Jökull - 01.01.2019, Page 67

Sgattoni et al.
composition of the water discharged (Wynn et al.,
2015). Seasonal variations in the glacial-geothermal
system may be responsible for these summer peaks,
as suggested by Wynn et al. (2015). However, in the
1955, 1999 and 2011 unrest episodes, the jökulhlaups
have been particularly catastrophic and have been ac-
companied by an increased seismicity rate and occa-
sionally by tremor (in 1999 and 2011), indicating a
more dramatic change in the volcanic/geothermal sys-
tem than usual.
The 2011 unrest was associated with increased
heat released by the volcano as indicated by the wa-
ter accumulation under the glacier that started a year
before (Guðmundsson et al., 2014). The tremor and
sudden sinking of ice cauldrons and jökulhlaup that
occurred in July 2011 may have been due to a geother-
mal event, a volcanic (magmatic) event such as a sub-
glacial eruption or a combination of the two, such
as a shallow magma intrusion leading to increased
geothermal activity. It is not straightforward to draw
conclusions about whether a subglacial eruption oc-
curred. According to Galeczka et al. (2015), the
chemical analysis of floodwater does not show evi-
dence that the water came into contact with magma.
The analysis of the tremor conducted by Sgattoni et
al. (2017) suggests that most of the signal is associ-
ated with volcano-related processes occurring at the
sites of the active cauldrons and an additional small
portion of the signal is associated with the flood it-
self. The duration of the tremor and the inferred flood
velocity are consistent with the tremor burst having
been generated by boiling phenomena and/or explo-
sions induced by the water release from the hydrother-
mal system. This would imply that a geothermal
source may have sufficed to generate the tremor and
the flood, with no need for a magmatic event. How-
ever, the notable increase of seismic activity inside
the caldera that continued for months after the jökul-
hlaup is hardly consistent with an exclusively geother-
mal event and may indicate that the unrest did involve
magma, although no eruptive products were identified
(Sgattoni et al., 2017). Concerning this, it must be
noted that, in a subglacial volcano like Katla, small
eruptions may escape the geological and historical
records because they are concealed by the glacier.
In addition to the tremor and increased caldera
seismicity, the 2011 unrest marked also the activa-
tion of a new seismic source on the southern flank.
Sgattoni et al. (2016b) suggested that this seismicity
may be associated with a shallow hydrothermal sys-
tem activated in association with the unrest episode,
although no direct evidence of hydrothermal activ-
ity was found. However, a possible connection to
magmatic processes cannot be ruled out. The ge-
ological study of the Gvendarfell area demonstrated
that the southern flank of Katla acted as the locus of
both tectonic activity and flank eruptions fed by ra-
dial dykes outside the caldera during its pre-Holocene
(to recent?) history. We have identified two areas of
magma ascent, corresponding to the Gvendarfell ridge
and the Gæsavatn lake, which indicates that volcanic
activity has already occurred close to the area where
the current south-flank seismicity is recorded. There-
fore, it cannot be excluded that the Gvendarfell seis-
mic cluster is related to renewed eruptive processes.
The lava bodies of silicic composition exposed in the
Gvendarfell area are comparable in size to the volume
occupied by the hypocenter distribution that was ob-
tained from relative location (Sgattoni et al., 2016a).
Thus, the intrusion of a similar lava body should be
taken into account as a possible source for the seismic
events, but clearly the depth of intrusion will signifi-
cantly influence the nature of associated geophysical
signals. As demonstrated by Krafla central volcano,
rhyolitic intrusion depths may vary from kilometres
(e.g. the rhyolitic magma intercepted by IDDP-1 bore-
hole at Krafla at a depth of 2.1 km; Elders et al., 2011)
to superficial intrusions beneath ice only tens of me-
tres in thickness (Tuffen and Castro, 2009).
Our reconstruction has profound implications on
Katlas’s hazard potential. We in fact observed that
both basaltic and rhyolitic volcanic activity outside
of the caldera seems controlled by radial dykes that
can cause flank eruptions of different compositions.
Moreover, activity in the caldera (likely controlled
by magma ascent along the ring faults) can be con-
nected to or trigger activity outside the caldera, as ob-
served in 2011. Inside the caldera, any surface ac-
tivity (geothermal or magmatic) is influenced by ice,
while magma transport to the flank can lead to sub-
66 JÖKULL No. 69, 2019