Jökull - 01.06.2000, Page 23

An ice-dammed lake in Jökulsárgil
process is only effective in situations where water
pressure is less than the ice pressure in the tunnel, and
is generally regarded as a minor contributory factor in
tunnel closure.
Ice collapse has been cited as a factor contribut-
ing to tunnel closure following the drainage of ice-
dammed lakes where little or no water is available to
melt the blocks or to carry them through the tunnel
(e.g. Kerr, 1934; Ricker, 1962). This process is un-
likely in channels deep within ice where compressive
stresses operate, as ice blocks are seldom generated
from the glacier in this type of environment, but wh-
ere extending forces dominate, crevassing or calving
may result in the disintegration of parts of the glacier
and ice blocks may find their way into tunnel systems.
Most analyses of tunnel closure assume that the
longitudinal compressive strain rate normal to the ax-
is of the tunnel is negligible and that the primary
closure process is that of deformation under the over-
burden pressure of ice. However, Jones et al. (1985)
assert that where an ice conduit is orientated across a
glacier, compression of the tunnel caused by glacier
movement may contribute to tunnel closure. They
describe a situation, analogous to that at Sólheima-
jökull, in which a glacier lies across a valley with
forward movement across the longitudinal axis of the
tunnel limited by a valley wall. Compression of the
glacier, under these circumstances would increase the
rate of tunnel closure over that existing solely due
to overburden pressure of ice. However, it must be
possible for the tunnel to move as a whole along
the bedrock through the sliding of the glacier. Thus
glacier sliding rates would not necessarily correspond
to lateral compressive closure rates and it would be
difficult to ascertain what proportion of the glacier
sliding velocity contributes towards closure by lateral
compression. This conduit closure mechanism is not
widely discussed elsewhere in literature dealing with
ice tunnel dynamics.
THE TUNNEL AT SÓLHEIMAJÖKULL
Water from Jökulsárgil enters Sólheimajökull via an
arcuate glacier portal on the northern side of the
glacier, and travels through the glacier by means of
a tunnel for c. 1 km, in an approximately straight line
(Figure 2). The ice above the tunnel is compressed
by glacier flow abutting against a bedrock obstacle to
the west, and exhibits evidence of compressive flow
through the presence of surface water and the lack
of crevassing over most of the tunnel length. Where
crevasses do exist, they are dominantly water-íilled,
again supporting the idea of compressive flow. At
the downstream portal the emerging water sometimes
undercuts the glacier, and ice blocks frequently cal-
ve into the water around the portal. The tunnel has a
diameter of approximately 5 m.
APPLICATION OL MODELS OL ICE
CONDUIT DYNAMICS TO THE
TUNNEL AT SÓLHEIMAJÖKULL
In order to identify the conditions under which tunn-
el closure, and hence ice-dammed lake formation,
would occur at Sólheimajökull it is necessary to apply
models of conduit dynamics to the tunnel at the site.
Nye’s (1953) model of tunnel closure due to over-
burden pressure and Hooke’s (1984) model of melt-
widening were used to ascertain the conditions under
which the tunnel would close. It is assumed that the
water flow in the tunnel is at atmospheric pressure and
that the conduction of heat from the flowing water is
efficient enough to keep the water at the melting po-
int during the flow. The effect of ice pressure on the
melting point of the ice surrounding the tunnel is neg-
lected, as suggested by Hooke (1984).
Measurements were taken at Sólheimajökull dur-
ing 1989, 1990 and 1991 for comparison with the
models described above. Although in many cases,
measurements taken identified a range of values, the
maximum and minimum of the data are used here as it
is the likely threshold values that define the conditions
required for the closure of the tunnel.
A) TUNNEL CLOSURE DUE TO OYERBURD-
EN PRESSURE
Tunnel closure due to ice deformation was calculated
using Nye’s (1953) theory of closure by deformation.
It should be noted that Nye’s theory was developed
for use with cylindrical tunnels fully enclosed in ice,
and the need to rely on this theory to calculate closure
rates for tunnels that deviate from this form is a
JÖKULLNo. 48 21