les bureaux
devient donc
ortee de tous
issociée à la
ans avoir à
udera à faire
ie de barre à
irs potentiels
ogrammétrie
el illimité.
ignement du
un probléme
ammetry and
lette, 1988.
v Tool: The
167.
2. PC-Based
s for Block
ammetry and
Redresseur
Archives of
S
igital Image
npany Inc,
ry. PE&RS,
NOVEL SOURCES OF CONTROL FOR AERIAL PHOTOGRAPHY
A.P.R. Cooper’, R. Swetnam" and A.J. Fox
British Antarctic Survey, Natural Environment Research Council, High Cross, Madingley Road, Cambridge, United
Kingdom.
"Department of Geography, The University of Edinburgh, Drummond Street, Edinburgh, United Kingdom.
KEY WORDS: Photogrammetry, Block adjustment, Aerotriangulation, Shape-from-shading
ABSTRACT
Sparse survey control data have constrained large-scale mapping of the Antarctic Peninsula. Extensive aerial photography
is available but photogrammetric techniques are difficult to use because surveyed points are very widely spaced. The
presence of extensive snow-covered areas with insufficient surface detail to allow the topography to be followed in a
stereo-plotter creates a further difficulty. In this study, control information was taken from various sources including
georeferenced satellite images, surveyed points and airborne barometric surveys. A shape-from-shading algorithm allowed
the elevation of snow-fields to be interpolated from barometric height information. These data provided control for the block
adjustment of a series of aerial photographs. A horizontal accuracy of x60 m and a vertical accuracy of +15 m was
achieved in a block of photographs containing only one conventionally surveyed control point. Such errors allow the
construction of an internally consistent large-scale map for an area where inadequate control had restricted mapping to
medium and small scales.
INTRODUCTION
Sparse control data and the presence of snow-fields with ^
insufficient surface texture for stereo-matching limit the
usefulness of aerial photography in Antarctica. New control
data are unlikely to be obtained because of the large areas :
to be covered, difficult access and severe climate. These Alexander
difficulties have limited the scope for large-scale mapping
even in regions well covered by aerial photography. In
coastal or mountainous areas of Antarctica the largest
scale maps with extensive coverage are at 1:200,000 or
1:250,000. To improve the quality of mapping in Antarctica, m
it is necessary to find means of using the aerial
photography. This paper presents new techniques for
creating sufficient control for a block of photographs that
only contains one survey point. The study area is located
in the north-east part of Colbert Mountains, Alexander
Island, off the west coast of the Antarctic Peninsula
(Fig. 1).
+“ ANTARCTICA} -
[]
Colbert
Mountains
DATA
The data used in this study are from four sources:
1) The study area contains one surveyed point in
the British Antarctic Survey (BAS) triangulation
network. This triangulation network was adjusted
to fit Geoceiver positions acquired in 1975/76
(Renner, 1982). The residual RMS error after this
adjustment was +11 m (Knight, 1986).
2) A Landsat TM image (scene ID 50719-12504
acquired on 18 February 1986) covers the area.
It is one of a mosaic of images that was block
adjusted to fit the BAS triangulation network by
the Institut für Angewandte Geodäsie (IfAG),
Frankfurt, using the method described in Sievers image for the effects of surface elevation, using a
and others (1989). IfAG have corrected the coarse digital elevation model (DEM) based on
Figure 1: Location map for the study area.
421
