International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
QUALITATIVE ASSESSMENT OF INLAND AND COASTAL WATERS BY USING
REMOTELY SENSED DATA
Asif M. Bhattia, *, Donald Rundquistb, John Schallesc, Mark Steeled, Masataka Takagie
a,e
Department of Infrastructure Systems Engg., Kochi University of Technology, Kami-City, Kochi, 782-8502, Japan.
b,d
Center for Advanced Land Management Information Technologies (CALMIT), School of Natural Resources,
University of Nebraska-Lincoln, Lincoln, USA.
c
Biology Department, Creighton University, Omaha, NE, USA.
*asifmumtaz.bhatti@kochi-tech.ac.jp
Commission VIII, WG VIII/4
Key words: Total suspended sediments, Chlorophyll-a, Secchi depth, Remotely sensed data
Abstract: The prime purpose of the research study was to elucidate the potential of remotely sensed data for estimation of water
quality parameters (WQPs) in inland and coastal waters. The useful application of remotely sensed data for operational monitoring of
water bodies demand for improved algorithms and methodology. The in situ hyperspectral Spectroradiometer data, water quality data
and Airborne Imaging Spectroradiometer for Applications (AISA) data of Apalachicola Bay Florida, USA were collected. The data
was analyzed to develop the models for assessment of total suspended sediment (TSS), chlorophyll-a (chl-a), and secchi depth. The
analysis of collected spectral data reveals that a peak reflectance in red domain was well correlated with chlorophyll-a concentration.
The optical depth is found to be strongly correlated with Chl-a and TSS. In order to examine the feasibility of multispectral data for
water quality monitoring; AISA data was integrated into band widths of ALOS/AVNIR-2 sensor. The combination of three bands,
band 2, 3 and band 4 was developed to correlate the remotely sensed data with TSS. The developed regression models showed good
correlation with water quality parameters and may successfully applied for estimation of WQP in surface waters. The research work
demonstrates an example for the successful application of remotely sensed data for monitoring the distribution of water quality
parameters in water bodies.
From the perspective of remote sensing, waters can generally be
divided into two classes: case-I and case-II waters (Morel,
1977). Case-I waters are those dominated by phytoplankton
(e.g. open oceans) whereas case-II waters containing not only
phytoplankton, but also suspended sediments, dissolved organic
matter, and anthropogenic substances for example some coastal
and inland waters (Gin, 2003). Remote sensing in case II waters
has been far less successful. Many scientists have pointed out
that this is mainly due to the complex interactions of four
optically active substances in case-II waters: phytoplankton
(chl-a), suspended sediments, coloured dissolved organic matter
(CDOM), and water (Novo et al., 1989; Quibell, 1991; Lodhi et
al., 1997; Doxaran et al., 2002). The spectral characteristics of
different water bodies and at different sampling points of the
same water body are not same. Optically active components in
the water bodies influence is qualitative and quantitative nature
of the spectral signatures. The main components responsible for
change in spectral signatures are yellow substance,
phytoplankton pigments, and non living suspended matters and
water itself. Remote sensing of water-constituent concentrations
is based on the relationship between the remote-sensing
reflectance, and the inherent optical properties, namely, the total
absorption and the backscattering coefficients (e.g., Gordon et
al., 1988). Chl-a concentration and total suspended solids (TSS)
are two important water quality variables influencing the
qualitative and quantitative nature of the spectral signatures.
The objective of present research is to develop relationships
between water quality parameters (WQPs) and remotely sensed
data (RSD) and to elucidate the spatial and temporal variation in
chlorophyll-a concentration and TSS in Apalachicola Bay,
Florida. The potential of simulated multispectral remote sensing
data for delineation of WQPs was comprehensively examined.
1. INTRODUCTION
Assessment of water quality parameters in water bodies is one
of the most scientifically relevant and commonly used
application of remote sensing. Water quality monitoring
requires regular and relevant observations which cannot be
obtained by conventional field monitoring campaigns.
Remotely sensed data with high spatial resolution and frequent
acquisition frequency offer solution to monitor variability of
water quality parameters up to several times per year.
Application of remotely sensed data allows to discriminate
between water quality parameters and to develop a better
understanding of light, water and substances interactions.
