Jökull - 01.12.1976, Qupperneq 47

acts on the ground as a buoyant inverted canti-
lever. The result is that the pressure on the
still-grounded ice is a little less than the ice
overburden pressure pjgHj in which Hj is the
glacier thickness, is the density of ice and g
is the acceleration of gravity. Nye estimated this
effect for a rnodel of a perfectly plastic glacier
with yield stress k and an average thickness H;.
The maximum vertical shear stress in the gla-
cier is k, but it drops to zero over a length L
of the grounded ice. The average force per
unit width on the glacier bed is given by F =
pjgHjL — kHj. The net overburden ice pressure
on the bed is pt = p^gHj — k (Hj/L). Nye esti-
mated L = i/£H; which for k = 1 bar provides a
reduction of 2 bars in the height T of the poten-
tial barrier.
For the three glacier lakes considered, the
width W of the potential barrier is a multiple
of the glacier thickness, for Grímsvötn W = 4Hj,
for Vatnsdalslón W = 10Hj and for Grænalón
W = 20Hj. The buoyant inverted cantilever acts
over a length L which can only be a fraction
of the given barriers width, W. Therefore the
cantilever effect cannot reduce the potential
barriers to nil.
The potential barriers in Figs. 3 and 4 are
derived from data on the average elevation of
the glacier bed. The effects of irregularities in
the bed are discussed by Björnsson (1974). Due
to the difference in densities between ice and
water the barrier is reduced in the downcutt-
ings. However, this effect alone cannot reduce
the threshold value to zero.
The jökulhlaups from Grænalón, Vatnsdals-
lón and Grímsvötn are triggered before the
potential barrier is reduced to zero. The poten-
tial barrier seems to be penetrated when the
barrier has been reduced to a certain value
(height 2—5 bars, width 2—7 km). The seal is
broken and a hydraulic connection is introduc-
ed through the barrier. The glacier is lifted
outside the barrier and a jökulhlaup results.
The tunnel effect is a subject to further study.
Melting by convected water may be focussed in
narrow downcuttings and help to open sub-
glacial channels. Such opening may not be an-
nihilated by deformation of the ice at the glacier
bottom when the potential barrier has been
reduced to a few bars. Fluctuations in the sub-
glacial water pressure may play an important
role in the penetration of the potential barrier.
GLACIER-SURFACE LAIÍES
Fig. If shows a cross-section of a lake which
is perched onto a glacier surface depression.
Nye (1976) has explained how such sujtraglacial
lakes can exist in a permeable glacier. Due to
the glacier surface depression water flows in
the vein systern towards the lake and the equi-
potential curve = 0 intersects the glacier bed.
Two types of glacier-surface lakes have been
observed in Iceland. First, lakes in the ablation
area where the depression is caused by differ-
ential surface ablation. Second, lakes at geo-
thermal areas where the depression is caused by
subglacial melting.
I.nkes in Ice Sink-Holes
Ice sink-holes are usually found at shallow
ice in the ablation area where the ice velocity
is very small. Although frequently observed on
glaciers in Iceland, sink-holes have seldom been
described. However, Kocli (1912) and Scheving-
Thorsteinsson (1955) have described the sink-
holes which for years have been observed
around the nunatak area Esjufjöll in Breida-
merkurjökull, SE-Vatnajökull. Fig. 5 shows a
cross-section of one trayshaped hole in this area
(based on Scheving-Thorsteinsson’s description
of Skálatangavatn). The hole collects meltwater
from the glacier surface and is drained engla-
cially through the bottom. The frequency of
dumping is of the order of hours and days.
Similar lakes have been described by Reid and
Clayton (1963).
Surface Lakes in Ice Cauldrons
Surface lakes have been observed in ice caul-
drons which forrn when water drains out of
subglacial water reservoirs. Fig. 6 shows such a
surface lake in the centre of an ice cauldron
Fig. 5. Cross-section of a typical ice sink-hole.
Mynd 5. Þversnið af smálóni við Esjufjöll.
JÖKULL 26. ÁR 45