Improved Geometric Modeling of 1960s
KH-5 ARGON Satellite Images for Regional
Antarctica Applications
Wenkai Ye, Gang Qiao, Fansi Kong, Xuwen Ma, Xiaohua Tong, and Rongxing Li
Abstract
Long-term observations of the Antarctic ice sheet will con-
tribute to a quantitative evaluation and precise prediction of
the sea level change induced by global changes in climate.
This paper proposes an improved rigorous geometric model-
ing method for the declassified
KH-5 ARGON
satellite images
collected in Antarctica in 1960s. The scanned film images
are preprocessed beforehand to enhance the quality for fur-
ther analysis. Systematic errors such as lens distortion and
atmospheric refraction are also considered and corrected.
A scheme is proposed to measure the ground control points
for the historical images based on modern image mosaic and
DEM
products. The bundle adjustment results of four blocks
in regions in East Antarctica present a geometric positioning
accuracy of less than one nominal pixel resolution (140 m)
in both horizontal and vertical directions, outperforming the
published results. A regional
DEM
of the ice sheet that repre-
sents the topography in 1963 is then generated from the stereo
ARGON
images for the first time, the evaluation of which shows
its consistency with the modern product but with great value
for studying the recent change history of the ice sheet.
Introduction
Environmental issues resulting from global climate change
can considerably influence human society and natural envi-
ronments. According to the Intergovernmental Panel on Cli-
mate Change’s report (IPCC, 2013), it is highly probable that
the increased warming and sustained mass loss of glaciers
and ice sheets would further accelerate the rate of global sea
level rise. It is of great significance to carry out comprehen-
sive observations over Antarctica given that it holds a major
proportion of the freshwater on earth. Numerous studies have
been performed to investigate the change of the ice sheets, the
mechanism of their interaction with the ocean and responses
to global climate change (Bindschadler and Vornberger, 1998;
Rignot
et al
., 2008; Zwally
et al
., 2015). Although long-term
analysis is desirable, it is often impractical due to the lack
of historical records. Although some early data collected by
in situ
measurements in the 1950s or even earlier are avail-
able, they only cover regions of a limited scale and cannot be
used to support change analysis of ice sheets on a large scale.
Fortunately, a remarkable data source, namely, Declassified
Intelligence Satellite Photography (known as
DISP
), collected
by a series of film-based reconnaissance satellites (
KH
series),
including
ARGON
,
CORONA
, and
LANYARD
, has been made avail-
able (McDonald, 1995).
DISP
provides a unique opportunity to
research early ice sheet configurations, thus greatly extending
the time period of the Antarctic surface observations.
Geometric processing is required as a critical procedure
to utilize these historical scanned satellite films for further
scientific applications. Nonetheless, the handling of these
film-based data presents a variety of challenges, such as the
imperfections of the early imaging mechanisms, artifacts of
the films, deformation through long-period storage, and oth-
ers (Zhou
et al
., 2002a; Galiatsatos
et al
., 2007; Gheyle
et al
.,
2011). Previous research associated with the processing of
film-based images primarily focused on mathematical model-
ing, orthophotos, and Digital Elevation Models (
DEM
) gen-
eration in non-polar regions. For example, Galiatsatos
et al.
(2007) proposed a methodology to geometrically process
CO-
RONA
imagery and generated a
DEM
in the area of Syria using
ground control measured from Ikonos imagery and map-based
contour lines. Surazakov and Aizen (2010) reconstructed the
image geometry of six
KH
-9 images captured in Central Asia
through Bundle Adjustment (
BA
) with Ground Control Points
(
GCPs
) from QuickBird images. Sohn
et al
. (2008) introduced
three mathematical modeling methods for geometrically pro-
cessing
CORONA KH-4B
images. They then generated a
DEM
in
Seoul with horizontal and vertical accuracies of 1.5 pixels.
Numerous related studies have managed to retrieve the
image geometry of historical scanned films with the aid of
additional
GCPs
, but it is not easy to perform the processing
steps in Polar Regions where the ground control is difficult to
access. Only three
ARGON
satellite missions (mission numbers
9034A, 9058A, and 9059A) captured the scenes of Antarctica
in the 1960s (Bindschadler and Seider, 1998). They are con-
sidered a precious resource by investigators for detecting the
surface changes of Antarctic ice sheets in the 1960s, e.g., the
margin changes of ice streams (Bindschadler and Vornberger,
1998), the large-scale structure changes on the surface of ice
shelves (Glasser
et al
., 2009), and the terminus changes of out-
let glaciers (Miles
et al
., 2013). Most of the studies employed
ARGON
imagery by registering to other more recent data using
GCPs
. However, the geometric reconstruction involved was
usually not extensively discussed, nor were the various dis-
tortions existing in the film-based images.
To reduce the geometry-induced errors and further exploit
these valuable historical images, some efforts were made to
establish a mathematical model for
ARGON
imagery in Polar
Regions. Sohn
et al
. (2000) applied least squares adjustment
to refine the Exterior Orientation (
EO
) parameters derived from
the space resection technique. The horizontal accuracy of
the orthorectified
ARGON
images was estimated to be approxi-
mately 150 m. Kim
et al
. (2001) employed a similar technique
to generate an orthorectified mosaic along the coast of Queen
Center for Spatial Information Science and Sustainable
Development Applications, College of Surveying and Geo-
Informatics, Tongji University, Shanghai 200092, China
(
).
Photogrammetric Engineering & Remote Sensing
Vol. 83, No. 7, July 2017, pp. 477–491.
0099-1112/17/477–491
© 2017 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.83.7.477
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
July 2017
477