Hyperspectral remote sensing allows accurate and potential use
of entire range of electromagnetic spectrum recorded in
extremely narrow wavebands for monitoring water quality on
multiple sites in water bodies. The operational monitoring and
useful application of remote sensing in water bodies demands
for improved methodology and sophisticated algorithms. Most
of the satellite sensors record data in only a few broad spectral
bands, the resolutions of which are too coarse to detect much of
the spectral ‘fine structure’ associated with optically active
substances in water (Goodin et al. 1993). The successful
quantification of water quality parameters using remote sensing
is affected not only by the type of waters under investigation,
but also by the sensor used (Liu et al. 2003). The remotely
sensed techniques for operational monitoring and management
of water quality parameters (WQPs) depend on the substance
being measured, its concentration, influencing environmental
factors and the sensor characteristics. Effectiveness of remotely
sensed data in water quality assessment of different water
bodies has been examined by numerous researchers (Han, 2006;
Dekker, 2001; Gitelson, 2007, etc.).
415
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
standard, simultaneously with incident irrradiance (Ecal).. Solar
zenith anggles ranged from
m approximately 20° to a maxim
mum of
55°. Meassurements were taken
t
over opticaally deep water and an
average of
o 10 consecu
utive spectra was
w
used. Thee term
reflectancee is defined as the ratio betweeen a reflected and
a an
incident quuantity of light. The ratio can coonsist of two raddiances,
two irraddiances, or rad
diance and irraadiance (Aas, 2009).
Percentagee spectral reflecttance R() was computed
c
as:
2. STUD
DY AREA
The Apalachicola National Estuuarine Research Reserve, 1 of 255
sites designatedd by the Natioonal Oceanic annd Atmosphericc
A
Administration,
covers approxim
mately 246,766 acres (figure 1))
B is connectedd
and has an averaage depth of thrree meters The Bay
t the Gulf of Mexico
to
M
throughh four major inllets: Indian Passs
and West Pass at the western end, and East Pass
P
and Lanarkk
R
Reef
at the easteern end. Most of
o the freshwaterr discharged intoo
t Bay flows from
the
fr
the Apalachhicola River (W
Wang et al. 2010)).
W
Water
in the Baay is moderatelyy stratified. The substrate of thee
B is predominnately soft silt and
Bay
a clay with so
ome sandy areass
(Dardeau et al. 1992). Apalachhicola Bay is a river-dominated
r
d,
b
bar-built
shallow
w estuary. It recceives freshwaterr flows from thee
A
Apalachicola,
C
Chattahoochee,
and
a Flint Riverr system (ACF)),
w
which
drains over
o
60,000 km
m2 of Georgia,, Alabama, andd
F
Florida
(Livingsston 2006). Tides in Apalachicolla Bay are mixedd,
w an uneven high and low tide and a range of 0.2 to 0.6 m
with
(Huang et al. 2002). Water quuality in estuarinne ecosystems iss
ny natural and annthropogenic factors.
affected by man
%R(O) [L(O)u / E(O)innc]u[E(O)cal / L(O)cal]u R(O)cal u100
R()cal is the reflectance of the Specctralon panel liinearly
o each radiometter. At
interpolateed to match the band centers of
each statioon a standard set
s of water quuality parameterrs was
measured. Samples were filtered to estim
mate chlorophylll-a by
meethods.
TSS
was
deterrmined
spectrophootometric
gravimetriically using pree-ashed and tareed filters. Filterrs and
retained particulate matterr were dried (600 °C for at leastt 24 h)
d
field cam
mpaign
and reweiighed. The data was collected during
d Land
carried ouut by the field crew of Centerr for Advanced
Managemeent Information Technologies (CALMIT),
(
Schhool of
Natural Reesources, Univerrsity of Nebraskka-Lincoln, USA
A.
