Jökull - 01.01.2012, Blaðsíða 140

E. Magnússon et al.
tion is not appropriate for slopes <8◦ (Figure 3), hence
some unmeasured areas within that slope range have
poorly constrained bedrock topography. For regions
with slopes >20◦ we expect the ice to be thin (typi-
cally <60 m) limiting the range of possible bedrock
elevation significantly.
The RES data, the glacier outlines with assigned
elevations, and the pseudo profiles with estimated
bedrock elevation, were used as an input to calculate
a preliminary bedrock DEM using the kriging inter-
polation function in Surfer. The preliminary DEM
was displayed in Surfer as a contour map with 20 m
interval on top of the surface DEM viewed as both
a shaded relief image and a contour map (20 m in-
terval). The RES data, the glacier outlines and the
pseudo profiles were posted as the final layer. These
data layers viewed simultaneously were used to man-
ually interpret and digitize the contour lines for the
bedrock topography with 20 m intervals. The loca-
tion of contours is more or less fixed along the RES
profiles and the glacier outlines by the preliminary
DEM, but elsewhere the contour locations are less
constrained. In our interpretation we attempt to con-
struct a bedrock topography which may produce the
known surface topography and crevasse pattern ob-
served in the LiDAR DEM. By adopting estimated
bedrock elevation at the pseudo profiles we are gen-
erally able to produce realistic landscape with some
exceptions where the derived relation is clearly in-
valid. This is the case at relatively gentle areas near
the glacier margin where the predicted ice thickness
is clearly overestimated. There we predict smooth in-
terconnection with the land outside the glacier margin.
In some areas of steep surface down-glacier of regions
with thick ice our simple relation (Figure 3) unrealis-
tically predicts thin ice, resulting in deep closed de-
pressions in the subglacial topography. In such cases
we prefer to construct a valley to favour the continu-
ity of the ice flux. This explains the lack of pseudo
profiles in some unmeasured areas in Figure 2 in ad-
dition to those where the surface slope is not between
8◦ and 20◦. In areas lacking both RES-measurements
and ice thickness estimates from our simple relation
(Figure 3) the probability of misinterpreting bedrock
topography is highest.
Each manually digitized contour line was next
taken and a new one linearly interpolated with 25 m
length interval between the points forming the derived
contour line. These interpolated contour lines along
with elevation at the glacier outlines were used to cal-
culate bedrock DEM with kriging interpolation. Be-
fore calculating the final bedrock DEM, the obtained
DEM was reviewed to check for erroneous contours.
Some additional elevation contours in between the 20
m contours were also added to the input dataset at few
locations to avoid negative ice thicknesses as well as
unwanted 20 m contour lines resulting from the krig-
ing interpolation. The DEM of the subglacial bedrock
was finally mosaicked with the regions interpreted as
ice free in the LiDAR DEM (Figure 6).
SUBGLACIAL TROUGHS
Results
The subglacial bedrock of Öræfajökull spans an ele-
vation range from ∼2100 m above sea level down to
at least 200 m below sea level (Figure 6). Such low
bedrock elevation can be found beneath both Svína-
fellsjökull and Fallsjökull outlet glaciers. Skaftafells-
jökull and Kvíárjökull also reach below sea level.
Marginal lakes will continue to grow and new ones
form in the next decades if the fast retreat of the Ör-
æfajökull outlets observed in recent years (Sigurðs-
son, 2011) continues, let alone if further warming
takes place as climate predictions imply (Lemke et
al., 2007; Rummukainen, 2006). If the present ice
cover would instantaneously disappear and the current
bedrock topography were maintained the lakes would
cover an area of ∼33 km2 and contain 2.4 km3 of wa-
ter (Figure 7; Table 1). The lake replacing Skafta-
fellsjökull would be the largest both in terms of area
(∼11 km2) and volume (∼0.8 km3) but the lake re-
placing Svínafellsjökull would be the deepest with a
maximum depth of ∼320 m.
Discussion
It has been put forward that the troughs beneath
Breiðamerkurjökull and Hoffellsjökull outlets of SE-
Vatnajökull were excavated into gently sloping sed-
iment plains during the Little Ice Age (Björnsson,
1996; Björnsson and Pálsson, 2004). If we assume
138 JÖKULL No. 62, 2012