Jökull - 01.01.2005, Blaðsíða 98

Heidi Soosalu and Páll Einarsson
sual picking of the events from the continuous low-
frequency volcanic tremor. Separate, low-frequency
volcanic earthquakes were either not recorded or they
were hidden in the tremor.
The volcanic tremor
Volcanic tremor at Hekla has only been observed dur-
ing eruptions. Tremor starts simultaneously with the
eruptive activity, thus its appearance in seismograms
can be taken as the seismic expression of the onset of
the eruption. The vigour of the tremor ceases together
with the eruptive activity.
On January 17, 1991, continuous low-frequency
volcanic tremor started at 17:02 GMT, marking the
onset of eruptive activity, thirty minutes after the first
observed earthquake (Soosalu et al. 2003). On Febru-
ary 26, 2000, the tremor appeared at 18:19, 79 min-
utes after the detection of the first earthquake (Soosalu
et al. 2005). The simultaneous beginning of the erup-
tion was verified by an eyewitness account by radio at
the same moment (see Figure 5b).
In 1991 the station SAU, 35 km west of Hekla,
was the closest digital station to record continuous
tremor data at the onset of the eruption. In 2000, in
addition to data from SAU, data were also available
from the station HAU, 15 km west of Hekla, and from
a few other more distant stations. The amplitude of
the volcanic tremor at Hekla rises rapidly, within min-
utes, and becomes the dominant feature in the seis-
mograms, effectively masking the earthquakes. At
the onset of the 1991 eruption, the maximum reduced
displacement (e.g. McNutt 1994) calculated from the
records of the station SAU was about 8 cm2 (Soosalu
et al. 2003).
Tremor remains most vigorous during the first
hour of the eruption, and subsequently declines. The
amplitude of the tremor oscillates. Episodes of high-
amplitude tremor, lasting from a few seconds to about
ten seconds, are separated by moments of lower am-
plitude. To follow the development of the vigour of
the tremor in a simple manner, we calculated its inten-
sity using the procedure presented by Þorbjarnardóttir
et al. (1997). The overall behaviour of the tremor dur-
ing the first hours of the eruption is shown in a graph
expressing its intensity at SAU in 1991 and both SAU
and HAU in 2000, together with the observed earth-
quakes (Figure 6). In 1991, intensity began to decline
sharply about one hour after the onset. This was not
observed in 2000.
The spectrum of the tremor at the onset of both
of the eruptions is remarkably similar. Generally it is
observed that volcanic tremor has a peaked spectrum,
typically with one dominant and a few subdominant
frequency peaks (e.g. Aki et al. 1977; Chouet 1992;
Seidl et al. 1981; Ferrick et al. 1982). The Hekla
tremor also had this pattern. The characteristic fre-
quency band of the tremor is 0.5–1.5 Hz and most of
the time one single outstanding peak existed within
the frequency band 0.7–0.9 Hz (Figure 8a-b). Occa-
sionally there were two or three approximately equal
high peaks. A handful of subdominant peaks sporad-
ically appeared within the 0.5–1.5 Hz band. The lo-
cation of the maximum peak was markedly constant
throughout the first hours, for which we have con-
tinuous data. The amplitudes of the peaks decrease
with time, but the spectral range remains the same.
Although the general pattern is similar at various sta-
tions, some local spectral differences due to path ef-
fects were observed in 2000, with data from several
stations.
The source of volcanic tremor is often inferred
to be shallow, thus the tremor consists mainly of
surface waves (e.g. McNutt 1986; Gordeev et al.
1990; Gordeev 1992; Goldstein and Chouet 1994;
Ripepe et al. 1996). Our particle motion analysis
of the data from the station HAU, both in 1991 and
2000, shows evidence for surface waves, particularly
Rayleigh waves, and a shallow source of the tremor
(Soosalu et al. 2003, 2005). The Hekla tremor atten-
uates faster with distance than the earthquakes, and
is indisputably visible at remote digital seismograph
stations only at the beginning of the eruption. This
also indicates a shallow origin for the tremor, shal-
lower than the earthquakes during the first hours of the
eruption. The tremor is closely related to degassing,
as it first appears when the conduit is open and is most
intense in the beginning, when the eruption has a vio-
lent, explosive phase. When eruption vigour subsides,
the tremor intensity also declines. Occasional tremor-
bursts may reflect gas-bursts.
98 JÖKULL No. 55