rophytes
icquired
uisition,
etection
yy in the
jitted to
ences in
and the
nd were
je. The
ion and
les: the
; related
| are the
jn The
in the
surfaces
linearly
letration
(24 cm
] (5 cm
equency
re more
:t of the
of the
ps and
by the
rease in
on the
its low
present
nd, HH
, 1995;
°C band
ation in
rne C
ig those
iss such
among
such as
| when
(Costa,
Results from the SIR-C experiment reported in Hess et al.
(1995) suggest that the use of multi-frequency SAR data can
improve the discrimination of vegetation in the tropical
floodplain. According to the authors, backscattering statistics
indicate that both C and L band are necessary for accurate
delineation of ground classes, such as flooded forest,
macrophyte stands and non-flooded forest.
According to Rosenqvist (1995) L band backscatter was
sensitive to changes in rice growth, reaching the highest
backscatter as the plants grow to their full length of about 100
cm. There was around 8 dB difference in the rice backscatter
from the beginning of the growing season to the period of
maximum growth.
This paper describes a study performed to assess the suitability
of multiband SAR data (L band and C band ) for
discriminating among stands of macrophytes with differences
in height and density in a Amazon reservoir. In this study,
although the SAR data had been acquired at different
incidence angles ( 35? and 76" for the L and C band
respectively), it was hypothesized that by combining L and C
wavebands one could separate macrophyte stands according to
the canopy height and density.
2. THE STUDY SITE
Tucuruí reservoir was selected as test site because it has been
subjected to a series of studies since 1988 (Abdon and Meyer,
1990). Tucuruí is the first large reservoir in operation in the
Amazon region. It is located 300 km south Belém, limited by
the coordinates of 3° 43' S/ 49° 12' W and 5° 15' S/ 50°00"W in
the Tocantins river basin (Figure 1).
Figure 1 - The study site location
At the maximum height (72 m) the reservoir surface is
estimated in 2 700 km 4. The water level can reach 68 meters
in normally dry years and even 58 meters in the extremely dry
years.
The study area includes a wide variety of aquatic vegetation.
The annual rise and fall of the reservoir’s water level imposes
a constant change in the area occupied by the different aquatic
plant genus along the year. Some groups of aquatic vegetation,
however, are dominant in the reservoir: a) the free-floating
such as Eichhornia sp., Salvinia sp. and Pistia sp.; b) the
emergent such as the Typha sp.; c) the floating leafed such as
Scirpus sp.
3. REMOTE SENSING DATA
The data used in this study are the following:
193
e airborne C-band SAR wide mode image acquired during
the SAREX 92 mission in Brazil. The images were
processed in the Canadian Centre for Remote Sensing
(CCRS) being submitted to slant to ground range correction
and antenna pattern correction.
e color aerial photography at the scale of 1:10 000 taken
concurrently to the SAREX 92 mission.
e orbital L band JERS-1 data (P397/R308 and P395/R307).
e georeferenced Thematic Mapper/Landsat digital images
Table 1 presents the main features of the SAR data used in this
study.
Table 1 - SAR data used in the study
SAR Date Pixel Incidence
data Spacing | Angle
C Band | April, 14 | 15 m x| -76?
1992 6.9m
L Band | March, 7 | 12.5 m x | + 35°
1994 12.5m
In spite of the two year difference in data acquisition, both sets
were obtained during the beginning of the rising water season.
Previous studies in the area have shown that the main variable
affecting the spread of macrophytes in the reservoir is the
water level. Therefore, the differences in the backscatter from
1992 to 1995 data set can be assumed to be more related to
changes in the wavelength than in target variables. Figure 2
shows the changes in water level in April 1992 and March,
1995.
4. ANCILLARY DATA
During the SAREX 92 mission ground conditions were
documented by aerial, boat and ground survey. The following
information was collected during boat survey for several
macrophyte stands: GPS coordinates, dominant genus,
qualitative information such as stand density, homogeneity
and height of the aquatic vegetation stands. These information
helped to produce a reference map based on the color aerial
photography visual interpretation. This map was taken to the
ground in the following year and the final reference map
produced (Novo et al. , 1996)
7241
74i
nel
718 | * +
nii + +
| mei. at s
714 4- "M TÀÓ
713+
712 +
744 +
n+ ii Tédté Sabbath
[ "- A --april 92 - march 94|
Figure 2 - Reservoir water level during the SAR data
acquisition.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
