INTERACTIVE ANALYSIS OF POLARIMETRIC SIR-C AND LANDSAT-TM DATA FOR THE 
SPECTRAL AND TEXTURAL CHARACTERIZATION OF THE LAND COVER IN SW AMAZONIA, 
BRAZIL 
Joäo Roberto dos Santos 
1 
Hermann Johann Heinrich Kux ‘ 
Manfred Keil ^ 
María Silvia Pardi Lacruz 
Dominic R. Scales 
2 
Instituto Nacional de Pesquisas Espaciais - INPE, Sáo José dos Campos, Brazil 
jroberto@ltid.inpe.br 
? German Aerospace Research Establishment (DLR), Oberpfaffenhofen, Germany 
keil(g)dfd.dlr.de 
Commission VII, Working Group 2 
KEYWORDS: Land use, Forestry, SAR, Regrowth, Amazonia 
ABSTRACT 
The objective of this study, is to analyze SIR-C, L-band data at several polarizations, combined with TM-Landsat data, 
to chracterize land use/land cover features in SW Amazonia (Acre State, Brazil). Segmentation techniques of TM- 
Landsat and textural classifiers for SIR-C were used for the identification of six land use/land cover classes. The 
spectral and textural characteristics of forest, regrowth and pasture types are briefly discussed. This interactive analysis 
of the optical and microwave sensor data will certainly contribute to monitor the dynamics of land use in Amazonia. 
1. INTRODUCTION 
In the frame of studies related to global change, there is a 
need to monitor tropical forests, not only to estimate the 
yearly deforestation rate, but also to follow the dynamics 
of land occupation by man. The frequent cloud coverage 
in certain tropical areas made data collection with optical 
spaceborne sensors an impossible task. Imaging radar 
offers the potential to map vegetation and land use 
classes in such areas, because its’ independence of 
weather conditions. 
A cooperation program between the National Brazilian 
Institute of Space Research (INPE) and the German 
Aerospace Research Establishment (DLR) is under 
development, with the objective to evaluate the 
capabilities of Remote Sensing techniques to control the 
environmental impact of deforestation in Amazonia and 
to map rainforest formations and land use (Keil et al., 
1995). As a result of this cooperation, the main objective 
of the present study is to evaluate the spectral and 
textural capabilities of SIR-C (Shuttle Imaging Radar) 
data at different polarizations (HH, VV, HV, LL and 
Total Power), specially at L-band, to characterize the 
changes of land cover in a section of West Brazilian 
Amazonia, in Acre State. An interactive analysis of 
Landsat-TM data, using the six optical bands, was also 
performed to identify the main land cover units along the 
Rio Branco-Sena Madureira road (Acre State). 
209 
2. AREA UNDER STUDY 
The area under study includes a section along the road 
Rio Branco - Sena Madureira (BR-364). This area is 
covered by different tropical vegetation formations and 
disturbed areas like pastureland and natural secondary 
succession. Since the opening of the BR-364 in the early 
70's, large deforestation activities took place, mainly for 
cattle raising activities, but also for colonization projects. 
In the last few years, since the suspension of 
governmental incentives, the speed of forest clearance 
decreased. According to Santos et al. (1994) the annual 
rate of gross deforestation in Acre State in the timeframe 
1991 to 1992 was of about 327 km”, and it is believed 
that this value is valid until today, due to the strong 
environmentalist pressure. The plantation of pasture, the 
improvements obtained in burning and logging, as well 
as the vegetal successions of deforested areas, strongly 
influence the equilibrium of the ecosystem in SW 
Amazonia. 
3. MATERIALS AND METHODS 
The main database of reference were Landsat TM 
images, bands 1-5 and 7, from July 1994. A SIR-C 
dataset of the April ‘94 Mission was used. The SPRING 
software package, developed at INPE, as well as EBIS, 
from DLR, were used. In July 1994 a field survey was 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996