corals within the region
equinoctial spring tides
xposure of inter-tidal and
ng in situ coral spectra
nn. Paga is an off-shore,
occur mostly in the reef
as while Laku point is à
>s grow in shallow, rock-
METHODS
Reef Substrates from
nsor: LISS-IV, onboard
lites have been the most
1 reefs for its high spatial
nted with three spectral
regions) and 10 bit level
tral mode has performed
( geomorphology of the
(Navalgund et al. 2010).
m, detection capability of
al colonies within a reef.
nels, positioned at 530 to
) nm usually fall short to
e". Cohabitation of macro
ae along with underlying
varying depths of water
e difficult. Atmospheric
the back-scattered signal
scattering. The back
pixel can thus be a mixed
geneity present in the
1 has been demonstrated
se study using a subset of
spectral) data acquired on
sf. The spectral behaviour
alysed with respect to Top
spectral radiance. No
med on this subset image
ures (in terms of mean
d thirty pixels for each
btained from the subset of
ire 2). This subset image
dard deviation stretch for
ubstrates. Four out of the
resent four different reel
ishable by their respective
Colour Composite (FCO)
xposed sand get well
cel classes, as a substrate
all the three channels. For
reef flat (free of any kind
. green tone as the sand i$
Zoomed view of Outer Reef Flat of Paga Reef
Field Photograph of
Outer Reef Flat of Paga Reef
Benthic ^
Ll l——
Figure 2. Appearance of reef substrates on LISS-IVMX FCC
and on field
The effect of water column in suppressing the magnitude of sandy
substrates from reef is visible in Figure 3 if one compares the
values of mean spectral radiance in all the three channels
represented by grey and cyan triangles. Benthic green
(chlorophyceae) and brown (phyaophyceae) macroalgae groups
can be differentiated in terms of pixels appearing in orange and
brown colours respectively. Spectrally, chlorophyceae group
dominates the phyaophyceae in all the three channels as shown in
Figure 3. The magnitude of difference in their spectral response is
minimal in the red band (spectral channel 2, 620-680 nm) while in
the green band (spectral channel 1, 530-590 nm) there is a slight
increase in this difference. In NIR (spectral channel 3, 770-860
nm) this difference magnifies drastically which is well evident in
Figure 3.
The fiflh pixel group represents mixed pixels (appearing in
different tones in the FCC) randomly selected from the *outer reef
flat' zone of Paga (Figure 2). This zone is naturally characterized
by diverse benthic and litho-substrates including live coral
colonies. The natural diversity allows this zone to appear as a
Tough textured’ zone adjacent to ‘smooth textured’
chlorophyceae dominated areas. Interestingly, the position of
mean spectral radiance of this mixed pixels lie very close to the
centre of the vertical distance representing the magnitude
difference in spectral response of sand on reef and benthic brown
algae categories in green and red bands (red filled triangles vis-à-
vis cyan and light green triangles). In NIR, sand on reef category
is replaced by chlorophyceae representing the upper limit of this
vertical distance as reef sand shows relatively less spectral
response due to water column absorption.
Thus, in NIR the mean spectral radiance of mixed pixels lie
Wihin the vertical range defined by chlorophyceae and
Phyaophyceae. So it can be inferred that sand on reef and benthic
Macro-alga contribute to the backscattered signal of these mixed
pixels. This fact is confirmed if one numerically calculates the
Spectral radiance of mixed pixels assuming that sand on reef and
phyaophyceae contributes in equal proportion to a mixed pixel
Signal in green and red bands while chlorophyceae and
Phyaophyceae in NIR. This is demonstrated in figure 3 by the red
outline triangles against the red filled/solid triangles. This
reaffirms the fact, that even in high (spatial) resolution, broad-
band, multi-spectral images, pixel-based spectral signature of
“oral colonies is dominated and obscured by other reef substrates.
10 Plot: Key
# Pure Sand
4 2
9 5 4 Benthic Green Algae
E 84 4 Benthic Brown Algae
Tu Sand on reef (submerged)
= 7: x a Mixed Substrate
Es £ Mixed Substrate
= $ . (numerical construct)
3 5 # : A
o a
5 4 3
= &
S $
& 4] $
s
2 *
0
1
0 TT FEV ET T TET T T ™ 1
500 : 550 $00 | 650 i 700 750 : 800 850: 900 Wavelength (in nm)
ies Spectral m Le Spectral À (er Spectral ems!
Channel 1 Channel 2 Channel 3
(530-590 nm) (620-680 nm) (770-860 nm)
Spectral radiance values have been plotted against the central wavelengths of the spectral channels present in LISS-IV MX sensor
Figure 3. Multi-spectral signatures of selected substrates of Paga
reef (TOA spectral radiance values observed from IRS-P6
LISSIVMX data acquired on 16™ March, 2005)
Hence there is a definite need to explore the hyperspectral domain
in remote sensing to understand the spectral behaviour of coral
colonies at all possible scales and modes of acquisition: in sifu,
air-borne and space-borne.
3.2. In situ Spectral Measurements and Data Processing
In situ coral spectra were collected during the equinoctial spring
tide (i.e. maximum negative tide = -0.09 m) of March, 2011 when
low tide exposures of reefs coincided with early hours of local
day time (i.e. 09:00 to 11:00 hrs) suitable for passive, proximal
sensing of coral colonies with no or minimal water column. Coral
reflectance spectra were collected with Analytical Spectral
Devices (ASD) Fieldspec3 spectroradiometer having a spectral
range of 350 to 2500 nm and spectral resolution of 3 nm (at 700
nm) and 10 nm (at 1400, 2100 nm). The sampling interval is 1.4
nm for 350-1000 nm wavelength region and 2 nm for 1000-2500
nm regions. The fibre optic probe has a Field of View (FOV) of
25° full conical angle. Since the objective was to study in situ
spectral reflectance of diverse coral communities, a point
sampling strategy was followed. The field spectroradiometer was
calibrated with reference to a Spectralon white plate and
thereafter multiple coral spectra were recorded from different
sample stations. For each station, a minimum of thirty reflectance
spectra was logged along with GPS coordinates, water depth and
water transparency (visual). Spectral measurements were carried
out for twenty two stations over three consecutive days during
09:00 to 10:30 hrs (to reduce illumination variations) when the
live coral colonies were submerged in less than 10 cm of clear,
water column and data logging was completed within 15-minute
period for each station. The field spectra were subsequently
processed with the help of ViewSpecPro software (version 5.6).
Twenty two hermatypic coral targets representing different
taxonomic genera and colony morphologies (with varying levels
of underwater polyp exposures) were sampled on field. Eight
sample stations (representing seven live coral genera and one
bleached coral, with least water depths) were later selected out of
these twenty two stations as pure samples. The details of these
eight coral targets are given in Table 1 and Figure 4 shows their
field photographs.