Param
meters
Minn
Max
Chl-a (μgg/l)
2.6
6
21.1
Seston (m
mg/l)
2.4
4
28.7
Secchi Depth
D
(m)
0.35
0.35
Water Depth (m)
1
5.45
Avg
SD
4..8
7.1
5.5
11.7
12.0
8.4
0..9
0.8
0.3
1..8
1.9
0.9
Med
Table 1. Descriptive statistic (Minimum,, Maximum, Meedian,
Averagge, Standard Deviation) of meassured water quallity
parameters
3.2 Airborne Imaging Spectroradiome
S
eter for Appliccations
(AISA) Data
Visible to near infrared (NIR)
(
hyperspecctral airborne im
maging
s
Spectroraddiometer is a vaaluable technology for remote sensing
of the eartth’s surface becaause of its combination of good spatial
and spectrral resolution. Hy
yperspectral rem
motely sensed daata was
acquired bby an aerial rem
mote sensing plattform. The instrrument
array inclluded an AISA
A Eagle hypersspectral imagerr from
Visible to Near Infrared (VNIR). The AISA
A
Eagle is a solidstate, pussh-broom instruument that haas the capabiliity of
collecting data with high
h spatial and sppectral resolutionn. The
e
is 390 to 1000
1
nm in up to 512
spectral raange of AISA eagle
bands. The sensor has an Inertial Navigattion System (INS) and
m) DGPS in orrder to
(Differentiial Global Posiitioning System
provide sppatially accurate data. The AISA
A Eagle pre-proccessing
software provides for the
t
automatic geometric correection,
g, and calculaation of at-platform
rectificatioon, mosaicking
radiance by
b applying calib
bration coefficien
nts referenced to
o wellcharacterizzed spectroradioometric targets (Mishra, 2007)). The
algorithm uses the DGPS and attitude infoormation from th
he INS
to perforrm geometric, georeferencinng and mosaaicking
operationss (Makisara et al.,
a 1994). AISA
A Eagle data ussed for
the presennt study were accquired in the sp
pectral range of 400 to
980 nanom
meter between 0330
0
and 0430 h (CST) on 3rd and
a 4th
April 20066 when the solaar zenith angle was
w close to 70
0o. The
sensor altiitude was (2.0733 km), and the image
i
was acquuired at
nadir at a spatial resolutioon of 2 m and sp
pectral resolution
n of 62
bands. Grround data indiicated low winnd (~ 3 m s-1), high
visibility (40
( km), and cleear skies. The site selected from
m flight
Figure 1. Map showing the reseearch study Areaa: Apalachicola
Bay, Florida
F
3. MATERIA
AL & METHOD
D
3 In situ meaasurements
3.1
Two independeent in situ dataasets were colleected for model
calibration and validation
v
respecctively. The grouund truth data off
A
Apalachicola
B
Bay
included water
w
sampling for laboratoryy
analysis (e.g. chl-a
c
and sestonn), on-site meaasurements (e.gg.
Secchi depth, water
w
depth), weather
w
observattions (e.g. windd
speed and skky condition) and upwellingg radiance andd
downwelling irrradiance hypersppectral data.
Two pairs off Ocean Optics USB 2000
0 hyperspectraal
r
radiometers
waas used (the duaal headed systeem, i.e., upwardd
looking and downward lookiing Ocean Opttics sensor) forr
welling irradiancee
acquiring upwellling radiance L(()u and downw
E inc just ab
E()
bove and below the waterr surface. Thee
h
hyperspectral
daata was collectedd in the spectral range of 400 nm
m
t 900 nm. To
to
T match their transfer functions, the intercalibration of th
he radiometer was
w accomplisheed by measuringg
t upwelling raadiance (Lcal) off a white Spectrralon reflectancee
the
416
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
lines covered ann area of approx
ximately 1.6 km
m2 in the vicinityy
of Apalachicolaa Bay. The imaage data were converted
c
to atp
platform
radiannce by applyinng the calibrattion coefficientss
p
provided
by AIS
SA processing so
oftware ‘Caligeo
o’ for subsequennt
p
processing.
remotely sensed data weere analyzed in a systematic manner.
m
The colleccted spectrum shhows that the quualitative nature of the
acquired ssignals are same,, however, the quuantitative naturre vary
from pointt to point in the water
w
body.
3 Atmospherric correction
3.3
The radiance received
r
by a sensor, Lt(i), at the top of
atmosphere (TO
OA) in a spectrall band centered at a wavelengthh,
i can be dividedd into the follow
wing components (Gordon et al.,
1983):
Lt ( Oi )
L r ( O i ) La ( O i ) T ( O i ) L g ( O i ) t ( O i ) L w ( O i )
Where Lr(i) annd La(i) represents radiances generated alongg
W
t optical pathh in the atmospphere by Rayleeigh and aerosol
the
scattering respeectively; Lg(i) is
i contribution arising
a
from thee
specular reflectiion of direct sun
nlight from the sea surface or thee
sun glint compoonent; Lw(i)is deesired water leaaving radiance; T
is direct atmosphheric transmittannce; and t is difffuse atmosphericc
t
transmittance
off the atmospheree.
