ul 2004
———
visible-
50 nm
todiode
gration
ftware.
at any
cally,
ibrated
ky and
sensor
surface
rom a
'ctional
"et. al.
> Water
as the
‘om the
motion
flected
ding to
el. al,
for the
In any
| Zenith
in the
can be
ince IS
from à
ly. The
ip that
(A) Is
(1)
factor
for a
ince of
gently
stream
film at
t.
ted by
shaded
of the
> direct
on the
far as
uration
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
(angles, the position of the panel with respect to the sun and to
the sensor) should be consistent from one measurement to next.
23 Sky radiance
The spectral sky radiance signal is measured by pointing the
ASD sensor towards the sky, in the same vertical plane in
which previous measurements have been performed. The
viewing zenith angle should be equal to that of a radiance beam
whose specular reflection on the water would yield a radiance
beam equal to the Zenith angle used for the seasurface
measurements. Measurements under cloudy skies are of little
interest since the sky radiance distribution becomes less
uniform, giving less information on the wavelength
dependency. The best conditions are those for a clear, uniform
sky.
24 Remote sensing reflectance
The reflectance Rrs (0, 6, A) is found according to following
formula:
Rrs (0, 6, à) = Lw(0, 6, à) / Ed(A) (2)
Where 0, $, X stand for zenith and azimuth angles, and the
wavelength at which the reflectance is calculated. Lw(0, 6, À)
is the waterleaving radiance at a particular wavelength and at
the viewing geometry. It is not possible to detect this
component directly. At a given wavelength, the measured
seasurface radiance is the combination of the waterleaving
radiance (Lw) and the sky radiance reflected from the
seasurface (Lr):
Lt(0,0) ^ Lr(0,9) * Lw(0,0) (3)
where Lt stands for the total radiance detected by the sensor
pointed towards the seasurface at a zenith angle 0 and an
azimuth angle ¢. So Lw can be found by subtracting Lr from
the total radiance. To find the Lr, the sky radiance data is used,
by supposing that a portion p of the sky radiance Ls(0,0") is
reflected by the seasurface:
Lr(0,4) = p . Ls(0,9^) (4)
Here, p is a proportionality factor, defined as the total skylight
reflected from the wave-roughened sea surface divided by sky
radiance at the same given zenith and azimuth angles. Finally,
starting from collected data, the reflectance at a particular
geometry and wavelength is calculated as:
Rrs = (Lt-p.Ls) / | (1/pg).Lg 1 (5)
assuming to a good approximation that Spectralon is a
Lambertian reflector (Mobley, 1999).
3. SAMPLING LOCATION AND PERIOD
Within the frame of the project, a set of research cruises have
been performed from right after the ice cover melts to late fall.
327
10 subregions of the ecosystem have been predefined, based on
the previous knowledge obtained from oceanographic
researches on the system. The set of parameters have been
measured following SeaWiFS protocols (Mueller and Austin,
1995 cited in Larouche 2000). The stations are shown in the
figures 1 and 2:
52°,
bai
38° 57° 56
‘ad T
62° 61° 59
70° 69° 68° 67° 65° 67 Of 63
Fig.1 The stations in the estuary and the Gulf of St. Lawrence.
* - Tél n .. air
Sea WiFS campaign
& 4 > s
Estuarine stations j es
f a+ s.
. aes p
e
e ea. $* m
1 su 3
»
i N t a+ ux e
A a er
f er
i 19
E. à
s Hy
$
$us
: 5
IK Ar s ta £ se". [.- se
Fig.2 Estuarine stations.
The radiance data is collected both at the stations and at the
transects between stations, while the ship is in motion. A
number of spectra has been obtained at each station and from
each transect (except for when it rained). A spectrum is
obtained by taking a series of 10 separate measurements from
each of the seasurface, panel, shaded panel and sky. Before
processing, the spectra must be checked to eliminate the
saturated ones either manually or automatically using a
software program. Furthermore, up to two spectra which
deviates highly from the average are removed. Then, the
remaining spectra are averaged for each series of measurements
and the reflectance spectra is computed for each seasurface-
panel-shaded panel-sky sequence.
4. PRELIMINARY RESULTS AND DISCUSSION
Given the high spatial variability, the first step is a spatial
classification of the spectral behaviour based on preliminary