Jökull - 01.01.2005, Blaðsíða 102

Heidi Soosalu and Páll Einarsson
eruptive activity. Apparently the tremor-producing
mechanism at Hekla could not start before the magma
conduit was opened, indicating that the tremor is
closely related to degassing of magma. Schick (1988,
1992) states that strong tremor is not necessarily ac-
companied by strong lava emission, but strong de-
gassing of a volcano does coincide with strong tremor.
In the light of our observations on the latest Hekla
eruptions, the Hekla tremor reflects the vigour of the
eruption rather than the amount of produced lava.
The sudden swarm of at least one hundred shallow
earthquakes at Hekla, which occurred in June 1991, is
an unusual phenomenon. It may have been a failed
attempt to revive the eruptive activity after its cessa-
tion on 11 March. Two former cases of resumed ac-
tivity at Hekla are known: the August 1980 eruption
continued after several months of quiescence in April
1981 (Grönvold et al. 1983), and the major eruption
in 1766-1768 died down for six months in between
(Þórarinsson 1967). The June 1991 earthquake swarm
was not accompanied by a strain signal indicating
the start of an intrusion, as was observed in January
1991 (Kristján Ágústsson 2000, pers. comm.). Inter-
estingly, the 1980–1981 eruption was similar in this
sense. In the initial phase of the eruption on August
17, 1980 an intrusion-related strain signal was ob-
served, but not when the eruption continued on April
9, 1981 (Ragnar Stefánsson 2003, pers. comm.). An
additional observation supporting an attempt to re-
sume the eruption is the three volcanic-looking earth-
quakes that occurred during the swarm.
Following both the 1991 and 2000 eruptions,
Hekla earthquakes have been few, and with an un-
usual low-frequency appearance, but with clear S-
wave arrivals. This points to brittle failure in the
crust rather than to a volcanic origin. Apparently,
the reason for the low-frequency appearance is that
the crust is still hot and weak after the eruption, and
breaks under low stress-drop. Low-frequency vol-
canic earthquakes typical for many volcanoes in the
world (e.g. Chouet 1996) have almost never been ob-
served at Hekla. The only known exception so far oc-
curred during the June 1991 swarm. High-frequency
earthquakes at Hekla proper are observed almost ex-
clusively during its eruptions. High-frequency Hekla
events are generated during times of high strain, i.e.
during an eruption or an attempt to resume an erup-
tion.
The frequency content of the few inter-eruption
earthquakes at Hekla can potentially be useful for
long-term anticipation of eruptions by giving hint of
a strain build-up. After 1991, the few Hekla earth-
quakes which occurred had a low-frequency appear-
ance, until small high-frequency events were detected
in February 1998 and July 1999. The re-appearance
of high-frequency events may indicate that stress is
starting to build up at Hekla and a new eruption is
in preparation. The signal is vague, though, because
the events are small and few. Our post-2000 eruption
dataset demonstrates that Hekla events have again had
a low-frequency appearance, until September 2004
when a clearly high-frequency earthquake occurred
in the central part of Hekla. Another high-frequency
event was observed at Hekla in March 2005. These
events may be the first seismic indications that stress
is building up again at Hekla. In addition, current tilt
observations suggest increasing magma pressure un-
der Hekla (Sturkell et al. 2005b).
Seismicity at the east end of the South Iceland
seismic zone is of interest because of its similarity to
inter-eruption seismicity in the Hekla-Vatnafjöll area.
We have studied the area east of 20◦12’W and ob-
served that the earthquakes mainly occur along two
N-S lineaments (Soosalu and Einarsson 1997, 2002).
The seismicity is highest in the area of mapped sur-
face faults (approximately 10 km in length), but in to-
tal, the epicentral lineaments are considerably longer,
about 20–30 km. Our observations are in harmony
with the boundary element calculations of Hackman
et al. (1990) which imply that the South Iceland seis-
mic zone faults have to be longer than observed on the
surface, or the zone cannot accommodate the required
transform deformation. Nearly all the hypocentres
are concentrated at 6–12 km depth, with a peak at 8–
10 km. This is consistent with the general pattern of
earthquake depths within the seismic zone; hypocen-
tres deepen towards the east (Stefánsson et al. 1993).
We interpreted earlier (Soosalu and Einarsson
1997) that the earthquake lineaments of the seis-
mic zone are associated with the Hellar fault (in the
102 JÖKULL No. 55