that presents also sections of bare soil. Those areas with
individuals showing a larger distribution of diameter and height
and that show two to three strata, present higher backscatter at
RADARSAT scenes. The roughness from the canopy of these
more developed formations, where the components “shade” and
“vegetation” are included in the resolution element (pixel), are
very helpful to identify these vegetation types (Bernardes, 1996;
Santos et al., 1998).
As a complementary information on the behavior of landscape
classes in RADARSAT images, Kux et al., (1998) reports that
primary forest areas in Acre show backscatter variations in the
range of c? = -8 to -6 dB, while areas of initial regrowth vary
between -10 to -8, showing the separability between these two
vegetation classes. On the other hand, those areas of fresh and
overgrown pasture present values close to -9 and -7 dB, and as
such they show a similar spectral behavior as those initial
regrowth areas.
As for the sampled areas discussed in this paper, those areas of
fresh and overgrown pasture show a slightly higher backscatter
amplitude at the October 25th datatake (y — 0.18 to 0.23), while
values decrease from 0.13 to 0.17 at the May 15th datatake,
probably due to the increase of the influence of the soil, which
is lower when the ground surface is dry.
When considering a qualitative analysis of these RADARSAT
scenes, it is to say that at this wavelength, the discrimination of
vegetation types is limited to vegetation types with high
differentiation of canopy closure, and also to bare soils and
burned areas. At the SAR scene from the dry season under study
the soil surface is dry enough to provide differentiation between
vegetation types with low aboveground biomass and clear-cut
areas.
5. CONCLUSIONS
Through our experience with C-band SAR data analysis we
learned that vegetation cover classes present most significant
backscatter values during the dry season. The main motivation
for this study was to analyze RADARSAT images from May
and October'96, to investigate the possibilities of correlation
between backscatter and biomass values. Generally speaking,
isolated RADARSAT data does not show high sensitiveness to
biomass. Nevertheless, this sensor system can be used to
discriminate among forest and non-forest areas, being useful to
monitor changes in the land use.
An important contribution of this sensor is its capacity to
discriminate neatly between primary forest (higher canopy
roughness) and initial succession (lower roughness). However
C-band data are not suitable for monitoring biomass in the
whole regeneration processes of tropical forests because, when
biomass is at the development stage of intermediate regrowth,
there is a superposition of spectral/textural attributes
(backscatter) occupied by both advanced regrowth and primary
forest. Future studies should concentrate efforts on the use of
combined temporal RADARSAT and JERS-1 data, to integrate
specific information of the canopy components with the vertical
structure with information of aboveground biomass distribution
of Amazonia.
ACKNOWLEDGEMENTS
The authors are grateful to the Conselho Nacional de
Desenvolvimento Científico e Tecnológico-CNPq (process
number 300677/91-0 and 381246/97-3), the Fundaçäo de
Amparo à Pesquisa do Estado de Säo Paulo-FAPESP (process
number 97/05475-2), the Canadian Space Agency under the
Applications Development Research Opportunity (ADRO), the
Canadian International Development Agency-CIDA and
Universidade Federal do Acre-UFAC/ Parque Zoobotânico.
REFERENCES
Bernardes, S., 1996. Índices de vegetagáo e valores de
proporçäo na caracterizaçäo de floresta tropical primária e
estádios sucessionais na área de influéncia da Floresta Nacional
do Tapajés-Estado do Parä. Dissertaçäo de Mestrado em
Sensoriamento Remoto, Instituto Nacional de Pesquisas
Espaciais-INPE, Säo José dos Campos. 105 p.
BRASIL, 1990. Projeto de proteçäo ao meio ambiente e das
comunidades indígenas: diagnóstico geoambiental e sócio-
económico. Área de influéncia da BR-364 trecho Porto Velho-
Rio Branco. Secretaria de Planejamento e Coordenagáo da
Presidéncia da Repüblica, IBGE/IPES, Rio de Janeiro. 132p.
Brown, S.; Gillespie, A.J.R.; Lugo, A.E., 1989. Biomass
estimation methods for tropical forest with applications to forest
inventory data. Forest Science, 35(4), pp. 881-902.
Dobson, M.C.; Ulaby, F.T.; Le Toan T.; Kasischke, E.S.;
Christensen, N., 1992. Dependence of radar backscatter on
coniferous forest biomass. IEEE Transactions on Geoscience
and Remote Sensing, 30(2), pp. 412-415.
Imhoff, M.L., 1995. Radar backscatter and biomass saturation:
ramifications for global biomass inventory. IEEE Transaction
on Geoscience and Remote Sensing, 33(2), pp. 511-518.
INPE/IBAMA - Instituto Nacional de Pesquisas Espaciais
INPE/Instituto Brasileiro do Meio Ambiente e dos Recursos
Naturais Renováveis - IBAMA, 1998. Deforestation 1995 -
1997: Amazónia. Sáo José dos Campos.
<http://www.inpe.br/amz.htm>.
Keil, M.; Scales, D.; Winter, R.; Kux, H.J.H.; Santos, J.R,,
1996. Tropical rainforest investigation in Brazil using
multitemporal ERS-1 SAR data. In: Second ERS-1 Application
Workshop, London, UK, pp. 241-248. (ESA SP-383).
Kux, H.J.H.; Santos, J.R.; Ahern, F.J.; Piestch, R.W.; Pardi
Lacruz, M.S., 1998. Evaluation of Radarsat data for
landuse/land cover dynamics in southwestern Brazilian Amazon
State of Acre. Paper submitted and accepted on the Canadian
Journal of Remote Sensing, (in press).
Luckman, A; Baker, J.; Kuplich, M.T.; Yanasse, C.C.F.; Frery,
A.C, 1997. A sutdy of the relationship between radar
backscatter and regeneration of tropical forest biomass for
spaceborne SAR instruments. Remote Sensing of Environment,
60, pp. 1-13.
530 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
Santo:
the be
tropic
Intern
Monit
Comn
Schmi
1997.
chang
Intern
Latin .
Uhl, (
pastur
Journz
Yanas
radar i
detect