The goal of atmospheric correction
c
is to
t remove thee
contributions off scattering in the atmospheree and reflectionn
from the water surface from thee TOA radiancees measured by a
A
Eagle dataa
sensor in the viisible region of the spectrum. AISA
w
were
atmospherrically correctedd by using FLAA
ASH (Fast Lineof-sight Atmosp
pheric Analysis of
o Spectral Hypercubes), a first-p
principles
atmosspheric correctioon algorithm for visible to nearr
infrared (NIR) hyperspectral data. FLAAS
SH atmosphericc
M
codde and typically consists of threee
correction uses MODTRAN
steps (Matthew
w et al., 2003). FLAASH usees the standardd
equation for speectral radiance att the sensor leveel, L, in the solarr
w
wavelength
ran
nge (neglecting thermal emission) from a flaat
L
Lambertian
surfface or its equivaalent (Vermote et
e al., 1994).
Figure 2. Sub-surface
S
hyp
perspectral reflecctance of Apalacchicola
Bay, USA
Figurre 3. AISA reflecctance of Apalacchicola Bay, USA
A
A R
B Re
La
(1 R e S )
(1 R e S )
W
Where
R is pixxel surface reflecctance, Re is surrface reflectancee
averaged over the pixel and a surrounding region,
r
S = thee
do of the atm
mosphere, La = the radiancee
spherical albed
b
backscattered
by
b the atmosphere, and A and
a
B are thee
coefficients thaat depend on atmospheric and geometricc
conditions but not on the surrface. Each of these variabless
depends on thee spectral rangee of the selected channel; thee
w
wavelength
inddex has been om
mitted for simpplicity. The first
t
term
in above eqquation correspoonds to radiancee that is reflectedd
from the surfacee and travels direectly into the sen
nsor. The secondd
t
term
correspondds to radiance frrom the surface that
t
scattered byy
t atmosphere into the sensor,, resulting in a spatial
the
s
blendingg,
mage processingg
or adjacency, effect. ENVI 4.3, digital im
u
to process the AISA data. The image wass
software, was used
first geometricaally rectified to
o UTM (Univeersal Transversee
M
Mercator)
projjection (Zone 16; Datum: WGS84). Thee
geometrically annd radiometricaally corrected AISA
A
image wass
u
used
in the analyysis.
L
The amouunt of TSS presennt in the water body
b
defines thee water
category. Chlorophyll-a concentration and TSS werre not
correlated well (Figure 4)) with the determ
mination coefficient of
linear relaationship R2 < 0.33.
0
It depicts that
t
chl-a was not
n the
only charaacteristic controllling water quality, confirming th
hat the
waters bellonged to typicall case-II water grroup (Gitelson, 2008).
2
F
Figure 4. Correllation between chhl-a and TSS
4 MODEL DE
4.
EVELOPMENT
T
meters in case II waters
The abilityy to monitor watter quality param
requires hhigh resolution remotely sensed data and su
uitable
techniquess to examine thee diverse nature of optical constituents
present inn the water body.
