2004
| ray
ects,
rof
dels.
such
the
rain,
nent
ints.
's of
face
erial
s of
most
Is in
tion
IS to
seful
nary
]ure,
area
the
and
hing
i the
s of
tural
acier
sting
gital
rt for
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
An example of the ranking Strategy is given in table 1 for the
height difference between matched points and preliminary
DEM. The theoretical height accuracy o; is derived from height
above ground.
Height difference to Ranking of points
reliminary DEM outside glacier
Ranking of points
on glacier area
<0.5*c, 2 classes up 1 class up
0.3%g, 10 1.0%g, I class up Same class
1.0*c; to 3.0*o, Same class Same class
3.0*c; to 10.0*c; 1 class down Same class
2 classes down | class down
> 10.0*c,
Table 1. Example of point ranking
2.4 Verification and valuation of results
After ranking based on quality estimation the matched DEM
points are sorted in different files depending on the estimated
quality class. Thus the “bad” points (usually classes 4 and 5)
quickly can be re-measured at an analytical plotter, while
“good” points (usually classes 1 to 3) can be checked at digital
stereo workstations. Time consuming correction of points at the
digital stereo workstation is reduced significantly.
Miss-classifying of points mainly depends on the quality of the
preliminary DEM. As DEM capturing in glacier areas often
means multi-temporal measurement, high quality preliminary
DEM is available in many cases.
Typical miss-classified points are points at the glacier tongue,
often classified as "bad" points caused by really existing terrain
changes (height and orientation). On the other hand these areas
are very important for glaciology and therefore best verification
of these points is of high interest.
In total the presented semi-automatic matching strategy
including the tool of knowledge based point analysing is about
five times faster than traditional analytical point measurement
and about two times faster than conventional point verification
at digital stereo workstations.
The geometric aspects used for searching of best suited stereo
models could be used further on for efficient point capturing in
image blocks with large overlaps (up to 80%) as usually
available when capturing images from low altitude in
mountainous areas. Building additional stereo models with the
images after next, the height accuracy can be improved for
valley regions. The allocation of the DEM points to the best
suited stereo models then could be automated by the presented
software tool. This tool could also be expanded to select best
suited images for ortho mosaics.
3. LASER SCANNING IN GLACIER AREAS
During the last decade airborne laser scanning has made a
decisive technical improvement and has become a standard and
well-accepted method for the acquisition of topographic data
for many applications. First investigations in high mountain
areas have shown good results (e.g. Favey, 2001). Also
accuracy evaluations in comparison with aerial image matching
have been made (Baltsavias et.al. 2001),
The results presented in this paper are based on data captured
for the EU-funded OMEGA project. Its main objective is the
development of an Operational Monitoring system for
European Glacier Areas, aiming to offer accurate and up-to-
755
date information based mainly on remote sensing technology
(Pellikka et al. 2001).
One major aspect for the achievement of the objective is the
generation and utilisation of digital elevation models from
spaceborne and airborne data. In OMEGA digital elevation
models of following sources are constructed: VHR satellite data
(IKONOS, EROS), aerial photography (analogue and digital),
terrestrial photography, airborne SAR, airborne laser scanning.
The method of DEM capturing by airborne laser scanning is
expected to reach high accuracy for mountainous applications
and will be therefore introduced and investigated in detail.
3.1 The principle of airborne laser scanning
Airborne laser scanning integrates a Global Positioning System
(GPS) receiver for determining the position of the sensor, an
Inertial Navigation System (INS) for determining the attitude of
the sensor and the scanning system using laser technology. All
components are time-synchronized. Different technical
solutions for the laser scanning system exist. With the laser
scanning system used in this study (Optech Airborne Laser
Terrain Mapper - ALTM 1225) the laser beam is swept
perpendicular to the ground track, thus producing an even
distribution of data points. The density and distribution of the
data points depend on the scan angle, the scan frequency, the
height above ground, the aircraft speed, the swath overlap and
the reflectance characteristics of scanned surface. A
comprehensive overview on laser scanning technology is given
by Ackermann (1999). The high accuracy and dense coverage
(more than 500.000 points per km? are possible) give the
possibility of generating high-quality DEM.
3.2 Data acquisition and pre-processing
In OMEGA the possibilities and limitations of airborne laser
scanning as an independent method for glaciological
applications are investigated and evaluated (Geist et al. 2003).
For this purpose 10 data acquisition flights were organised by
the Institute for Geography, University of Innsbruck and carried
out between 10/2001 and 9/2003 over glacier areas in the Rofen
valley, Otztal Alps, Austria.
The laser scanner data acquisition was conducted by TopScan
GmbH, Rheine, Germany, with an Optech ALTM 1225 laser
scanner (see table 2).
Measuring Frequency 25.000 Hz
Scanning Angle +/- 20°
Scanning Frequency 25 Hz
1064 nm
2000 m
Laser Wavelength
Max. operating altitude
above ground
Table 2. Parameters of the Optech ALTM 1225 laser scanner
After the acquisition the raw data were pre-processed by
TopScan. The pre-processing comprises the determination of
the absolute position of the laser scan system during the flight
by analysis of the time-synchronized GPS and INS data,
calculation of the relative coordinates, System calibration and
finally calculation and delivery of the coordinates in WGS 84
format. The primary product of data acquisition are coordinates
(x, y, Z) of single reflections. A detailed overview on the pre-
processing steps is given by Wever and Lindenberger (1999).
Data of two permanent GPS receiving stations (Krahberg and
Patscherkofel) were used for the differential correction. The