Jökull - 01.01.2005, Síða 103

Seismic characteristics of the Hekla volcano
west) and the Leirubakki fault (in the east) which
are visible at the surface (Einarsson and Eiríksson
1982). However, the hypocentres are displaced about
1 km east from the faults at the surface, and thus the
faults should be dipping approximately 80◦. Recent
field mapping has revealed two formerly unknown
faults, Skarðsfjall fault east of the Hellar fault (Ein-
arsson et al. 2002; Figure 2) and a fault east of the
Leirubakki fault (Einarsson et al. 2003). The faults in
the Leirubakki area are currently named as the west-
ern (formerly known) and the eastern (newly found)
Leirubakki fault (Figure 2). The recently mapped
faults are located above our earthquake lineaments
and are more likely the origins for them.
Although the volcano Torfajökull is next to Hekla,
its seismic behaviour is completely different. It is far
more active, as small earthquakes occur persistently.
Two sorts of events are identified: high-frequency
events in the western parts of its caldera, hypothesised
to be related to a cooling, but solidified magma cham-
ber (Soosalu and Einarsson 1997) and low-frequency
events in the south, which apparently are related to ac-
tive magma (Soosalu and Einarsson 2003). Torfajök-
ull seismicity is volcano-related and not affected by
the transform tectonics of the South Iceland seismic
zone.
CONCLUSIONS
The seismicity at Hekla and its immediate surround-
ings is quite unique and has a dual nature. In non-
eruptive periods there is little seismicity and the few
earthquakes that do occur are not related to the vol-
cano itself. Instead, they have the same characteris-
tics as the seismic activity in the South Iceland seis-
mic zone located to the west of Hekla. The ultimate
eastern terminus of transform tectonics of the South
Iceland seismic zone therefore lies between the volca-
noes Hekla and Torfajökull.
The Hekla eruptions in recent decades have been
quite similar in size and general behaviour. Volcano-
related seismicity occurs at Hekla in the form of an
initial earthquake swarm, continuous low-frequency
tremor and eventual sporadic small earthquakes dur-
ing later phases of the eruption. The eruption-related
seismicity starts only tens of minutes earlier with a
swarm of hundreds of small earthquakes which in-
crease in size towards the onset of the eruption. The
sizes of earthquakes, ML < 3, culminate around the
very start of the eruption. They turn subsequently to
a slight decline in size, continue for a few hours and
then stop altogether. Only few earthquakes occur dur-
ing the later phases. Rather little seismic energy is re-
leased during opening of eruptive conduits at Hekla,
corresponding to a single event of ML 3.4 in 1991 and
ML 3.2 in 2000, respectively. All the detected events
during eruptions have been high-frequency, volcano-
tectonic earthquakes.
Low-frequency volcanic tremor begins simultane-
ously with the onset of the eruption, when a conduit
is open, and is closely related to degassing. Within
minutes it becomes the dominant element in the seis-
mic records. It is most violent during the first hours,
continues throughout the eruption and fades away to-
gether with it. The characteristic frequency band of
Hekla tremor is 0.5–1.5 Hz, with one or a few domi-
nant peaks at 0.7–0.9 Hz. Large attenuation of tremor
with distance compared to the eruption earthquakes
indicates that the tremor has a shallower origin than
the earthquakes. Particle motion observations point to
a large amount of surface waves in the tremor signals.
With close seismograph stations the initial earth-
quake swarm related to the onset of an eruption can
be detected soon after it begins, and with combined
use of seismicity and strain observations it is possi-
ble to foresee Hekla eruptions on a short-time scale,
approximately within an hour. The spectral low-
frequency character of the inter-eruption seismicity at
Hekla proper and its change to high-frequency seis-
micity when the strain is building up may provide a
tool for long-term forecasting of Hekla eruptions, but
its validity needs testing with future observations.
Acknowledgements
The colleagues at the Science Institute, University of
Iceland, at the Icelandic Meteorological Office and at
the Institute of Seismology, University of Helsinki are
thanked for help and advice. The Icelandic Meteo-
rological Office provided the digital SIL data. The
National Power Company of Iceland funds the ana-
logue seismograph stations. H. Soosalu has had finan-
cial support from NorFA, Suomalainen Konkordia-
JÖKULL No. 55 103