Digital Elevation Model
(DEM) of Castle Bay, Alaska
||The skies over the coast of Southwest
Alaska are often too cloudy to permit photogrammetric data collection.
NGS has established a technology validation site at Castle Bay, in Southwest
Alaska, to experiment with various types of mapping imagery. Synthetic
Aperture Radar (SAR) systems may well provide imagery with sufficient metric
characteristics to support shoreline mapping applications. The image above
is a digital elevation model (DEM) formed from SAR data, and rendered by
engineers at NGS. The scene is of a 5 nautical mile stretch of the north
coast of Castle Bay. The DEM used in this image was constructed at Sandia
National Labs (SNL) from interferometric SAR (IFSAR) data acquired by NGS
in the Fall of 1997, aboard the ERIM DCS aircraft.
||The scene below shows a 1 nautical mile
section of this same area. The image is a 1 meter resolution slant plane
image formed by engineers at NGS using image formation code written at
SNL, and from the raw phase history data acquired aboard the ERIM DCS aircraft.
The pronounced drainage pattern visible in the scene is on the south slope
of Chiniak Mountain, a 3400 ft. peak located approximately 1 mile from
the coast. This project was a collaboration between NGS, the Office of
the Coast Survey, and the United States Air Force. Preliminary results
from the experiments were published at the ISPRS Commission III Symposium
in Columbus, Ohio in July of 1998.
SAR Image of the Northwest Arm of Castle Bay, Alaska
||We have mapped the North shore of Castle
Bay using photogrammetry, airborne SAR, and RADARSAT fine mode imagery.
The figure below illustrates the scale and phenomenology differences among
the photography, airborne SAR, and spaceborne SAR data. The area shown
in the scene is about 1 nautical mile.
The aerial photography was flown at 25,000
feet with a resolution of about .5 meters. The airborne SAR was flown at
14,000 feet with a depression angle of 30 degrees and with a resolution
of 1 meter. The RADARSAT imagery was acquired from space and has a resolution
of approximately 8 meters. Aside from scale differences, the images differ
because of phenomenology, SAR layover effect, and height of tide. We have
ongoing experiments to assess the accuracy of the SAR derived shoreline
by comparing it to the photogrammetrically derived shoreline.
|| Scale differences:
photo, airborne and spaceborne SAR
||In support of the airborne SAR data collection
in the Fall of 1997, NGS conducted a major field campaign to deploy radar
reflectors. The photo at below left shows a NGS surveyor using GPS to position
a radar reflector. The right photo shows a detailed view of the reflectors,
which were 0.6 meter triangles constructed of aluminum. The field team
deployed reflectors at 45 sites in the Castle Bay area.
Positioning a radar reflector
of radar reflector
||The image below illustrates the appearance
of a .6 meter reflector in a 1 meter resolution SAR image. This image was
formed by NGS engineers using the SNL image formation code.
Radar reflector in the SAR image
||The image below shows an analysis of shoreline
mapped from the RADARSAT fine mode imagery. The residual vector represents
the difference from the photogrammetric shoreline plotted in its correct
geographic orientation over an orthophotograph of Castle Bay. Over the
segment of the coast shown the two shorelines agree to within 28 meters
at the 95% confidence level. We have examined the causes of this
difference and we documented our results in a paper presented at the IEEE
conference in Seattle, Washington in September of 1999.