9.3.2 Photogrammetric Workflow
9.3.2.1 Aerotriangulation
In the present study we selected the Austrian Gauss-Kru ¨ger coordinate system (strip
M31) as a coordinate reference. As already indicated in the previous section, we
built on available photogrammetric projects carried out by federal (BEV) and
regional mapping (TIRIS) authorities. All photogrammetric work of the present
study was carried out using an ImageStation of Intergraph and Bentley’s CAD
MicroStation. We manually measured 364 distinct natural GCPs in the various
stereomodels of 2006 of which we finally used 221 in the AT of the 2003 photo
block. Due to the difficulties arising from the imaging geometry, i.e. foreshortening
and occlusions, cloud cover and shadows of the 2003 image data we had to measure
all tie points semi-automatically. This means that prospective points were selected
manually, followed by automatic point transfer using image matching. The quality
of the geo-referencing process was checked (1) absolutely by measuring cadastral
triangulation points provided by BEV (mainly mountain peaks), and (2) relatively
by superimposing contour lines derived from the 2006 and 2009 DEMs onto the
2003 stereomodels at a later stage of the photogrammetric workflow. Planimetric
accuracy of single point measurements is better than Æ20 cm. Height accuracy is in
the order of Æ25–30 cm. Areas with limited photo overlap may suffer occasionally
from small systematic model deformations which, however, were not quantified.
The accuracy of the 2009 stereo models was checked visually by superimposing
all available GCPs and official triangulation points, onto the respective
stereomodels. Control measurements confirmed a similar accuracy as for the
2003 result.
9.3.2.2 Digital Elevation Models
A primary source of glaciological studies are high-resolution multi-temporal DEMs
which can be efficiently provided either by airborne laser scanning (ALS) or by
digital photogrammetry (Baltsavias et al. 2001; Wu ¨rla ¨nder et al. 2004; Abermann
et al. 2009, 2010). DEMs, along with digital orthophotos and glacier boundaries
form the basis of a glacier inventory system (Eder et al. 2000; Ka ¨a ¨b 2005).
Data capture for surface reconstruction of the three glacial stages consisted of
two steps: (1) automatic computation of surface points with a grid spacing of 5 m
for the whole study area using ImageStation Automatic Elevations (ISAE) of
Intergraph applying image matching and (2) manual adaptation of erroneous results
located in areas with cloud cover, shadows, occlusions, missing texture and low
contrast, and filling up the voids by 3D mapping of additional surface points using
image data of all three epochs. Finally, the data captured was intermeshed to form a
triangulated irregular network (TIN) using the MGE Terrain Analyst of Intergraph.
Grid-based DEMs of the same size (12 Â 12 km, see Fig. 9.1) with a grid spacing of
9 Glaciological Studies at Pasterze Glacier (Austria) Based on Aerial Photographs
179
9.3.2.1 Aerotriangulation
In the present study we selected the Austrian Gauss-Kru ¨ger coordinate system (strip
M31) as a coordinate reference. As already indicated in the previous section, we
built on available photogrammetric projects carried out by federal (BEV) and
regional mapping (TIRIS) authorities. All photogrammetric work of the present
study was carried out using an ImageStation of Intergraph and Bentley’s CAD
MicroStation. We manually measured 364 distinct natural GCPs in the various
stereomodels of 2006 of which we finally used 221 in the AT of the 2003 photo
block. Due to the difficulties arising from the imaging geometry, i.e. foreshortening
and occlusions, cloud cover and shadows of the 2003 image data we had to measure
all tie points semi-automatically. This means that prospective points were selected
manually, followed by automatic point transfer using image matching. The quality
of the geo-referencing process was checked (1) absolutely by measuring cadastral
triangulation points provided by BEV (mainly mountain peaks), and (2) relatively
by superimposing contour lines derived from the 2006 and 2009 DEMs onto the
2003 stereomodels at a later stage of the photogrammetric workflow. Planimetric
accuracy of single point measurements is better than Æ20 cm. Height accuracy is in
the order of Æ25–30 cm. Areas with limited photo overlap may suffer occasionally
from small systematic model deformations which, however, were not quantified.
The accuracy of the 2009 stereo models was checked visually by superimposing
all available GCPs and official triangulation points, onto the respective
stereomodels. Control measurements confirmed a similar accuracy as for the
2003 result.
9.3.2.2 Digital Elevation Models
A primary source of glaciological studies are high-resolution multi-temporal DEMs
which can be efficiently provided either by airborne laser scanning (ALS) or by
digital photogrammetry (Baltsavias et al. 2001; Wu ¨rla ¨nder et al. 2004; Abermann
et al. 2009, 2010). DEMs, along with digital orthophotos and glacier boundaries
form the basis of a glacier inventory system (Eder et al. 2000; Ka ¨a ¨b 2005).
Data capture for surface reconstruction of the three glacial stages consisted of
two steps: (1) automatic computation of surface points with a grid spacing of 5 m
for the whole study area using ImageStation Automatic Elevations (ISAE) of
Intergraph applying image matching and (2) manual adaptation of erroneous results
located in areas with cloud cover, shadows, occlusions, missing texture and low
contrast, and filling up the voids by 3D mapping of additional surface points using
image data of all three epochs. Finally, the data captured was intermeshed to form a
triangulated irregular network (TIN) using the MGE Terrain Analyst of Intergraph.
Grid-based DEMs of the same size (12 Â 12 km, see Fig. 9.1) with a grid spacing of
9 Glaciological Studies at Pasterze Glacier (Austria) Based on Aerial Photographs
179