Jökull - 01.01.2016, Page 7

The subglacial topography of Drangajökull ice cap, NW-Iceland
3 outliers, is 2.8 m and 2.0 m, respectively. For com-
parison the RMS difference for the surface elevation
profiles at same locations is 0.4 m.
Figure 4. The difference between the bedrock el-
evations observed from the RES profiles (using
Cgl=1.70×108 m s−1) and from photogrammetric
DEM in 1994 when the same area was ice free (Magn-
ússon et al., 2016). Positive values indicate RES
elevation above the 1994 DEM. The x-axis signi-
fies glacier thicknesses at corresponding locations
at the time of the RES survey in 2014 (also using
Cgl=1.70×108 m s−1). Location of the comparison
data is shown in Figure 2. – Mismunur botnhæðar,
mæld með íssjá, og landhæðar af korti frá 1994 (y-
ás) áður en svæðið (2. mynd) fór undir jökul í fram-
hlaupi. Jákvæð gildi þýða að botn mældur með íssjá
liggur hærra en landhæð frá 1994. Mæld þykkt jökuls
í mars 2014 er sýnd á x-ás. Gert er ráð fyrir að
ferðahraði rafsegulbylgjunnar í gegnum jökulinn sé
1.70×108 m s−1.
Processing of the Bedrock DEM
Below we describe how the Drangajökull bedrock
DEM is constructed from the RES profiles and from
existing DEMs of Drangajökull surface and sur-
roundings. To obtain a continuous smooth bedrock
DEM fitting the Lidar DEM of 2011 (Jóhannesson
et al., 2013) at the intersection of ice and snow free
ground terrain, the margin of the ice cap and the
attached ice and snow patches was manually digi-
tized from a shaded relief representation of the 2011
DEM (2 m×2 m cell size). This results in an area
of 157 km2, which is significantly larger than the
144 km2 area considered as the dynamically effec-
tive part of the ice cap in 2011 (Magnússon et al.,
2016; the attached ice and snow patches excluded as
previously done for Drangajökull by Jóhannesson et
al. (2013)). The attached patches are to some degree
just snow fields, while others include ice. The most
prominent of those, north of Kaldalón (Figure 1) was
surveyed with RES revealing maximum ∼30 m thick-
ness in the winter 2014 (∼25 m relative to 2011 sur-
face elevation), but most of this data show < 20 m
thickness.
In addition to the RES profiles and the 2011 Lidar
DEM, DEMs from 1985, 1994 and 2005 (Magnús-
son et al., 2016) were used to construct the bedrock
DEM (Figure 2). The parts used from the 1985 and
1994 DEMs were ice free at the time of acquisi-
tion, but were covered with glacier ice in surges in
the 1990s and early 2000s. The 1994 DEM patches
are considered more accurate than the RES profiles
and therefore used instead of the profiles where the
two data sets overlap (Figure 4). The snowfields at-
tached to the ice cap at the time of the Lidar survey in
2011 were substantially smaller when the aerial pho-
tographs were acquired in 2005. These parts of the
2005 DEM together with a small nunatak sticking out
of the SE-part of the ice cap in 2005 (covered in 2011)
were merged into the bed construction.
To obtain the first draft of the bed DEM we first
calculated the glacier thickness at the time of the 2011
DEM at all locations of bed elevation data, listed
above, by subtracting the bed elevation from the sur-
face elevation of the 2011 DEM at corresponding lo-
cations. The thickness at the margin of the ice cap and
the attached ice and snow fields is therefore 0 m by
definition. To avoid 0 m thickness at the ice and snow
patches in the interpolation, typically surrounded with
0 m values from the margin, a thickness value for ev-
ery 50 m×50 m grid cell of the patches was approx-
JÖKULL No. 66, 2016 7