Jökull - 01.01.2016, Page 15

The subglacial topography of Drangajökull ice cap, NW-Iceland
from: i) the difference between the Lidar DEM the
DGNSS observations carried out during RES-survey
(late March 2014) and ii) from Pléiades DEM in May
2015 (Belart et al., in open review, 2016). It should be
noted that the presented uncertainties are due to uncer-
tainties in the surface DEMs and the uncertainties in
the seasonal correction. Volume uncertainties caused
by the RES survey are not included. These correspond
to ±0.4 km3 but the error due to this is common in all
values and could, therefore, cause an addition shift of
all values, but would not cause relative changes in the
volume values.
Since the autumn of 1946 Drangajökull has lost
∼17% of its volume (Figure 10). From the autumn of
1994 to the autumn of 2011 the ice cap volume was
reduced by ∼10% corresponding to an annual volume
loss of ∼0.6 %. For comparison the Langjökull ice
cap, ∼170 km SW of Drangajökull suffered an an-
nual volume loss of ∼0.7% in the period 1996–2006
(Björnsson and Pálsson, 2008); Langjökull is however
almost twice as thick on average. In 2011–2014 the
changes in the autumn volume of Drangajökull were
insignificant. The seasonal changes of 0.8–1.1 km3
obtained from the two spring DEMs correspond to
5.5–7.5% volume variations relative to the 2014 au-
tumn volume. This very high annual volume change
is equivalent to about 1/3 of its total volume loss since
the autumn of 1946.
DISCUSSION
Development of ice volumes and future evolution
of glacier river drainage
About 71% of the total volume of Drangajökull is
stored in the part of the ice cap draining towards
west; Leirufjarðarjökull, Kaldalónsjökull and SW-
Drangajökull (Figure 8 and Table 1). In 1946–2011
the area of these outlet glaciers were only reduced by
∼3% and their mass loss was on average 0.16±0.05 m
water equivalent per year during the same period
(Magnússon et al., 2016). No glaciated area in Ice-
land has been reported so close to equilibrium since
the 1940s. The eastern part of Drangajökull has how-
ever lost mass at rates ∼3 times faster than the western
part and has reduced in area by 21% over the same
time span (Magnússon et al., 2016). This east-west
spatial trend is also observed from autumn 2011 to
autumn 2014, when a significant volume increase is
observed on the western part of Drangajökull, while
significant volume loss is observed on the eastern part.
If the decline of Drangajökull continues with a
similar east-west trend it is clear that the western part
of Drangajökull, already storing more than 2/3 of the
ice cap volume, will prevail for much longer than the
eastern part. This also means that the glacial runoff
in the rivers on the east side would vanish much ear-
lier than in the rivers on the west side. As the glacier
becomes thinner the water divides will migrate closer
to the divides obtained assuming no ice loading (red
lines in Figure 9). This trend will cause the drainage
basins of Selá and Mórilla to expand at the expense
of the drainage basins of Bjarnarfjarðará and Reykjar-
fjarðarós.
Horizontal shift between surface and bed topogra-
phy at ice divides
One of the most pronounced features observed when
comparing the surface and bedrock DEMs of Dranga-
jökull is the mismatch between the glacier surface
crest and the topographic ridge in the glacier bed.
This is particularly pronounced for the ridge extend-
ing from the ice cap summit, Jökulbunga, towards the
southernmost part of Drangajökull (Figure 11). The
mismatch typically corresponds to horizontal shift of
300–600 m of the glacier surface crest relative to the
bedrock ridge in southern and westerly direction. At
one location the shift exceeds 700 m. It is this mis-
match, which produces the large difference between
the water calculated drainage catchments with and
without ice loading. Data from a simulated atmo-
spheric dataset (hindcast for 1957–2011, Rögnvalds-
son et al., 2011) reveals that significant winter precip-
itation at the glacier surface crest is most often associ-
ated with strong northeasterly winds, > 10 m s−1 (Fig-
ure 11). This favours snow drift and enhanced snow
accumulation on the lee side of the ridge, i.e. below
and to the west and the south of the ridge. The dis-
placement therefore suggests that wind re-distribution
is an important control not only in snow distribution
(Magnússon et al., 2016; Belart et al., in open review,
2016) but consequently also in the ice cap geome-
JÖKULL No. 66, 2016 15