The radiance measured by NISTAR can be represented as (1)
where M is the flux
leaving the entire Triana-viewed hemisphere in all directions and ADM
is the effective
anisotropic directional model for the wavelength interval
. The value of
the ADM depends on the scenes comprising the hemisphere and their
angular positions relative to the satellite and Sun as given by the
solar zenith
o,
viewing zenith
, and relative
azimuth
angles.
Estimation of the flux from the Triana radiance requires a value of ADM
for each
measurement because the scene changes as the Earth rotates under the
satellite. For simplicity, the shortwave Triana ADM is designated XT, while the longwave ADM is
.
The shortwave ADM can be estimated from the EPIC data at time t as , (2)
where Ri is the weighted average reflectance for
the 443, 645, 870, and 905 nm channels for EPIC pixel i, A is the pixel
area, µo = coso,
N is the total number of pixels used, and X is the
normalized bi-directional reflectance factor for scene type K that is
determined through the cloud screening process and the geography
corresponding to the location of pixel i. The values of X will
be taken from lookup tables like those used by CERES (e.g., Suttles et
al., 1988). Figure 27 shows a Galileo flyby image of Earth, showing
sun-glint region. Figure 28 depicts the pattern of X for clear
ocean from Minnis and Harrison (1984) simulated for three of the Triana
positions viewing the Pacific Ocean. Note, the bright area near the
middle of the globe corresponds to the regions most likely to be
affected by sun-glint. The reflectance for clear ocean increases toward
the limb because of enhanced Rayleigh and aerosol scattering.
The albedo for the hemisphere is, (3)
where S is the insolation over the viewed sunlit portion of the Earth.
The value of X NIRT can be derived using
the average reflectances from the 870 and 905 nm channels to compute Ri from (2). The near-infrared albedo can be
estimated from (3) using the measured near-infrared radiance LNIR(t) and XNIRT.
Unless there are significant differences in the weightings of Ri for the near infrared and the total shortwave in (2)
or if near infrared bi-directional reflectance models become available,
then the ratios for the radiances and fluxes will be almost identical.
With proper temporal averaging, these albedo values can be compared
directly to GCM-derived albedos to verify the climate model
calculations.
The longwave ADM could also be estimated as , (4)
where LIR is the narrowband infrared radiance for
some imager pixel i and is the limb-
darkening function from some established lookup table (e.g., Suttles et
al., 1989). Although the EPIC does not have infrared channels, it may
be possible at some point in the Triana program to use infrared
radiances from the matched pixels from other satellites as discussed
earlier or from CERES as discussed below. In that manner, it would be
possible to provide a reasonable estimate of the OLR for the hemisphere
viewed by Triana.