hyperspectral images. As defined in the previous section, LiDAR-derived waterleaving reflectance is calculated by adding together the LiDAR-derived water
column volume reflectance and seafloor reflectance. This water leaving reflectance
is then used to scale the hyperspectral water-leaving reflectance across all wavelengths based on the ratio between the LiDAR-derived water-leaving reflectance at
532 nm and water-leaving reflectance at the closest band in the hyperspectral
imagery. The resulting image is referred to as a ‘‘color-balanced’’ mosaic of
hyperspectral imagery (Wozencraft et al. 2008). Figure 7.5 shows a mosaic of atsensor radiance images compared with a color-balanced mosaic of water-leaving
reflectance for Hilo Bay, Hawaii, HI. The LiDAR data were collected by a
SHOALS sensor and the hyperspectral imagery were collected by a CASI-1500.
7.2.4 Constrained Optimization Modeling
LiDAR-derived water column information can be used to constrain the inversion
of the hyperspectral radiative transfer equation. A spectral seafloor reflectance
image is defined here as bottom reflectance across all viable wavelengths of the
hyperspectral imagery (typically limited to just the visible portion of the spectrum). One method for deriving this parameter using just hyperspectral imagery is
by inverting the radiative transfer equations to simultaneously solve for seafloor
reflectance, spectral water column volume reflectance, and spectral water column
attenuation using an iterative non-linear least-squares approach (Lee 2003). As an
alternative, LiDAR data can be used to constrain this inversion, thereby increasing
the accuracy of the derived reflectance. For example, Tuell et al. (2005a) identify
homogenous areas in the LiDAR-derived reflectance image, and calculate water
column attenuation in these areas such that the difference between hyperspectralderived depth and LiDAR-derived depth is minimized. Tuell and Park (2004)
suggested the creation of LiDAR water column layers by interpolating among the
homogenous areas, which Tuell et al. (2005b) refined by using the LiDAR-derived
water column attenuation to scale hyperspectral-derived water column attenuation.
This approach yields spatially-varying water column attenuation at every wavelength of the hyperspectral data for use in the inversion of the radiative transfer
equation. The resulting spectral seafloor reflectance image is improved by
reducing the impact of assumptions regarding constant water column attenuation
throughout a hyperspectral scene. Figure 7.6 shows example reflectance products
generated from SHOALS LiDAR and CASI-1500 hyperspectral data collected in
Fort Lauderdale, FL. The water-leaving reflectance image on the left is an example
of the color-balanced mosaic described in Sect. 7.2.3, while the image on the right
is the derived hyperspectral seafloor reflectance.
LiDAR-derived water depth, water column attenuation, and bottom reflectance
can also be used to constrain the number of parameters in a combined atmospheric-oceanographic spectral optimization model for deriving water column
properties and seafloor reflectance (Kim et al. 2010). The oceanographic portion of
182
J. M. Wozencraft and J. Y. Park
column volume reflectance and seafloor reflectance. This water leaving reflectance
is then used to scale the hyperspectral water-leaving reflectance across all wavelengths based on the ratio between the LiDAR-derived water-leaving reflectance at
532 nm and water-leaving reflectance at the closest band in the hyperspectral
imagery. The resulting image is referred to as a ‘‘color-balanced’’ mosaic of
hyperspectral imagery (Wozencraft et al. 2008). Figure 7.5 shows a mosaic of atsensor radiance images compared with a color-balanced mosaic of water-leaving
reflectance for Hilo Bay, Hawaii, HI. The LiDAR data were collected by a
SHOALS sensor and the hyperspectral imagery were collected by a CASI-1500.
7.2.4 Constrained Optimization Modeling
LiDAR-derived water column information can be used to constrain the inversion
of the hyperspectral radiative transfer equation. A spectral seafloor reflectance
image is defined here as bottom reflectance across all viable wavelengths of the
hyperspectral imagery (typically limited to just the visible portion of the spectrum). One method for deriving this parameter using just hyperspectral imagery is
by inverting the radiative transfer equations to simultaneously solve for seafloor
reflectance, spectral water column volume reflectance, and spectral water column
attenuation using an iterative non-linear least-squares approach (Lee 2003). As an
alternative, LiDAR data can be used to constrain this inversion, thereby increasing
the accuracy of the derived reflectance. For example, Tuell et al. (2005a) identify
homogenous areas in the LiDAR-derived reflectance image, and calculate water
column attenuation in these areas such that the difference between hyperspectralderived depth and LiDAR-derived depth is minimized. Tuell and Park (2004)
suggested the creation of LiDAR water column layers by interpolating among the
homogenous areas, which Tuell et al. (2005b) refined by using the LiDAR-derived
water column attenuation to scale hyperspectral-derived water column attenuation.
This approach yields spatially-varying water column attenuation at every wavelength of the hyperspectral data for use in the inversion of the radiative transfer
equation. The resulting spectral seafloor reflectance image is improved by
reducing the impact of assumptions regarding constant water column attenuation
throughout a hyperspectral scene. Figure 7.6 shows example reflectance products
generated from SHOALS LiDAR and CASI-1500 hyperspectral data collected in
Fort Lauderdale, FL. The water-leaving reflectance image on the left is an example
of the color-balanced mosaic described in Sect. 7.2.3, while the image on the right
is the derived hyperspectral seafloor reflectance.
LiDAR-derived water depth, water column attenuation, and bottom reflectance
can also be used to constrain the number of parameters in a combined atmospheric-oceanographic spectral optimization model for deriving water column
properties and seafloor reflectance (Kim et al. 2010). The oceanographic portion of
182
J. M. Wozencraft and J. Y. Park