Jökull - 01.12.1990, Qupperneq 166

the thickness of the floating ice cover has increased
and the volume of the jökulhlaups decreased (Bjöms-
son 1988; Björnsson and Guðmundsson, in prep.).
Björnsson et al. (1982) have suggested that the heat
flux from the geothermal area can be explained by pen-
etration of water into the hot boundaries of a magma
body at shallow depth. The observed cooling trend of
the geothermal area in Grímsvötn may be explained by
lack of magma refill (Bjömsson, 1988; Guðmundsson,
1989).
The activity in Grímsvötn in 1983-84 had very dis-
tinct seismic characteristics. Only a few earthquakes
had been located in the Grímsvötn area before Decem-
ber 1982. Then there was a distinct increase in activity,
and in the following 6 months 20 locatable events oc-
curred there. On May 28, 1983, an eruption broke
out near the southern caldera wall, accompanied by
an intense earthquake swarm (Einarsson and Brands-
dóttir, 1984; Grönvold and Jóhannesson. 1984). The
eruption was small and lasted only a few days. After
the initial outbreak, seismicity dropped to a very low
level and remained low for 4 months. In September
and October 1983, seismic activity increased again,
and more than 39 events occurred in Grímsvötn in the
following 11 months. On August 21, 1984, a burst
of continuous tremor appeared on seismographs as far
as 130 km distant from Grímsvötn. The tremor lasted
about an hour. It originated in the Vatnajökull area as
judged from the relative amplitude on different seis-
mographs. After this event only very few earthquakes
have been found to originate in the Grímsvötn volcano.
The seismicity pattem can be interpreted as the
result of magmatic activity in the Grímsvötn volcano
in the following way. Magma began flowing into a
crustal magma chamber beneath the SE flank of the
volcano in late 1982. In December that year, strain in
the chamber roof passed the elastic limit and seismic
activity increased. On May 28 1983, the chamber wall
failed, and a dyke propagated to the surface, resulting
inaneruption. Pressureinthechamberdropped, strain
in the roof went below the elastic limit, and seismicity
stopped. Magma continued flowing into the cham-
ber, however, and in September 1983, the strain in the
roof again reached the elastic limit. Seismic activity
increased and continued until August 1984. Then the
chamber wall failed again, resulting in a small eruption
that did not reach the surface of the glacier. The pres-
sure in the chamber dropped and the seismicity also.
Now the flow of magma into the chamber was discon-
tinued and the activity stopped. The main evidence for
the eruption is the tremor burst and the sudden drop in
earthquake activity that followed.
The suggestion of a subglacial eruption in Gríms-
vötn in August 1984 is supported by aerial observa-
tions. A reconnaissance flight in August 20, 1984,
revealed that the ice surface had an unusual number of
depressions that could have been formed by localized
melting of the floating ice shelf from below.
Jökulhlaups were reported in the rivers Skaftá and
Þjórsá in connection with volcanic activity in western
Vatnajökull in 1783 (Þórarinsson, 1974). This activ-
ity is generally thought to be related to the Grímsvötn
system, including the gigantic Laki eruption (Þórðar-
son, 1990). The jökulhlaup in Þjórsá would indicate,
however, that the westemmost systems, Bárðarbunga
or Loki were involved in these events as well. The
same applies to the volcanic activity in 1766, which
caused a jökulhlaup in Þjórsá.
Since 1955, twenty jökulhlaups in Skaftá have
originated in the geothermal areas beneath the two
cauldrons located on the Loki Ridge, about 10 km
to the northwest of Grímsvötn (see Bjömsson, 1988).
This apparently reflects a change in geothermal activ-
ity since jökulhlaups were much smaller from this area
before (Bjömsson, 1977, p. 73). Björnsson (1977,
p. 75-76; 1983) pointed out that this change coin-
cided with a reduction in the power of the Grímsvötn
geothermal area, and argued that it may have been
caused by magmatic intrusion into the Loki area from
the Grímsvötn volcano. It is noteworthy that seismic-
ity increased in the Vatnajökull region at about this
time (Tryggvason, 1973).
There is a seismic indication that a small erup-
tion occurred in 1986 during a Skaftá jökulhlaup from
beneath the easternmost ice cauldron. The flood in
Skaftá began on November 29, and on November
30 and the following day short bursts of continuous
tremor were recorded on seismographs around Vatna-
jökull. Relative amplitudes were consistent with a
source near the eastem cauldron. In a reconnaissance
162 JÖKULL, No. 40, 1990