Jökull - 01.01.2016, Page 4

Eyjólfur Magnússon et al.
The RES setup is a transmitter (∼3 MHz centre fre-
quency) and a receiver on separate sledges, 30 m
apart, at the centre of the corresponding antenna (see
Mingo and Flowers, 2010, for further details on the
RES survey instrument). This single line setup was
towed by a snow scooter, equipped with a differential
global navigation satellite system (DGNSS) receiver.
The raw RES data are backscatter images where the
x-axis corresponds to the RES-measurement number
and the y-axis is the travel time of received backscat-
tered transmission relative to the triggering time of the
measurement. The receiver measurement is triggered
by the direct wave propagating along the surface from
the transmitter. Each RES-measurement, correspond-
ing to a column in the raw backscatter image, has an
assigned time and rough location from a GNSS in-
strument on the RES receiver sledge. The 3D loca-
tion of the mid-point, M, between the receiver and
transmitter antennae was inferred from the high accu-
racy position and elevation (vertical accuracy <0.5 m)
from the snow scooter DGNSS. The DGNSS and the
RES receiver files were synchronised using the GNSS
timestamps in both files and positions shifted by the
fixed distance between M and the DGNSS antenna.
In cases where the DGNSS failed (<10% of the RES-
profiles) we used planar locations from the GNSS in-
strument on the RES receiver sledge combined with
surface elevation derived from Lidar DEM of Dranga-
jökull in 2011 (Jóhannesson et al., 2013) corrected to-
wards March 2014 using the 2014 DGNSS data. This
generally should result in surface elevation accuracy
of ∼1 m. The backscatter data were then migrated
using 2D Kirchhoff migration (e.g. Schneider, 1978)
to compensate for the width of the radar beam (200–
300 m) in the along track direction of the survey line.
By combining the 2D migration, fixed separation be-
tween the receiver and transmitting antennas as well
the 3D location of M, new backscatter images were
produced (Figure 3). The x- and y-axis of these im-
ages correspond to driven profile length and eleva-
tion in m a.s.l., respectively. The pixel dimension of
these amplitude images, is dx=5 m and dy=1 m, corre-
sponding roughly to the sampling density when mea-
suring with ∼1 s interval at ∼20 km hour−1, and the
80 MHz vertical sampling rate.
Figure 2. Distribution of data used for construction of
the bedrock DEM overlain on a contour map (50 m
elevation interval shown with grey lines) of the sur-
face of Drangajökull ice cap in 2011 (Jóhannesson et
al., 2013). Blue lines represent RES profile data. Red
line indicates the glacier margin in 2011 (enveloping
the area considered the dynamically active part of the
ice cap (Magnússon et al., 2016)). The black margin
lines also includes snow and ice patches attached to
the ice cap in 2011. Purple clusters show areas of the
1985, 1994 and 2005 DEMs (Magnússon et al., 2016)
within the extended glacier margin (black) that were
ice and snow free during acquisition of respective
aerial photographs. Green dots show locations where
thin snow and ice was roughly estimated based on
approximation explained in main text. The cyan box
shows location of data used in the comparison shown
in Figure 4. – Dreifing íssjársniða og annara gagna
sem notaðar voru til að skorða botnhæðarkort af
Drangajökli. Bláar línur sýna íssjársnið mæld í mars
2014. Fjólubláir klasar sýna hluta úr hæðarkortum
frá 1985, 1994 og 2005 af svæðum sem voru ís-
og snjólaus þegar loftmyndirnar, sem hæðarkortin
byggja á, voru teknar. Grænir klasar sýna svæði þar
sem þykkt þunns snjós og/eða íslags var gróft áætluð
með nálgun sem skýrð er í megintexta.
4 JÖKULL No. 66, 2016