Jökull - 01.01.2016, Blaðsíða 33

First documented surge of Kverkjökull, central Iceland
The 2011 ALS measurements were carried out on
10th September 2011. They have a mean point den-
sity of 0.3 per m2 and were aggregated onto a 5 x 5 m
regular grid based on the mean point density being
slightly lower than that for the 2007 ALS data (Ice-
landic Meteorological Office and Institute of Earth
Sciences, University of Iceland, 2013: DEMs of Ice-
landic glaciers; Jóhannesson et al., 2013). The abso-
lute vertical accuracy of the 2011 DEM is estimated
to be better than 0.5 m but the relative accuracy may
be expected to be similar to the 2007 DEM.
The 2011 DEM (Figure 2B) was corrected ver-
tically for the geoid height using the NKG96 geoid
model for Iceland which is practically identical to the
more recent 2011 geoid model (National Land Survey
of Iceland, 2016). In contrast, the absolute vertical po-
sitioning of the 2007 DEM was derived on the basis
of the calculated altitude above sea-level of the base
dGPS station (see above). In order to eliminate pos-
sible vertical biases, the 2011 DEM was shifted ver-
tically based on elevation differences in 130 random
points off-glacier, snowfields and ice-cored moraines
(yellow triangles in Figure 2A). We interpolated a sur-
face through our points of difference and found no
spatial pattern, suggesting that both ALS datasets are
spatially coherent and indeed uniform in tilt and pitch
and yaw (i.e. IMU) corrections.
The 2007 and 2011 DEMs were each hillshaded to
produce a 3D visual impression of the glacier surface
with unprecedented spatial resolution when compared
to other analyses of glacier surges. Each DEM was
analysed for its representation of ice surface texture,
specifically by (i) simply using a local (between adja-
cent grid cells) slope, and (ii) calculating local rough-
ness as the elevation range between adjacent cells in a
detrended elevation grid. Elevation in each DEM was
detrended by subtracting the mean elevation within a
4 x 4 cell (i.e. 20 m) moving window from the local
(per grid cell) elevation. The local relief of crevasses
and seracs was quantified by calculating the difference
in elevation between adjacent detrended grid cells, i.e.
over a horizontal distance of 10 m.
After the vertical shifting of the 2011 DEM, the
DEMs were subtracted on a cell-by-cell basis to give
a new grid: a DEM of difference (DoD) where posi-
tive values represent an increase in elevation between
August 2007 and August 2011, and negative values
represent a decrease in elevation between those dates
(Figure 2B).
Satellite images
We checked Loftmyndir ehf. (Loftmyndir.is) for aerial
photographs and they had no imagery of Kverkjökull
for the time period 2007 to 2011. We therefore ob-
tained WorldView imagery (Table 2) that were resam-
pled to a common 0.5 m grid cell size for efficiency
of data management purposes. Our georeferencing of
the WorldView images could only be to ∼2 m via co-
registration to the ALS data and orthorectification was
not possible because the images have a very oblique
view and belong to a slightly different strip and to a
different time of day, which together cause long and
deep shadows. The images are from single strips and
are not stereo-pairs so they could not be used pho-
togrammetrically. Nonetheless, the WorldView im-
ages could be used for manual measurement of plan-
form and surface texture changes via on-screen digi-
tising, both at an unprecedented spatial resolution for
glacier surge analyses. Our manual feature tracking
focussed on linked points, such as the ends of neigh-
bouring major crevasses, and linear features such as
longitudinal foliation and major dirt band edges, be-
cause we were rather uncertain of finding individual
points is multiple images.
Bed topography and ice thickness
Bed topography and ice thickness are not usually
known for surging glaciers outside of Iceland but
Murray et al. (2012) have reported measurements of
ice thickness in their surge analysis. In Iceland, many
subglacial bed profiles have been interpolated to es-
timate subglacial bed topography (Björnsson, 1988;
Björnsson et al., 2003; Aðalgeirsdóttir et al., 2005).
However, bed elevation has not been measured along
the main trunk of Kverkjökull. The only bed eleva-
tion measurements that exist in the Kverkfjöll area
are; (i) with uncertainty of ±50 m elevation and from
a single traverse made in 1988 across the southern rim
of the northern caldera that identified ice thickness of
∼200 m (Finnur Pálsson pers. comm.; Björnsson and
JÖKULL No. 66, 2016 33