Jökull - 01.12.1974, Blaðsíða 22

be exceeded. The slope from Bárdarbunga
ancl the Kverkfjöll ridge north of the lake por-
vides enough pressure gradient in the ice to
ensure lifting in the area of the barrier. But
now consider the actual case. Fig. 13 is a map
of the pressure distribution around the lake
just before a jökulhlaup begins. Accorcjing to
this map, there is still a seal south-east of Gríms-
vötn. In a 2 km wide area the ice overburden
pressure p^ is still 2 bars higher than the pos-
sible subglacial hydrostatic water pressure P.
Accordingly, jökulhlaups woulcl not be expect-
ed to start until the water level in Grímsvötn
has risen 120 m. This suggests that the jökul-
hlaups were triggered by some other process.
As the lifting allowed water to escape from the
lake, it is possible that melting by the convected
water helps to open the subglacial channels.
However, such opening is counteracted by slow
deformation of the ice at the glacier bottom.
On the other hand, the map of the pressure
distribution has been based on the average
elevation of the glacier bed as deduced from
gravity survey and it does not account for varia-
tions in p( and P due to irregularities. Seismic
soundings (Holtzscherer, 1954, and data cited
by Thorarinsson, 1965) have suggested that the
bed topography is very rough in the area east
of Grímsvötn. Therefore, variations in the
height of the potential barrier can be expected.
The seal is situated on the south-east slope
of the subglacial ridge east of Grímsvötn.
Throughout previous centuries erosion by jökul-
hlaups must have cut down waterways at the
head of the subglacial Skeidarárjökull valley. A
narrow valley would not be detected by gravity
survey. A valley woulcl have two effects, both of
which would decrease the threshold value of the
potential barrier. The increased thickness of the
ice in a valley would tend to increase the thres-
hold value of the potential barrier. However,
the increased water pressure in a valley would
tend to decrease the threshold value. Due to
the difference in densities between ice and
water, the net effect would be to reduce the
threshold value. (Downcutting of 200 m would
reduce the threshold value to zero.) The second
effect is mechanical in nature. The shear stress
along the sloping side of a valley woulcl reduce
the ice pressure in the bottom of the valley,
thus reducing the threshold value of the poten-
tial barrier. Fluctuations in the glacier flow
also cause variations in the overburden ice pres-
sure.
One can also argue that as the lifting of
the ice cover on the lake proceeds the still-
grounded ice east of Grímsvötn will be bent
upwards. This effect alone may reduce the
pressure at the bed of the grounded ice by
2 bars (Nye, in press). Waterways that lie deep
enough to reduce the threshold value to zero
when the water Ievel of the lake has risen 100
m may very possibly exist.
The escape of water from Grímsvötn
Once the potential barrier has been reduced,
the seal around Grímsvötn is broken and water
can escape from the lake. As the volume of the
lake is rnuch larger than the volume of the
subglacial waterway the level of the lake is
harclly affected when water is forced down the
subglacial route. The area of the lake is 30—40
km2 and the cross-section of the subglacial
waterway may be of the order of 100 m2. Thus
only a minor drop of 0.1— 0.2 m could force
water down the 50 km subglacial route to Skeid-
arársandur. The obvious consequence would be
that P could lift the overlying glacier along the
whole route as soon as the seal is broken.
Once a jökulhlaup has started frictional heat
from the flowing water will melt the ice, form-
ing subglacial tunnels. The tunnels will tend
to close due to the overburden pressure. How-
ever, the rate of enlargement of the tunnels can
be expected to exceed the rate of closure (Lie-
stöl, 1956; Nye,jn preparation). Therefore, the
jökulhlaup continues through subglacial tun-
nels, even though the glacier may have settled
back onto the rim surrounding Grímsvötn.
The closing off of the jökulhlaup
A surprising feature of the jökulhlaups from
Grímsvötn is that the water level in the lake
does not drop to the level of the subglacial
ridge at 1100 m. Once the water level has
dropped about 100 m the outflow from the lake
stops in a few hours, as shown in Fig. 5. When
this occurs the ice overburden pressure p( at
the rock rim exceeds the hydrostatic water pres-
sure P by 5 bars, as shown in Fig. 14. The tun-
nels may close rapidly due to plastic deforma-
tion at the rim.
20 JÖKULL 24. ÁR