Jökull - 01.01.2019, Side 80

Variation in b-value of caldera earthquakes of Bárðarbunga Volcano
the deflation period of Bárðarbunga (Gudmundsson et
al., 2016), when the deflation had reached the limit to
activate the caldera fault. They then continued for the
remaining time of the collapse and eruption. These
large events make up the bulk of the population B
events.
Greenfield et al. (2018) report a study of the seis-
micity of the Askja and Bárðarbunga volcanoes, in-
cluding a study of the b-value of Bárðarbunga caldera
earthquakes. They ran a network of up to 70 seis-
mographs around these volcanoes, beginning in 2008.
Their catalog of Bárðarbunga earthquakes was contin-
uous between January 2013 and the end of 2015, i.e.
only a part of our study period. On the other hand,
their catalog includes earthquakes of smaller magni-
tude. In general, their b-values are higher than the
ones reported here for the periods where the studies
overlap. For the period before the collapse, 2013 to
July 2014, they determine a rather stable b-value of
1.2-1.4, whereas our value is around 0.9, also quite
stable. They report a drop in the b-value during the
collapse, but do not note the severe deviation from
linearity in the magnitude distribution. In the post-
collapse period, March-December 2015, both stud-
ies show variable b-values, beginning at 1.3-1.5 and
decreasing with time. The systematic difference be-
tween the results of the two studies may be caused by
the different methods to determine the b-value. In the
study of Greenfield et al. (2018) it is assumed that the
distribution is linear. In our study we try to account for
the apparent deviation from linearity by fitting three
line segments to the magnitude distribution. Our b-
value is then determined from the slope of the middle
segment, that always has a smaller average slope than
the whole curve.
The reasons for temporal changes in b-value in
the caldera area are not likely to be found in struc-
tural differences. There was no significant change
in the spatial distribution of the hypocenters during
the period of study while the b-value changed. In-
stead, we suggest that the changes are due to tempo-
ral variations in stress in the hypocentral volume at
Bárðarbunga caldera. Slight inflation was detected in
the volcano prior to the failure of the magma cham-
ber wall in August 2014 (Sigmundsson et al., 2015).
The high b-value and low seismicity following the
collapse of the caldera is consistent with low stress
above a deflated magma chamber. The increase in
seismicity in the post-collapse period concurs with re-
sults of GPS-measurements in the region around the
volcano that show re-inflation (Sigmundsson et al.,
2018). The drop in b-value is also consistent with in-
creasing stress in the caldera region.
Geodetic methods in conjunction with seismic-
ity have proved very successful for the monitoring of
activity of Icelandic volcanoes (e.g., Sturkell et al.,
2006; Sigmundsson et al., 2018; Einarsson, 2018).
One of the main obstacles, however, is the glacier cov-
erage of some of the most active volcanoes, such as
Katla, Grímsvötn, and Bárðarbunga, that limits the
usefulness of geodetic methods. Geodetic stations
have to be located on nunataks or bedrock sites out-
side the glacier, which usually means that they are at
considerable distance from the center of pressure vari-
ation within the volcano. We have here documented a
case where the b-value of caldera earthquakes varies
systematically with expected pressure variations. The
correlation between the two therefore may be diag-
nostic, and both can be used as stress indicators. The
relationship might be calibrated on a suitably located
volcano using GPS measurements of inflation and the
analysis of stress perturbations around a magmatic
system, e.g. by Albino et al. (2010), to constrain bet-
ter the failure limit of a magma chamber. The b-value
may thus add to the set of parameters for volcano
monitoring and may be particularly useful for sub-
glacial volcanoes where other methods are difficult.
CONCLUSIONS
1. The b-value of caldera earthquakes associated
with the ongoing activity of the Bárðarbunga vol-
cano shows systematic variations that appear to reflect
physical processes in the caldera.
2. In the period preceding the rupture of the caldera
wall in August 2014 the seismicity of the caldera was
moderate. The b-value was 0.83 for the period 2010
until the magma chamber wall ruptured and a lateral
dike propagated away from the caldera.
3. Earthquakes associated with the subsequent slow
collapse of the caldera in response to the dike prop-
JÖKULL No. 69, 2019 79