COASTAL WATER CHLOROPHYLL ESTIMATION USING LANDSAT TM
Maycira Pereira de Farias Costa
Instituto Nacional de Pesquisas Espaciais
Caixa Postal 515 - CEP 12201
Säo Jose dos Campos, SP - Brazil
ABSTRACT
This work was performed in Ubatuba, Säo Paulo coast, to assess Landsat 5/TM
information of chlorophyll concentration (Chla) studies in the ocean. Twenty
four water sampling stations were determined and the following parameters
were sampled: chlorophyll a, yellow substance, total suspended solids and
Secchi depth. TM digital image was prev
iously corrected and an average of
nine pixels values of each coordinate point was obtained from TM band 1,2
and 3. A linear correlation analysis between water parameters and
it was observed the correlation
reflectance data was
applied and
coefficients between Chla and TM],
respectively:0.84, 0.92, 0.86, 0.91 and 0.71. A model to estimate Chla from
TM bands was determined by usin
resulting model included TM2 band (R
a
TM2, TM3, TM1/TM2 and TM3/TM2,
stepwise multiple regression. The
adjusted=0.84).
KEY WORDS: Remote Sensing, Ocean water chlorophyll.
1.0-INTRODUCTION
The phytoplankton pigments (chlorophyll
a) synthetize organic matter from
inorganic matter by using the solar
energy. It is responsible for around 95%
of marine photosynthesis, being the main
primary productor of the ocean and
regulating the CO, levels in the
atmosphere (Perry, 1986).
The chlorophyll pigment can be used as an
indicator of primary production level,
physical oceanographic phenomena, etc
(Tyler and Stumpf, 1989).
The conventional sampling methods for
determining chlorophyll concentration are
expensive, time consuming. These aspects
explain the poor spatial distribution of
the resulting data sets and prevent their
interpolation and extrapolation. Remote
sensing data can minimize those time and
spatial sampling problems by providing a
synoptic view of the area under study
(Perry, 1986; Platt and Sathyendranath,
1988).
The remote sensor signal results from the
interection between the solar radiation
and both water and atmosphere (Kirk,
1986).
Pure water presents high absorption in
the red and infrared region of the
eletromagnetic spectrum. The water
spectral response is changed by their
optically active components such as
pigments, organic and inorganic matter
and organic dissolved substances.
Suspended inorganic matter are the main
light scatters within the aquatic
230
environment. Size, shape and
concentration are the main factors
explaining the amount of scattering by
inorganic matter( Novo et al., 1989).
Yellow substances are mainly absorbed in
the short wavelenghts. At high
concentrations they cause a decrease in
chlorophyll model sensitivity (Tassan,
1988).
Each phytoplankton pigment presents its
typical absorption curve. The pigment
composition varies according to the
phytoplankton species. Chlorophyll a is
the main pigment and absorbs at 435nm and
670-680 nm. The chlorophyll concentration
in the water can be detected through
remote sensing techniques since changes
in its spectral absorption and scattering
coeficients affect water color.
Since the early fifties, water color has
been used as an indicator of water
components such as chlorophyll
concentration, inorganic matter and
yellow substance. In 1972, with the
Landsat program, there was an increase in
the use of remote sensing data for
estimating chlorophyll concentration
(Sturm, 1980). The poor spectral and
spatial MSS/Landsat resolution prevented
the operational use of this sensor system
for primary production assessment. With
the advent of CZCS featured to ocean
color monitoring, there was an increase
in chlorophyll algorithm development
(Gordon et al. 1983). With: the CZCS
discontinuity in 1986, TM/Landsat data
started to be considered for chlorophyll
model development (Lathrop and Lillesand,
1986; Tassan, 1987; Grunwald et al. 1986;
Braga, 1988; Novo and Braga, 1991,
Khorran et al. 1991; Ekstrand, 1991).
PM/I
chlc
betv
inoi
and
Khor
In |
mar:
sti;
coni
Gual
Con:
the
that
app:
encc
obj:
ass:
est:
conc
env:
The
Nort
soul
Thi:
bec:
inte
Unix
198:
cle:
coni
als
The
maii
of '
wate
in
proc
2.1.
Wate
in
of
loc:
to -
isl:
naui
4000000" 0
Fig
site