SPECTRAL CHARACTERISTICS OF SELECTED HERMATYPIC CORALS
FROM GULF OF KACHCHH, INDIA
Nandini Ray Chaudhury
EHD/ABHG/EPSA, Space Applications Centre, ISRO, Ahmedabad — 380015 — nandinire@sac.isro.gov.in
Commission VIII, WG VIII/6
KEY WORDS: Ecosystem, Hyper spectral, Multispectral, Image, Identification
ABSTRACT:
Hermatypic, scleractinian corals are the most important benthic substrates in a coral reef ecosystem. The existing, high (spatial)
resolution, broad-band, multi-spectral, space-borne sensors have limited capability to spatially detect and spectrally discriminate coral
substrates. In situ hyperspectral signatures of eight coral targets were collected with the help of Analytical Spectral Devices FieldSpec
spectroradiometer from Paga and Laku Point reefs of Gulf of Kachchh, India to study the spectral behaviour of corals. The eight coral
targets consisted of seven live corals representing four distinct colony morphologies and one bleached coral target. The coral spectra
were studied over a continuous range of 350 to 1350 nm. The corals strongly reflected in the NIR and MIR regions with regional central
maximas located at 820 and 1070 nm respectively. In the visible region the live coral spectra conformed to “brown mode” of coral
reflectance with triple-peaked pattern at 575, 600 and 650 nm. All coral spectra are characterized with two distinct absorption features:
chlorophyll absorption at 675 nm and water absorption at 975 nm. The live and the bleached corals get distinguished in the visible region
over 400 to 600 nm region. Water column over the targets modifies the spectral shape and magnitude. First and second-order derivatives
help in identifying spectral windows to distinguish live and bleached corals.
1. INTRODUCTION
Worldwide, hermatypic or reef-building corals occupy a keystone
position in the overall functioning of reef ecosystem. Live coral
cover of a reef habitat is one of the most widely accepted
parameters of coral reef health (Green et al. 2000). Space-borne,
optical remote sensing has sensor limitations in automated
extraction of live coral cover parameter from space data. Digital
delineation of this parameter has so far mostly used the expert
knowledge of local reef systems in terms of specific
geomorphological features and associated benthic classes
(Hochberg and Atkinson, 2003). The existing, medium to high
resolution, broadband, multi-spectral sensors have limited
capabilities to provide synoptic data on live coral cover at
community level. Poor penetration of sunlight and low signal to
noise ratio of backscattered reflectance from deeper reef habitats
are considered to be the major limiting factors to the performance
of these sensors (Green et al. 2000). Therefore, live coral cover
assessment through space-based imaging has been successfully
confined to emergent, shallow, reef flat environments preferably
at low tides. Relative abundance of reef biota and litho-substrates
at different bathymetric depths with varying water column (i.e.
depth, quality, etc.) add high degree of natural variability to coral
reefs as optically complex, shallow water, remote sensing targets.
This poses significant challenge in remote assessment of reef-
sale biodiversity in terms of spatial detection and spectral
characterization of individual corals at colony scale.
As an underwater remote sensing target, coral reefs appear as
Aes of diverse “substrates” or “bottom types" when viewed
Tom the space (Hochberg et al. 2003). Spectral nature of the
space-borne data has been recognized as the basic link between
coral reef substrates and remotely sensed images (Hochberg and
Atkinson, 2003). Substrate detection and quantification efforts
have applied a deterministic, in situ data based, “reef-up”
approach which often use proximal remote sensing of corals and
other reef benthos. This approach calls for spectral cataloguing to
identify distinct spectral features of a substrate. In situ reef-up
approach has demonstrated the potential applications of field
spectroscopy and hyperspectral analyses to characterize the
reflectance properties of individual reef substrates which usually
comprise an image (Goodman and Ustin, 2002). In situ
hyperspectral signatures of corals and other reef benthos obtained
with the help of portable field spectroradiometers have mostly
been used for spectral discrimination of reef substrates (Hochberg
and Atkinson, 2000, Hochberg et al. 2003, Hochberg and
Atkinson, 2003, Kutser et al. 2003) and for health of corals
(Holden and LeDrew, 1998; 1999; Clark et al. 2000). In situ reef
spectra have been also used as input to simulate top-of-
atmosphere spectral reflectance using radiative transfer models
(Lubin et al. 2001).
In situ spectral reflectance of corals per se has been recognized as
a fundamental parameter in coral reef remote sensing (Hochberg
et al. 2004). Spectral characteristics of coral organisms at colony
or at community level have been correlated primarily with coral
pigments, fluorescence and colony morphology (Hedley and
Mumby, 2002; Hochberg et al., 2004; Joyce and Phinn, 2002 and
2003).
Broad-band, multi-spectral signatures obtained from image-pixels
of high (spatial) resolution Indian Remote Sensing (IRS) satellite