Jökull - 01.12.1991, Síða 38

flow in this area at the time. After 1960 the patch
decreased in size and it disappeared in the early
eighties. No holes or openings similar to the ones
formed during the eruptions in 1934 and 1983 have
been observed in this area.
An expedition that visited Grímsvötn on October
5, 1945, reported a circular sinkhole in the “eastern
part of the northwest corner of the depression”
(Áskelsson, 1959). The walls were vertical, the depth
close to 100 m while the diameter was smaller.
Strong smell of sulphur was found in the vicinity of
the crater. Áskelsson (1959) considered that the
crater was created a few days earlier by a small erup-
tion producing mostly steam but also an ash layer
covering the northwestern part of the Grímsvötn
depression. No eruption column was ever seen during
the course of the jökulhlaup that started on
September 16, 1945, culminating around September
27, and no indications of an eruption were seen dur-
ing inspections from the air on September 22, 26 and
October4 (Hannesson, 1958).
Was the sink crater observed by Áskelsson in
1945 created by a volcanic eruption? The question
whether a subglacial mound is situated at the site of
the sinkhole cannot be answered, as its exact location
is unknown. The ash layer which Áskelsson (1959)
considered to have been formed in an eruption short-
ly before October 5 existed on August 30 and had
probably done so for some years. Moreover, no signs
of locally increased heat flow in this area can be seen
from the air photos of 1946. Hence, we find no sup-
port for Áskelsson's (1959) suggestion.
A sinkhole with about 50 m high vertical walls
and a diameter of about 200 m was observed in the
eastern part of Svartibunki at the culmination of the
jökulhlaup in 1954 (Þórarinsson, 1974). Small ice
blocks floated on the water covering the bottom of
the hole (Fig. 14). The hole apparently maintained its
shape for at least seven weeks (July 21 - September
7) but the vertical air photos (Table Al) show that it
had collapsed by September 15. A smaller sinkhole
was formed in the southwest corner of Grímsvötn at
the culmination of the 1954 jökulhlaup. The firn sur-
face surrounding the smaller hole was crevassed but
not stained by ash or tephra. This hole was still visi-
ble on September 15. No eruption column was
observed in 1954 and Þórarinsson (1974) suggested
that the large sinkhole was created by a steam explo-
sion. Jóhannesson (1983) has suggested that a small
eruption took place near the end of the jökulhlaup in
1954.
Air photos from 1945 onwards, show that a caul-
dron has existed in Svartibunki where the sinkhole
formed, at least since 1945. This indicates that the
hole formed above a local geothermal upflow zone.
Furthermore, according to Þórarinsson (1955) and
Eyþórsson (1960), a sinkhole with a powerful fuma-
role was observed at the same location in June 1955
and 1960. On both occasions the sinkhole had verti-
cal walls and a diameter of about 30 m.
The area surrounding the cauldron has been cov-
ered with a dense network of radio-echo soundings
and the site of the sinkhole of 1954 is shown on the
bedrock map on Fig. 2. There are no signs of a
mound indicating the eruption of magma onto the
base of the glacier under the sinkhole. Hence, the
suggestion that a volcanic eruption occurred at this
site in 1954 should be discarded.
Þórarinsson (1974) suggested that the sinkholes in
Grímsvötn were created by steam explosions. A
steam explosion may occur where a hydrothermal
reservoir is sealed by poorly or impermeable caprock
(Williams and McBirney, 1979). Boiling in the reser-
voir leads to increased fluid pressure which may
exceed the lithostatic load. Fracturing in the caprock
may open a channel for the fluid to the surface and a
sudden release of pressure will cause rapid boiling
and catastrophic expansion of the fluid. In Grímsvötn
a steam explosion may take place because a sudden
pressure release occurs during a jökulhlaup, about 10
bars in a fortnight.
The sinkholes may also be created by locally
enhanced melting of ice after increased upwelling of
hydrothermal fluid in reservoirs of high permeability
due to the pressure release during the jökulhlaups.
Within zones of upwelling of hydrothermal fluids in
high temperature geothermal areas, the temperature
gradient in the vertical fluid column follows the
boiling point curve and the ratio of steam and liquid
is presumably stable. A sudden pressure release
reduces the boiling point below the actual tempera-
ture, throughout the column, and leads to vigorous
36 JÖKULL, No. 41, 1991