b
Band raatio algorithms were
developedd to establish thee relationship beetween the refleectance
and selectted water qualitty parameters. Band-ratioing
B
did
d not
universallyy give the best results but in many
m
cases did; band
ratioing has been suggestted as the most appropriate app
proach
The sub-surfacee spectral reflecctance and atm
mospherically aree
depicted in figuure 2 and figurre 3 respectivelyy. The acquiredd
spectral data of the Apalachiccola bay, USA represents first
r
reflectance
peakk in the green domain near 550
5 nm and thee
second reflectan
nce peak in the red domain show
wing turbid waterr
w
with
low CDOM
M. The peaks near 700 nm clearly
c
proof thee
p
presence
of chllorophyll in the bay. The field, laboratory andd
417
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
elsewhere (Leg
gleiter et al. 20005). In order to
t elucidate thee
p
potential
of muultispectral rem
mote sensing foor water qualityy
m
monitoring,
the AISA data is integrated in too band width of
A
ALOS/AVNIR/2
2 sensor as folloow;
Band 1
Band 3
500
³
³
420
6900
610
R ( O )d O
Band 2
R ( O )d O
Band 4
³
³
600
520
890
760
o
optically active
In case 2 waters, the presence of other
constituennts (OACs) effeect the nature of
o the signals an
nd the
discriminaation in these con
nstituents is com
mplex. Dall'Olmo
o et al.,
2003 provvided evidence that a three ban
nd reflectance model,
m
originally developed fo
or estimating pigment
p
conten
nts in
b used
terrestrial vegetation (Giteelson et al., 2003), could also be
m
relates piigment
to assess chl-a in turbidd waters. The model
concentrattion to reflectancce R(i) in three spectral bands i:
Pigment concentration = R750*(R670-1-R7000-1)
R ( O )d O
R ( O )d O
4 Chlorophyyll-a
4.1
IIn case I waterss, concentration
ns of chlorophylll-a can be quitee
satisfactorily esstimated with satellite imagees by using ann
empirical modeel and interpreeting the receivved radiance at
a
different waveleengths (Gordon and Morel 19883). However, inn
case II waters ow
wing to complexxity of the waterr constituents thee
r
retrieval
of chl-aa is difficult taskk and needs advvance techniquess
and approaches.. The pronounceed scattering/abssorption featuress
of chlorophyll-aa are: strong absorption
a
betweeen 450–475nm
m
(blue) and at 67
70nm (red), and
d reflectance maaximums at 5500
n (green) and near
nm
n 700nm (Reed Peak). A varieety of algorithmss
h
have
been deveeloped for retrievving chl-a in tuurbid waters. Alll
are based on thee properties of thhe reflectance peeak near 700 nm
m
and the ratio off that reflectancee peak to the reeflectance at 6700
n (Gitelson et al., 2008). Gitellson, 1992 studieed the behaviourr
nm
nce peak near 700nm
7
and conncluded that thee
of the reflectan
700nm reflectannce peak is impo
ortant for the reemote sensing of
inland and coasttal waters with regard
r
to measurring chlorophylll.
H
Han,
2005 poin
nted out that hee spectral regionns 630–645 nm
m,
660–670 nm, 6880–687nm and 700– 735 nm were
w
found to bee
p
potential
region
ns where the fiirst derivatives can be used too
estimate chlorophyll concentraation. Dekker, 1991 mentionedd
t
that
the scatterinng and absorptioon characteristicss of chlorophyll-a can be studdied when moore than one band is usedd.
H
Hoogenboom
e al., 1998 dettermined that a ratio using ann
et
A
Advanced
Visibble–Infrared Im
maging Spectrom
meter (AVIRIS))
b
band
located neaar 713 nm with the band at 667nnm was the most
sensitive for chlorophyll retriev
val for inland waters.
w
A similarr
r
ratio
(R674/R705) has been demonnstrated to be opptimal for inlandd
d Kaufmann, 2000). Three typess
lakes and riverss (Thiemann and
of independent variables weree tested: singlee spectral bandd,
r
ratios
of spectraal bands, and combinations of multiple
m
bands too
develop linear regression eqquations and r2 values weree
computed.
hip b/w chl-a annd 3 band model
Figgure 6. Relationsh
f
The develooped model is off the following form;
-1
-1
1
Chl a ( Pg / l ) m[ R 750 * (R 670 - R 700 )] n ; where m and
n are empirical coefficientts.
NIR-2 (Band3/Band1) is
The relatioonship between chl-a and AVN
demonstraated in figure 7.
Figurre 7. Relationshiip b/w chl-a andd AVNIR-2 band
ds
reflectance ratio
f
The develooped model is off the following form;
m n
Chl a (Pg / l) m[Log(B3 / B1 )]2 n[Log(B3 / B1 )] l ;where m,
and l are empirical
e
coefficcients.
