A digital camera system was used to acquire high-resolution digital images at a
resolution of approximately 15 cm per pixel. The camera system was set to collect
data in three bands: 455–565 nm (blue/green), 560–640 nm (red), and 760–900 nm
(near IR). Mounted on a helicopter, the images were collected from an altitude of
1,500–2,000 ft (457–609 m), minimizing the need for post-processing due to
atmospheric conditions. A high-resolution digital video camera was mounted next
to the multispectral camera to simultaneously record natural-color video. Multiple
in-water teams collected in situ water samples from the upper 0.5 m of the water
column during the overflight, and the sample locations were marked using handheld GPS units. Water samples were filtered in the laboratory to determine the
suspended sediment concentrations and total sediment mass per sample.
Although no post-processing for atmospheric corrections was necessary
because of the low-altitude collection, a geometric correction was applied in order
to spatially match and mosaic the multispectral photos to each other. The process
was repeated on overlapping still images pulled from the natural-color video to
create a second mosaic to assist with image interpretation and visualization. The
resulting mosaics were then further corrected spatially using existing high-resolution LiDAR bathymetry, choosing image-to-image control points and ‘‘slaving’’
the mosaics onto the LiDAR master (Fig. 2.5).
In order to calibrate the image mosaic, suspended sediment concentration (SSC)
values from the in situ water samples were plotted against the digital numbers (DN
values) of various image bands, band combinations, and band ratios. Several types
of regressions were applied to the various plots, with results indicating that a
exponential regression applied to the ratio of Band 1 (blue/green) to Band 2 (red)
resulted in the best fit (r
2 = 0.75, p \ 0.001) (Fig. 2.6). This exponential regression equation was applied to the entire mosaic image resulting in a digital map
representing the SSC in the upper 0.5 m of the water column (Fig. 2.7a).
Fig. 2.5 Map showing natural-color mosaic created from aerial video stills and location of water
samples collected 7 April 2005; overlain on 1999 aerial photography of the south coast of the
island of Moloka‘i. Water samples from transect A were not used in this analysis as they fall
outside the boundaries of the mosaic. Inset upper right: photo of sediment-laden filters displayed
as transects I (left) to A (right), from nearshore (top) to offshore (bottom) (Cochran et al. 2008)
44
S. A. Cochran
resolution of approximately 15 cm per pixel. The camera system was set to collect
data in three bands: 455–565 nm (blue/green), 560–640 nm (red), and 760–900 nm
(near IR). Mounted on a helicopter, the images were collected from an altitude of
1,500–2,000 ft (457–609 m), minimizing the need for post-processing due to
atmospheric conditions. A high-resolution digital video camera was mounted next
to the multispectral camera to simultaneously record natural-color video. Multiple
in-water teams collected in situ water samples from the upper 0.5 m of the water
column during the overflight, and the sample locations were marked using handheld GPS units. Water samples were filtered in the laboratory to determine the
suspended sediment concentrations and total sediment mass per sample.
Although no post-processing for atmospheric corrections was necessary
because of the low-altitude collection, a geometric correction was applied in order
to spatially match and mosaic the multispectral photos to each other. The process
was repeated on overlapping still images pulled from the natural-color video to
create a second mosaic to assist with image interpretation and visualization. The
resulting mosaics were then further corrected spatially using existing high-resolution LiDAR bathymetry, choosing image-to-image control points and ‘‘slaving’’
the mosaics onto the LiDAR master (Fig. 2.5).
In order to calibrate the image mosaic, suspended sediment concentration (SSC)
values from the in situ water samples were plotted against the digital numbers (DN
values) of various image bands, band combinations, and band ratios. Several types
of regressions were applied to the various plots, with results indicating that a
exponential regression applied to the ratio of Band 1 (blue/green) to Band 2 (red)
resulted in the best fit (r
2 = 0.75, p \ 0.001) (Fig. 2.6). This exponential regression equation was applied to the entire mosaic image resulting in a digital map
representing the SSC in the upper 0.5 m of the water column (Fig. 2.7a).
Fig. 2.5 Map showing natural-color mosaic created from aerial video stills and location of water
samples collected 7 April 2005; overlain on 1999 aerial photography of the south coast of the
island of Moloka‘i. Water samples from transect A were not used in this analysis as they fall
outside the boundaries of the mosaic. Inset upper right: photo of sediment-laden filters displayed
as transects I (left) to A (right), from nearshore (top) to offshore (bottom) (Cochran et al. 2008)
44
S. A. Cochran