Jökull - 01.06.2000, Side 22

Fiona S. Tweed
HISTORY OF ICE-DAMMED LAKES IN
JÖKULSÁRGIL AND THE POTENTIAL
FOR ICE-DAMMED LAKE FORMATION
Sólheimasandur and Skógasandur (Figure 1) to the
south of Sólheimajökull are primarily composed of
sediments from jökulhlaups induced both by vul-
canological activity and by the outbursts of ice-
marginal ice-dammed lakes (Einarsson et al. 1980;
Maizels, 1991). Water issuing from Jökulsárgilsjökull
and the surrounding valley slopes has previously been
dammed by Sólheimajökull, forming an ice-dammed
lake in Jökulsárgil. Árni Magnússon’s account of Sól-
heimajökull given in Chorographia Islandica, c. 1705
(summarized by Thorarinsson, 1939) describes an ice-
dammed lake in Jökulsárgil which drained every ye-
ar. An ice-dammed lake at the site is also marked
on the Danish General Staff map of 1904, during a
period when Sólheimajökull dammed the Jökulsárgil
canyon, the ice front terminating close to the end of
the Fjallgil canyon (Figure 2). The lake was subst-
antial enough to appear in Thorarinsson’s (1939) table
of the largest lakes in Iceland to have been dammed in
historical times, covering an area of 0.2 km2. Thor-
arinsson (1939) also cites Jökulsá as one of the most
dangerous of the Icelandic glacial rivers, because of
the jökulhlaups that occurred at this time. According
to Sigurðsson (personal communication, 1990), jökul-
hlaups from Jökulsárgil were frequent during the first
third of this century, possibly happening every year.
The last known occurrences of jökulhlaups from Jök-
ulsárgil during this time took place in the summer of
1936, based on accounts from Einar H. Einarsson, a
farmer at Skammadalshóll, Mýrdalur. Sólheimajökull
then receded, the potential for damming up the Jök-
ulsárgil waters was lost and jökulhlaups ceased.
However, the glacier began to advance again in the
late 1960s and has gradually blocked off clear access
for the water from Jökulsárgil. The river has been
flowing from Jökulsárgil under Sólheimajökull as the
glacier has advanced across the valley, a bedrock
obstruction to the west forcing the snout southwards
(Figure 2). Observations made from 1989-1991 and
again in 1996 indicate that the river maintains a tunn-
el through the glacier and probably does so for most of
the year. If the glacier sustains its current position or
continues to advance, it is likely that the tunnel could
seal periodically and an ice-dammed lake may form
in the valley again, giving rise to the potential for jök-
ulhlaups. The conditions required for closure of the
tunnel can be ascertained through the application of
models of ice conduit dynamics.
ICE CONDUIT DYNAMICS -
PROCESSES AND PREDICTIVE
MODELS
A number of processes operate in ice conduits that
theoretically govern their open/closed status. Water
flowing through conduits enlarges the cross-sectional
area by melting of ice from the tunnel walls. This
process is discussed by Liestpl (1956), Röthlisberger
(1972), Shreve (1972), Weertman (1972), Nye (1976)
and Lliboutry (1983). The expansion rate under this
process is generally dependent on the slope of the
tunnel, the water temperature and the flow rate of
water passing through the tunnel. Hooke (1984) sug-
gests that subglacial conduits are not always full of
water, and provides a model to predict melt rates from
ice tunnel walls under situations of open channel flow
at atmospheric pressure, rather than under conditions
of pressure conduit flow operating in full tunnels.
Various conduit closure mechanisms compensa-
te for this melt-widening process by decreasing the
cross-sectional area of the tunnel. Conduits have the
capacity to close by deformation as ice deforms und-
er its own weight (Nye, 1953). The rate at which the
ice deforms is primarily dependent on the thickness
of ice above the tunnel as this controls the over-
burden pressure. Nye’s (1953) model, usually used in
some form to calculate tunnel closure rates, implies
that conduits are circular in cross-sectional form and
bounded solely by ice; in reality this presupposition is
almost always challenged by the nature of subglacial
conduits which deviate markedly from this idealised
cross-sectional shape.
An additional ice conduit closure effect is discus-
sed by Shreve (1972) and Jones et al. (1985), who
comment on the accumulation of frazil ice onto tunn-
el walls with the onset of sub-zero temperatures. This
20 JÖKULL No. 48