Jökull - 01.01.2017, Qupperneq 18

Microearthquakes on the eastern flank of Katla volcano, S-Iceland
sian (Figure 9). We note that the differential measure-
ments applied in relative location are generally more
precise than absolute phase picks and inherently min-
imize the potential effects of velocity structure bias.
Trade-off between focal depth, origin time and veloc-
ity may also affect the absolute locations of Jeddi et al.
(2016), but this is difficult to evaluate. We estimated
the relative error based on the waveform correlation
coefficient. However, the error of differential mea-
surements based on correlation also depends on the
characteristics of the waveform at each station (Sgat-
toni et al., 2016b). We have not taken this dependence
into account as it would require an extensive synthetic
study. Furthermore, we have assumed a laterally uni-
form velocity model in our relative locations. This
may cause bias that can either be avoided by apply-
ing a robust 3-D velocity model or by solving for the
wave geometry at each station together with relative
location as Sgattoni et al. (2016b) did.
Figure 9. Histogram of residuals derived by relative
location after the final inversion (132 events of fam-
ily 1). – Súlurit sem sýnir tímafrávik eftir afstæðar
staðsetningar allra skjálfta í flokki 1.
With limited information about the focal mech-
anisms of the eastern events we can only speculate
about their physical cause. We can, however, exclude
glacial processes. Icequakes are observed in glacial
streams worldwide (e.g., Roux et al., 2008; O’Neel
and Pfeffer, 2007), also in the temperate glaciers
of Iceland, e.g. in Skeiðarárjökull (Roberts, 2005;
Magnússon et al., 2005). However, these occur pre-
dominantly in association with floods or rain, while
the two known biggest eastern-event swarms occurred
in November and December. More importantly, we
can robustly distinguish the events’ focal depths from
the surface.
We have established that the events are small
(ML ≈ 0), of VT character and very clustered lat-
erally. They are also clustered in depth and their aver-
age depth is about 3.5 km. The magnitude range and
VT character suggest a brittle fault dimension of tens
of meters (assuming constant stress drop scaling (e.g.,
Abercrombie, 1995)), i.e. smaller than the dimensions
of the clusters. The absolute depth is not robustly de-
termined. We can speculate about tectonic, volcanic
or hydrothermal origin. The events isolation in space
clearly argues against the underlying cause being a
distributed stress field in the upper crust. Therefore,
a simple tectonic explanation is unlikely. Besides, the
events occur in an area outside the rift, which is not
expanding on a short time scale in this area (LaFemina
et al., 2005). The clustering of the events suggests a
local force field. This might associate with a localized
pressure or buoyancy source, i.e., fluids. The fluid
might be magma, hydrothermal brine or gas. How-
ever, the VT character of the events suggests brittle
failure while volcanic sources involving fluids are of-
ten associated with long-period (LP) events (Chouet,
1996).
The magma candidate leads us to speculate about
an ascending cryptodome, such as Soosalu et al.
(2006) suggested as an explanation of events beneath
Goðaland, on the western flank of Katla. Felsic ex-
trusives are found on the eastern flank near Sandfells-
jökull. However, the Goðaland events are LP events,
but swarms of VT and hybrid events have been regis-
tered elsewhere in association with dome rising, e.g.,
at Montserrat (Miller et al., 1998).
Hydrothermal fluid may become pressurized due
to a local heat source or due to changes in permeabil-
ity that circulates fluids by a heat source. A pressure
increase would lower the effective normal stress on
fractures and faults, thereby decreasing their stabil-
ity. It may also induce fracture dilation or even cause
tensional fracturing, affecting permeability and effec-
JÖKULL No. 67, 2017 13