felt,
oke
9].
ints
o of
000
the
ver
| in
not
eed
ion
of
alts
for
1dy
the
the
ke
ted
the
of
Novo, Evlyn
WFI/CBERS imagery simulation for understanding water pathways from Amazon River to the floodplain
Evlyn Novo*, Yosio Shimabukuro*, Maycira Costa **
Instituto Nacional de Pesquisas Espaciais, Brasil
Divisäo de Sensoriamento Remoto
Evlyn@ltid.inpe.br
Yosio@ltid.inpe.br
**University of Victoria, Canada
maycira@office.geog.uvic.ca
KEY WORDS: WFI camera, floodplain monitoring
ABSTRACT
This paper describes an experiment performed to simulate WFI/Camera spatial features using a TM-Landsat digital
mosaic re-sampled to 258 m x 258 m, 100 m x 100 m resolution. The re-sampled mosaic and selected sub-scenes from
the original TM data were submitted to digital processing in order to map different water types found in the Amazon
River floodplain. The mosaics were classified and classification results for selected reaches were compared. The results
support that in spite of the degraded spatial resolution WFI images offer a good deal of useful information on water
types.
INTRODUCTION
The fundamental problem in understanding the spatial and temporal distribution of water types in large floodplains is its
non-uniformity in space and time. In a river basin such as the Amazon River, this problem is even bigger. Amazonian
rivers and lakes vary seasonally in water quality and sediment load as a function of the geochemistry and hydrology of
their catchment basins (Junk and Furch, 1985). The discharges of Amazon River tributaries are variable and out of
phase with the main stem (Richey et al., 1986). Moreover, in rivers such as Rio Negro (a Black Water river) and
Madeira (a White Water river) the discharge can vary for about a factor of 10 from the low to the high water season.
This large annual amplitude in the Amazon River and its tributaries results in a variable pattern of periodic flooding of
Amazonian lowlands. The nature and origin of the water in the floodplain is also dependent on other sources such as
direct precipitation and water from local tributaries. The proportion of those different water types, however, is not
known because of the variable transfer of water, sediment and materials from rivers to floodplains and from floodplains
to rivers. In spite of the huge research efforts for collecting ground information (Fisher and Parsley, 1986, Richey et al.,
1986), there is still a need for better understanding of those exchanges.
Remote sensing data have been used to study the river dynamics and the biological diversity of the lowland forests
(Salo et al., 1986) , suspended sediments in surface water (Mertes et al., 1993), and the zones influenced by river water
(Mertes et al., 1995) in the Amazon River basin. The various water types have distinctive spectral reflectance making it
possible to distinguish among them and even to quantify their optically active components (Kirk, 1994). The synoptic
view provided by satellite images make them an useful tool for monitoring the influx and dispersion of river water into
the floodplain lakes. In spite of the potential of optical remote sensing data for monitoring water quality variables, cloud
cover limits the effective use of the technology (Clark, 1983). It took a 10-year period to assemble 29 almost cloud-free
TM-Landsat images to build the Amazon River floodplain mosaic (Shimabukuro et al., 1998). The majority of cloud-
free images, however, were restricted to the high-water season, preventing the assessment of seasonal changes in the
distribution of water types in the floodplain. Limnological and hydrological studies require a much more frequent
imaging than the rate provided by present land research satellites such as Landsat and SPOT.
On October 14, 1999 the China-Brazil Earth Resources Satellite (CBERS) was successfully launched carrying on board
a wide field of view imager (WFI). WFI (table 1) is a CCD pushbroom camera operating in two bands with the potential
of acquiring low resolution but wide swath (890 km) images in about every five days. The satellite orbital pattern
provided by CBERS allows for a complete coverage of the Amazonia in less than a week and increases the possibility
of acquiring data without major natural changes in water type distribution. The high imaging frequency also increases
the probability of acquiring cloud-free images. This high frequency of data acquisition makes this camera potentially
useful for studying the spatial distribution of water in the Amazon River basin floodplains.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1027