4.2 Totall Suspended Ma
atter (TSM)
TSS conccentrations reguulate light attennuation in inlannd and
estuarine systems. In coastal
c
waters, light scatterin
ng by
suspendedd particles stronngly affects lighht propagation in the
water coluumn (Lee et all., 2005), and determines
d
to a large
extent the magnitude of surface
s
reflectannce (Sathyendran
nath et
d McKee, 200
04 demonstrated the
al., 19899). Miller and
applicationn of MODIS (Teerra) 250m data to quantify TSS
S using
a linear rregression moddel relationship established beetween
MODIS bband-1 (620–67
70nm) and in situ
s
measuremeents of
concentrattions of inorgaanic-dominated TSS in the coastal
c
northern Gulf of Mexiico. Water collour associated
d with
estuarine systems is typiccally characterizzed by high levvels of
M and chloro
ophyll,
total susppended solids (TSS), CDOM
exhibiting a complex mixture of contributing colour
R
b/w
w chl-a and reflecctance ratio
Figure 5. Relationship
IIt was observed that the ratio off R700/R675 is welll correlated withh
T developed model
m
is of thee
chlorophyll-a concentration. The
following form;
m
Chl a ( P g / l ) m ( R 700 / R 675 ) 2 n ( R 700 / R 675 ) l ; where m,
n and l are empiirical coefficientts.
418
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
constituents (Buukata et al., 199
95). Total suspennded matters (orr
s
seston)
represen
nt living organicc matter (mainlyy Phytoplankton))
and inorganic suspended
s
solidds (tripton). Triipton (Inorganicc
m
material
& detriitus) mainly conttribute to scatterring of light withh
low absorptionn. Absorption is
i normally neglected for thee
inorganic particcles such as suuspended sedim
ments. Empiricaal
r
relationships
beetween spectral properties and total suspendedd
m
matter
showed good
g
correlation
n with NIR/Green band ratio ass
illustrated in figure 8. The develloped model is as
a follow;
is found too be well correlaated with reflecttance ratio of R750
7 /R560
(NIR/Greeen). The simplee band ratio techhnique is effecttive in
monitoringg SD by means of
o remotely senssed data.
Figure 100. Relationship b/w
b secchi depth
h and reflectancee ratio
SD (m) m( R750 / R560 ) n ; where m and n are emppirical
coefficientts.
Figure 8. Relationship betweeen TSS and refllectance ratio
TSM ( mg / l )
CLUDING REM
MARKS
5. CONC
m ( R815 / R560 ) n ; Wheere m and n aree
Remote seensing is propoosed as a usefu
ul tool for monitoring
water quallity parameters up
u to several tim
mes per year andd offer
valuable data
d
on the seasonal variability of water quality
y. The
research work
w
demonstrattes the feasibilityy of hyperspectrral and
multispecttral remotely seensed data for monitoring
m
the spatial
and tempooral variations of
o water quality parameters. Thee band
ratio apprroach is effectivve for developm
ment of water quality
q
algorithmss and to minnimize the efffect of confouunding
environmeental variables. It
I was found thaat the simple twoo band
reflectancee ratio R700/R670 is well correlated with chl-a
concentrattion. The threee band model R750*(R670-1-R7000-1) is
found to be
b predictor of chl-a concentrattion in case-II waters.
w
The logarithmic ratio of ALOS/AVNIR--2 band 3 and band
b
1
ncentration in thee study area. How
wever,
was related with chl-a con
mplexity of wateer and wide bannd width, the accuracy
due to com
was not hhigh. The ratio of NIR and green
g
domain is
i best
predictor of
o TSS. In casee of multispectraal remote sensin
ng, the
developedd 3 band modell including ALO
OS/AVNIR-2 band 4,
band 3 annd band 2 is welll correlated witth TSS. Empiriccal and
semi-empiirical algorithm
ms are easy too use; howeveer, the
coefficientts used in empirrical algorithms are derived from
m data
sets that do
d not necessarilly represent all natural
n
variationns. The
developedd algorithms are based on the lim
mited data set. More
M
in
situ water quality data, hyp
perspectral data and multispectral data
d validated the models.
m
Moreov
ver, the
is needed to calibrated and
q
variabless in the
spatial andd temporal variaability of water quality
Apalachicola Bay needss investigation. It is importaant to
a
as ann integral part off water
incorporatte water quality assessment
resources and environm
mental planningg and manageement.
ol for
Remotely sensed data is effective annd efficient too
q
parametters in
monitoringg the distributiions of water quality
inland andd coastal waters and to suppo
ort water manag
gement
strategies.
empirical coefficcients.
The relationship
p between chl-a and AVNIR-2 (Band
(
s
3/Band1) is
demonstrated inn figure 9 and thee model is shownn below.
