en- 
lar 
3), 
um 
ul- 
Dp. 
ion 
uc- 
eo- 
T8. 
ta. 
ng, 
ifi- 
ind 
ANALYSIS OF THE BACKSCATTERING SIGNALS OF JERS-1 IMAGE FROM SAVANNA AND TROPICAL 
RAINFOREST BIOMASS IN BRAZILIAN AMAZONIA. 
Joäo Roberto dos Santos 
Maristela Ramalho Xaud 
Maria Silvia Pardi Lacruz 
Instituto Nacional de Pesquisas Espaciais — INPE 
Av. dos Astronautas, 1758 CP. 515 Säo José dos Campos-SP-Brasil 
Fax.: +55 12 3456449 jroberto@ltid.inpe.br 
Commission VII, Working Group 6 
KEY WORDS: tropical rainforest, savanna, biomass, regrowth, JERS-1, Amazonia, inventory. 
ABSTRACT 
The objective of this study is to show the behavior of the backscatter signals from a JERS-1 image correlated to the variability of 
biomass of the forest and savanna formations, at the contact zone of these vegetation units. This study is part of a larger project of 
characterization and monitoring of biomass distribution using remote sensing data, at the Central-North section of Roraima State 
(Northern Brazil), located between geographical coordinates N 2? 45"- N 3° 15° and W 61° 00°- W 61° 30°. The JERS-1 scene was 
processed using the ENVI software package. The backscatter values of the following vegetation classes were extracted: primary and 
secondary forest, savanna grassland, parkland savanna and savanna woodland. During the field campaign, dendrometric parameters 
(such as DBH, height) crown cover percent and botanic identification of forest formation samples were performed to elaborate 
physiognomic profiles and to estimate the biomass by allometric equations. For the areas covered by savanna types, the biomass was 
obtained by clear-cut and weighting of herbaceous, shrub and arboreal strata. Diagrams showing the changes of backscatter values, 
as related to the amplitude of biomass variation at each vegetation unities are presented. A TM/JERS-1 image composite is also 
presented to better show a product where the delineation and spatial distribution of the biomass variation of the contact zones 
forest/savanna has been made. 
INTRODUCTION yellow podzolic, associated to lateritic crusts and to 
hydromorphic dystrophic laterite. 
The continuous inventory of those contact regions of savanna 
and tropical rainforests of Brazilian Amazonia is nowadays ar: 
important task, because these areas are suffering a strong impact 
due to deforestation. The results of these studies will be a 
relevant contribution for a more accurate analysis of the global 
vegetation changes, which is an almost impossible task with 
data from optical sensors, because of that, microwave sensor 
data has to be used. The objective of this study is to show the 
relationship among spectral and textural data obtained from 
JERS-1 images and biophysical parameters (biomass values) 
from forest and savanna formations. Furthermore a 
physiognomic-structural characterization of the vegetation cover 
is made and the biomass is presented, taking into account men 
induced changes in the landscape of the area under study. 
  
  
  
  
  
Figure 1 - Section of the study area in a color composite of TM- 
Landsat and a JERS-1 image (see Colour plate 7). 
DESCRIPTION OF AREA UNDER STUDY 
The area under study has a size of 3,000 Km? and is located at 
the northern Center of Roraima State, North of Brazilian 
Amazonia, between geographical coordinates N 2° 45’- N 3° JERS-1 images of path/row 418/295 and 419/296 of May 31st 
15’ and W 61? 00"- W 61° 30° (Figure 1). The climate of this ‘96 were obtained from the NASDA’s Global Rain Forest 
DATA ACQUISITION AND METHODS 
  
region, according to Kóppen classification, is of the type Awi, 
presenting a tropical rainy dominance with a well defined dry 
season, with an average yearly rainfall of 1,800 mm. 
Geologically this area is made up of non-consolidated sandy 
sediments with clay sediment (Formaçäo Boa Vista). Recent 
alluvium, such as sands, gravel and clays, is also found. At this 
large pediplanation surface the following soil types are found: 
yellow dystrophic latosol, red-yellow dystrophic latosol, red- 
Monitoring Project (GRFM) through an agreement with INPE, 
to perform this study. A TM-Landsat image, at bands 1 to 5 and 
7, referring to path/row 232/58 from January 17th *96, at digital 
format, was also used as complementary source of information. 
Both JERS-1 scenes were processed using ENVI (The 
Environment for Visualization Images) - version 2.6, while for 
the TM image, INPE's SPRING software - version 2.0.4 (INPE, 
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 523