Jökull - 01.12.1984, Qupperneq 17

Fig. 1. Index map of SE-Iceland, showing the
Grímsvötn caldera within the Vatnajökull ice
sheet and the location of the nearest seismo-
graphs used in this study: AB (Aðalból), MI
(Miðfell), KV (Kvísker), KK (Kirkjubæjark-
laustur), SB (Snæbýli), HF (Hafursey), SK
(Skammadalshóll). GF is a telemetered station
operated intermittently on the caldera rim of
Grímsvötn.
1. mynd. Yfirlitskort af Suðausturlandi, semsýnir
þá jarðskjálftamœla landsnetsins, sem nœstir eru
Grímsvötnum: AB (Aðalból), Ml (Miðfell), KV
(Kvísker), KK (Kirkjubæjarklaustur), SB (Snœ-
býli), HF (Hafursey), og SK (Skammadalshóll).
GF er mœlir sem starfrœktur hefur verið á Gríms-
fjalli, þegar aðstæður hafa leyft.
epicenters is located in the Grímsvötn area, but
only a few epicenters are seen near Bárðarbunga
and Hamarinn. To demonstrate the increase of
seismicity in Grímsvötn, magnitude is plotted
against time in the lower part of Fig. 2. The
increase begins in Dec. 1982 to March 1983 and
continues until the outbreak on May 28. This
long term precursor to the eruption was not
recognized until after the eruption because of the
usual delay in seismogram analysis.
THE EARTHQUAKE SWARM ON
MAY 28
An earthquake of magnitude 2.9 at 02:30 h on
May 28 marks the beginning of a dense earth-
quake swarm in Grímsvötn. About 60 locatable
events occurred in the next 9 hours (Fig. 3), 18 of
them of magnitude 3 and above, one event
reached magnitude 4. The swarm was most
intense between 7 and 10 h, and then it gradually
faded out. The last earthquake occurred at 11:47
h and after that no locatable earthquake occurred
in Grímsvötn for more than 3 months (Fig. 2).
A remarkable feature of the swarm is the vari-
able appearance of the earthquakes on the
seismograms, especially with regard to frequency
content and the ratio of maximum amplitude to
coda length. This is clearly seen in Fig. 4, which
shows part of the seismogram from station HF.
Needless to say, this leads to large discrepancies
between duration magnitude and magnitude
based on maximum amplitude for individual
earthquakes. Two events of equal coda length
may have quite different maximum amplitude
and vice versa. Since the earthquakes are tightly
clustered in space, this feature can hardly be a
path effect and must be attributed to the source,
e.g. different depth or stress drop.
The epicenters of the swarm earthquakes are
tightly clustered, slightly to the southeast of the
Grímsvötn caldera. This is somewhat surprising
since the eruption site was in the caldera, near
the southern rim, about 5 km from the center of
the epicentral cluster (Fig. 3). The question
arises, whether or not this is due to a systematic
error in the location procedure. A combination
of uneven station distribution and velocity ano-
malies could lead to systematic mislocation of
epicenters. In the case of Grímsvötn, however,
we have found no reason to suspect the location
procedure. P-wave rays from RRISP-shots (Geb-
rande et al. 1980, Einarsson 1979) passing
through this area did not show any travel time
anomalies large enough to explain mislocation of
5 km. We must conclude that most of the earth-
quakes preceding the 1983 eruption originated a
few kilometers outside the caldera, near the SE
rim. This applies both to the long term precur-
sory events and the swarm earthquakes on May
28.
THE VOLCANIC TREMOR
Shortly after the earthquakes ceased on May
28, continuous tremor became visible on the
nearest seismographs. The tremor was first seen
at approximately 12h on the Kvísker record (KV
JÖKULL 34. ÁR 15