Figure 9. Relatiionship b/w TSS
S and ALOS/AV
VNIR-2 (3 band)
reflectannce ratio
TSS(mg/ l) m[(B4 B2 ) /(B3 )]2 n[(
[ B4 B2 ) /(B3 )]l ; where m, n
a l are empiriical coefficients.
and
4
4.3
Secchi Deepth (SD)
The measuremeent of water trannsparency has beeen attempted byy
v
various
method
ds, most commoonly based on light
l
attenuationn
p
principles
(Mob
bley 1994). The estimation
e
of ligght attenuation inn
w
water
bodies is not
n a trivial task
k, and therefore simpler methodss
h
have
been propposed for the operational
o
estim
mation of waterr
t
transparency
(G
Gomez, 2009). Th
he best known iss the Secchi discc,
w
which
is a black
k and white discc 20 cm in diam
meter that is usedd
t estimate wateer transparency visually,
to
v
by meaasuring the depthh
at which the dissc is no longer visible. The maain problem withh
estimating wateer transparencyy with the SD
D is the spatiaal
significance of the
t samples, wh
hich are expensive to obtain andd
r
refer
to single-ppoint measuremeents. Remote seensing can be ann
ideal tool for monitoring
m
water transparency. The
T secchi depthh
OWLEDGEME
ENTS
ACKNO
The authoors acknowledge the data suppport provided by
b the
CALMIT, University of Nebraska-Lincol
N
ln, USA and finnancial
support prrovided by KUT, Kochi, Japan.
419
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
waters. Remote Sensing of Environment, 65, pp. 333–340.
Huang W., et al., 2002. Modeling wind effects on sub tidal
salinity in Apalachicola Bay, Florida. Estuarine, Coastal and
Shelf Science, 55, pp. 33–46.
Legleiter, C.J. and Roberts, D.A., 2005. Effects of channel
morphology and sensor spatial resolution on image-derived
depth estimates. Remote Sensing of Env., 95, pp. 231–247.
Lee, Z. P., et al., 2005. A model for the diffuse attenuation
coefficient of downwelling irradiance. J. Geophys. Res. 110:
C02016, doi:10.1029/2004JC002275.
Liu Y., et al., 2003. Quantification of shallow water quality
parameters by means of remote sensing. Progress in Physical
Geography, 27, pp. 24–43.
Livingston, R.J., 2006. Restoration of Aquatic Systems. Boca
Raton, FL: CRC Press.
Lodhi A., et al., 1997. The potential for remote sensing of loess
soils suspended in surface waters. Journal of the American
Water resources Association, Vol. 33, No. 1, 111–117.
Makisara, K., et al., 1994. A system for geometric and
radiometric correction of airborne imaging spectrometer data.
In 1994 International Geoscience Remote Sensing Symposium,
pp. 851 – 853, Geosci. and Remote Sens. Soc., Piscataway, N. J.
Matthew W., et al., 2003. Atmospheric correction of spectral
imagery: Evaluation of the FLAASH algorithm with AVIRIS
data. Proc. SPIE Algorithms and Technologies for
Multispectral, Hyperspectral, and Ultraspectral Imagery,
IX(5093), 474-482.
Miller, M. & Mckee, B.A., 2004. Using MODIS Terra 250m
imagery to map concentrations of total suspended matter in
coastal waters. Remote Sensing of Env., 93, pp. 259–266.
Mishra, D. R., et al., 2007. Enhancing the detection and
classification of coral reef and associated benthic habitats: A
hyperspectral remote sensing approach. Journal of Geophysical
Research, 112, C08014, doi:10.1029/2006JC003892.
Mobley, C.D., 1994. Light and Water: Radiative Transfer in
Natural Waters, London: Academic Press.
Morel A. et al., 1977. Analysis of variation in ocean color.
Limnology and Oceanography, Volume 22, Issue 4, 709-722.
Novo, E., 1989. The effect of sediment type on the relationship
between reflectance and suspended sediment concentration. Int.
Journal of Remote Sensing, Vol. 1 0, pp. 1283 - 1289.
Quibell, G., 1991. The effect of suspended sediment on
reflectance from freshwater algae. Int. Journal of Remote
Sensing, 12, pp. 177–182.
Sathyendranath, S., et al., 1989. A three component model of
ocean color and its application to remote sensing of
phytoplankton pigments in coastal waters. Int. J. Remote Sens.
10: 1373–1394.
