August 10, 2010 / Vol. 8, No. 8 / CHINESE OPTICS LETTERS 721
Satellite retrieval of inherent optical properties in the
turbid waters of the Yellow Sea and the East China Sea
Tingwei Cui (
www
ÉÉÉ
)
1∗
, Jie Zhang (
ÜÜÜ
###
)
1
, Junwu Tang (
///
ÉÉÉ
)
2
,
Yi Ma (
êêê
ÀÀÀ
)
1
, and Song Qing (
ttt
)
1
1
First Institute of Oceanography, State Oceanic Administration, Qingdao 266061, China
2
National Satellite Ocean Application Service, State Oceanic Administration, Beijing 100081, China
∗
E-mail: cuitingwei@f io.org.cn
Received January 11, 2010
In situ-satellite match-ups of radiometric data are established in the turbid waters of the Yellow Sea
and the East China Sea. Inherent optical properties (IOPs) are retrieved by match-up radiometric data
and multi-band quasi-analytical algorithm (QAA). By comparing in situ spectra-retrieved IOPs with the
satellite ones of moderate resolution imaging spectroradiometer (MODIS) and medium resolution imaging
spectrometer ( MERI S), the accuracy of satellite-derived IOPs is quantified. The median of the absolute
percentage difference is found to be approximately 20% for th e total absorption coefficient a
t
(λ) at green
and blue-green bands, and 30% for particulate material backscattering coefficient b
bp
(λ) throughout the
visible bands. The spatial pattern and temporal variability of IOPs along the eastern coast of China are
clarified based on satellite images and the QA A model.
OCIS codes: 010.4450, 280.4788.
doi: 10.3788/COL20100808.0721.
Inherent optica l prop erties (IOPs), including abso rption
coefficient, scattering phase function, and single scatter-
ing albedo
[1]
, are considered inherent to seawater because
they are determined only by the medium (seawater) it-
self, independent of the observation geometry and envi-
ronmental conditions (e.g., illumination and wind speed).
The spatial pattern and temp oral variability of IOPs can
provide important information about ocean properties
and ocean color constituents
[2−3]
.
IOP retrieval mo dels can be divided into two cate-
gories: e mpir ical and semi-a nalytical (SA). The empir-
ical model, based on statistical regression
[4−7]
, is easy
to implement; however, it is limited to r egional appli-
cation. The SA model, with a clear theoretical foun-
dation, represents the development trend. Most deriva -
tion of the SA models are analytical; however, a few em-
pirical relationships are also involved and need be as-
sessed for their applicability to specific waters. Typical
SA models include the Carder algorithm
[8]
, the Ga rver-
Siegel-Maritorena (GSM) algorithm
[9,10]
, and the qua si-
analytical algorithm (QAA)
[11]
.
The waters along the eastern coast of China are char-
acterized by high turbidity. The accuracy of satellite-
retrieved IO Ps is still an open question. Part of the rea-
son lies in the difficulty of obtaining a match-up dataset
between in situ measurements and synchronous satellite
observations. In this letter, the QAA model
[12]
is applied
to satellite images of spectral normalized water-leaving
radiance L
wn
(λ) by moderate resolution imaging spec-
troradiometer (MODIS) and medium resolution imaging
sp e ctrometer (MERIS). The satellite-retrieved IOPs are
compared with the in situ L
wn
(λ)-retrieved ones in the
turbid c oastal waters of the Yellow Sea (YS) and the
East China Sea (ECS). The seasonal characteristics and
the spatial distribution of IOPs along the coast of China
are compared and assessed.
The Changjiang River pours a huge amount of fresh
water and suspe nded particulate material (SPM) into the
YS and ECS which a re located on the continental shelf
of the west bank of the Pacific Ocean. The seasonal vari-
ability of SPM and chlorophyll concentration, as well as
transparency, are significant.
The coastal parts of YS a nd ECS are characterized by
high turbidity. The particle backscattering coefficient
b
bp
(λ) at 532 nm in the study area spans three orders of
magnitude from nearly 0.001 to 1.0 m
−1[13]
. The beam
absorption coefficient a
t
(λ) at 412 nm varies in the range
of 0.1 and 10 m
−1[6]
.
In situ optical data were acquired from the ocean op-
tics experiment conducted in spring (March, April) and
autumn (August, September) of 2003. These data are
regarded as the most comprehensive, high-quality bio-
optical observation results a long the China Seas
[14]
. The
dataset and the adopted measurement methodologies are
described in detail by Tang et al.
[14]
. Spectra l remote
sensing reflectance R
rs
(λ) (sr
−1
) is calculated by
R
rs
(λ) = (L
sw
− rL
sky
)ρ
p
/L
p
π, (1)
where L
sw
, r L
sky
, and L
p
are the measured radiances
of the ocean, reflec ted sky, and re ference plank, respec-
tively; and ρ
p
is the known reflectance o f the reference
plank.
MERIS Level 2 (L2) data products of reduced reso-
lution were collected from the European Spa ce Agency
(ESA) within the framework of the DRAGON project.
MERIS water-leaving reflectance ρ
w
(λ) was converted to
remote sensing r eflectance R
rs
(λ) by
R
rs
(λ) = ρ
w
(λ)/π. (2)
MODIS L2 and Level 3 (L3) products were downloaded
from the National Aeronautics and Space Adminstration
(NASA) Goddard Space Flight Center Distributed Ac-
tive Archive Center (GSFC DAA). MODIS L
wn
(λ) was
1671-7694/2010/080721-05
c
2010 Chinese Optics Letters