The input of fresh water is very limited, since no strong
river reaches the system. Tides, which present a
maximum amplitude of 1,20 m, are the main agents for
estuarine circulation and for the mixture between
chemical-physical properties. By the time of this
experiment, a very small drainage basin, with an area
of 1.339 Km?, was responsible for most of the fresh
water input. Tidal creeks also represent an important
source for fresh water (collected basically from rain),
and may present up to 8 Km of extension and
maximum depths of 7,0 m.
These small water bodies were developed over
Pleistocenic terraces, which are covered by tropical rain
forests. Expressive mangroves have established in their
margins, contributing for the high productivity of the
area. The decomposition of leaves from these types of
vegetation regularly promotes the "brownishment" of
the water color, mainly after huge storm rains, which is
related to the introduction of humic and fulvic acids in
the system. These dissolved organic compounds are
optically referred as "yellow substances".
The particulate inorganic sediments in the water
column predominantly come from Pleistocenic sandy
terrace's erosion and from bottom sediment's
resuspension, since river input is inexpressive. The
highest concentrations occur near to the bottom,
associated to maximum current speeds, both on ebb
and flood tides. Bottom currents always present
threshold velocities high enough to erode and
transport fine and very fine sands, which predominate
in the bottom (Bonetti Filho et al., 1995). The water
transparency in the area is rarely higher than 2,0 m
and total suspended solids present an average
concentration between 30 mg/l and 40 mg/l.
2. METHODOLOGY
Although ideal conditions for water quality monitoring
from orbital platforms occur when simultaneous in situ
data are available, many authors agree that a previous
knowledge of the study site's oceanography may lead to
a good qualitative approach (Robinson, 1984).
During this experiment, the extensive cloud coverage
of the area did not permit the acquisition of real time
data, and demanded the use of previously acquired
LANDSAT-5/TM images for the investigation.
Despite this, oceanographic data were collected aboard
research vessels during two field trips, on August 1993
and February 1994.
In these occasions a series of water quality parameters
was collected, following transects along the estuarine
front and in 25 hour anchor stations inside the system,
to help in the interpretation of the front's
78
oceanographic characteristics. Suspended sediment
concentration (fractionated on its organic and
inorganic compounds), currents (speed and direction),
salinity, temperature and pH were sampled and gave a
very good support to the interpretation of orbital data.
The time series was composed by three continued years
of images, generated in the following dates: April 17,
1984; May 22, 1985; and September 14, 1986. Their
WRS reference was 220.77. For the three images,
meteorological and tidal information were analyzed in
detail. LANDSAT-5 bands TM-1, TM-2 and TM-3
were used for the study of water surface compounds,
band TM-4 for water/land separation and band TM-6
for brightness temperature determination.
The images were co-registered and processed in full
resolution (pixels with 30 x 30 m for reflected bands
and 120 x 120 m for thermal band) and the working
scale, in the system's monitor, was 1:85.333. The
Brazilian SITIM-340 image processing system was used
for digital treatment, which was composed by a number
of steps previously described by Bonetti Filho et al.
(1994). Basically, they were:
a) Histogram's Analysis: conducted to characterize
digital numbers distribution and to determine the
values for atmospheric correction;
b) Atmospheric Correction: applied to minimize the
atmospheric contribution to the scene, following the
method of the darkest pixel subtraction (Chavez, 1975);
c) Digital Numbers Surveying and Densitometric
Transects: determined by direct digital reading along
the estuarine front, perpendicular to the coastline, to
help in the quantification of the differences detected in
the time series;
d) Water/Land Separation: applied from the
generation of a mask, derived from water classification
on TM-4, used to maximize the performance of the
posterior contrast enhancement;
e) Linear Contrast Enhancement: the function was
determined from histogram's analysis, individually for
each TM band, in order to increase visual
discrimination;
f) Pseudo-Color: generated by the equalization of each
band into 14 classes (5 classes for TM-6, due to its
lower spectral dynamics) and the latter application of a
mnemonic color palette;
g Color Composition: obtained from the use of the
enhanced bands TM-3, TM-2 and TM-1 (RGB),
without land information.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
O m Ey oe
< nm Z4 t u N -J]
e]
Pg x^ € VL ] rt m e-