Thiemann, S. and Kaufmann, H., 2000. Determination of
chlorophyll content and trophic state of lakes using field
spectrometer and IRS-1C satellite data in the Mecklenburg
Lake district, Germany. Remote Sens. of Env., 73, pp. 227–235.
Vermote, E.F., et al., 1994. Second Simulation of the Satellite
Signal in the Solar Spectrum (6S), 6S User Guide Version 6.0,
NASA-GSFC, Greenbelt, Maryland, p 134.
Wang H., et al., 2010. Detecting the spatial and temporal
variability of chlorophyll-a concentration and total suspended
solids in Apalachicola Bay, Florida using MODIS imagery. Int.
Journal of Remote Sensing, Vol. 31, No. 2, 439–453.
REFERENCES
Aas, E., et al., 2009. Conversion of sub-surface reflectances to
above-surface MERIS reflectance. International Journal of
Remote Sensing, Vol. 30, No. 21, 5767–5791.
Bukata, R.P., et al., 1995. Optical Properties and Remote Sens.
of Inland and Coastal Waters. Boca Raton, FL: CRC Press.
Dall'Olmo, G., et al., 2003. Towards a unified approach for
remote estimation of chlorophyll-a in both terrestrial vegetation
and turbid productive waters. Geophysical Research Letters, 30,
1038. doi:10.1029/2003GL018065.
Dardeau, M.R., 1992. Estuaries. In Biodiversity of the
Southeastern United States, C.T. Hackney, S.M. Adams, and
W.H. Martin (Eds.), NY: John Wiley & Sons, pp. 614–744.
Dekker, A.G., et al., 1991. Quantitative modelling of inland
water quality for high-resolution mss systems, IEEE
Transactions on Geosciences and Remote Sens., 29, pp. 89–95.
Dekker, A.G., et al., 2001. Comparison of remote sensing data,
model results and in situ data for total suspended matter TSM/in
the southern Frisian lakes. The Science of the Total
Environment, 268,197-214.
Doxaran, D., et al., 2002. A reflectance band ratio used to
estimate suspended matter concentrations in sediment-dominant
coastal waters. Int. Journal of Remote Sens., 23, pp. 5079–85.
Gitelson A, 1992. The peak near 700 nm on reflectance spectra
of algae and water: relationships of its magnitude and position
with chl. concentration. Int. J. Remote Sens. 13 3367–73.
Gitelson A A, et al., 2003. Relationships between leaf
chlorophyll content and spectral reflectance and algorithms for
non-destructive chlorophyll assessment in higher plant leaves.
J. Plant Physiol., 160, 271–82.
Gitelson A., et al., 2007. Remote chlorophyll-a retrieval in
turbid, productive estuaries: Chesapeake Bay case study.
Remote Sensing of Environment, 109, 464–472.
Gitelson A., et al., 2008. A simple semi-analytical model for
remote estimation of chlorophyll-a in turbid waters: Validation.
Remote Sensing of Environment, 112, 3582–3593.
Gin, K.Y.H., et al., 2003. Spectral irradiance profile of
suspended marine clay for the estimation of suspended
sediment concentration in tropical waters. International Journal
of Remote Sensing, 24, pp. 3235–3245.
Gomez, D., et al., 2009. Monitoring transparency in inland
water bodies using multispectral images. International Journal
of Remote Sensing, Vol. 30, No. 6, 1567–1586.
Gordon, H. R., et al., 1988. A semianalytic radiance model of
ocean color. Journal of Geophysical Research, 93,
1090910924.
Gordon, H. R., and A. Y. Morel, 1983. Remote Assessment of
Ocean Color for Interpretation of Satellite Visible Imagery: A
Review of Lecture Notes on Coastal and Estuarine Studies. 113
pp., Springer-Verlag, New York.
Goodin, D. et al., 1993. Analysis of suspended solids in water
using remotely sensed high resolution derivative spectra.
Photogrammetric Engg. and Remote Sens., 59, pp. 505–510.
Han Z. and X. Yun, 2006. Suspended sediment concentrations
in the Yangtze River estuary retrieved from the CMODIS data.
Int. Journal of Remote Sensing, Vol. 27, No. 19, 4329–4336.
Hoogenboom, H.J., et al., 1998. Simulation of AVIRIS
sensitivity for detecting chlorophyll over coastal and inland
420