'e}s de processamento digital de
imagens TM/Landsat. Para reduzir a dimensionalidade dos dados e
consequentemente o tempo de processamento, utiliza-se o Modelo
Linear de Mistura Espectral, implementado no Sistema de
Processamento de Informa{\c{c}}{\~o}es Georreferenciadas -
SPRING/INPE, que transforma as bandas originais TM3(0,63-0,69m m),
TM4 (0,76- 0,90m m)e TM5(1,55-1,75m m), em
imagens-fra{\c{c}}{\~a}o {"}sombra{"},
{"}vegeta{\c{c}}{\~a}o{"} e {"}solo{"}, numa abordagem que tem
permitido automatizar a discrimina{\c{c}}{\~a}o das {\'a}reas
de fisionomia florestal daquelas desflorestadas, resultante de um
processo antropog{\^e}nico din{\^a}mico que vem se
caracterizando ao longo das {\'u}ltimas d{\'e}cadas. Isso em
raz{\~a}o do aumento da fronteira agropastoril ou mesmo,
ocasionado pela explora{\c{c}}{\~a}o madeireira. A
caracter{\'{\i}}stica amig{\'a}vel na opera{\c{c}}{\~a}o de
manuseio do SPRING, o procedimento metodol{\'o}gico adotado e a
s{\'e}rie de informa{\c{c}}{\~o}es geradas sobre o
desflorestamento, de r{\'a}pida manipula{\c{c}}{\~a}o de
an{\'a}lise, trazem um ganho de qualidade {\`a} comunidade
t{\'e}cnico-cient{\'{\i}}fica que lida com as quest{\~o}es
ambientais do pa{\'{\i}}s,sobretudo naqueles estudos que modelam
as transforma{\c{c}}{\~o}es e seus efeitos em n{\'{\i}}vel
global. ABSTRACT: For many years, the Brazilian Institute for
Space Research, INPE, has been promoting the interpretation of
images provided by Landsat satellite to monitor the evolution of
the extent and rate of gross deforestation in the Brazilian
Amazon. This effort has generated results for the period from 1978
to 2001. INPE (2002)presents the study, covering the years of 2000
(estimate based on the analysis of all scenes that cover the
Brazilian Amazon)and 2001 (estimate basead on a sample), updates
the historical series on the extent and rate of gross
deforestation in the Amazon. It provides an updated insight on the
deforestation issue and allows to explore the origins of
deforestation offering indicators to guide the public policies in
the region. In addition to the extent and rate of gross
deforestation for the nine Amazon states, this study makes
available information on the distribution of the mean rate of
deforestation by large vegetation types and by classes of area
size. Due to the geometric problems, appearing during the
development of the manual interpretation of multitemporal TM
images ( image hardcopies with different scales, overlays
digitizing/scanning, complexity of deforestation pattern), the
availability of these deforestation maps in a digital format have
been restricted for some areas. During the last few years, an
effort to find a methodology to overcome this problem has been
successful. The objective of this work is to establish an
operational procedure, actualy in use at INPE, to identify and map
deforested areas in Brazilian Amazon based on digital
classification of Landsat TM images. The Linear Spectral Mixing
Model avaliable in the software package SPRING, developed at INPE,
transforms the original Landsat bands: TM3 (0,63-0,69 mm), TM4
(0,76-0,90mm)e TM5 (1,55-1,75mm)into shade vegetation and soil
fraction-images in which forest and deforested areas present a
large distinction. The Brazilian Amazon requires 229 Landsat TM
images to cover the entire region, in order to demonstrate the
operational feasibility of this approach, which is been applied to
the critical region ( {"} ARCO DO DESFLORESTAMENTO{"} ), where
governamental fiscalization actions have a highest priority.",
conference-location = "Cochabamba, Bolivia",
conference-year = "11-15 nov. 2002",
copyholder = "SID/SCD",
label = "10351",
language = "pt",
targetfile = "INPE 9425.pdf",
urlaccessdate = "21 maio 2024"
}