Jökull - 01.01.2010, Page 36

Brandsdóttir et al.
components, in addition to a few purely normal events
with similar N5–15◦E strike. A majority of the focal
mechanisms and distribution of aftershocks thus indi-
cate N-S aligned right lateral movement along the two
major fault zones, although focal solutions along the
E-W oriented epicentral zone further west may be in-
dicative of strike-slip motion on both N-S and E-W
oriented faults.
A lower hemisphere equal area projection of the
T axes (Figure 9) shows a strong horizontal NW-
SE alignment (130◦–140◦), with an average of 138◦.
Similarly, plots of the P axes show a NE-SW trend
of steeply dipping as well as horizontal axes associ-
ated with normal and strike-slip events, respectively.
These directions are consistent with both strike-slip
and normal earthquakes being generated by oblique
movement with respect to the 103◦ spreading direc-
tion from NUVEL-1A (DeMets et al., 1994). The
NW-SE direction is close to being perpendicular to
volcanic fissure swarms within the Reykjanes Penin-
sula Rift Zone and Western Volcanic Zone (Clifton
and Kattenhorn, 2006). Mapping of Holocene surface
faults within the SISZ (Einarsson, this issue) shows
individual N-N20E aligned fault systems to be made
up of a series of NE en echelon fault strands (Figures
1 and 5) indicative of variable obliquity of the book-
shelf faulting along the SISZ.
DISCUSSION
Aftershock occurrence frequency
A histogram of earthquake occurrence in one hour
bins (Figure 11) shows aftershock frequency decay,
interrupted by an increase in seismicity May 31–June
1, June 2–4, 6–7 and 8–9. Although wind records
from local weather stations (Ingólfsfjall and Eyrar-
bakki) reveal a clear inverse correlation between the
number of earthquakes recorded and wind strength
the increase in seismicity in early June is also ob-
served in the number of CMM located events with
higher signal-to-noise ratio as well as the number of
events Mlw ≥3 located by the SIL network (Figure 2
and black stars in Figure 11). The earthquake activity
thus deviates from the empirical Omori’s Law, which
states that aftershock activity should decay exponen-
tially with time (Utsu, 1965).
A fault reaches steady state over time, when slip
rate has decreased to the tectonic loading rate. As
the aftershock rate does not simply scale with stress
rate, reloading of faults by afterslip can trigger after-
shocks over various time periods, thus adding com-
plexity to the rate-and-state dependent friction law
and making it difficult to infer the mechanisms re-
sponsible for earthquake triggering on the basis of ob-
servations of stress changes (Helmstetter and Shaw,
2009). Although it is likely that the Reykjafjall event,
which occurred within 3s of the Ingólfsfjall event (De-
criem et al., 2010), was generated by near-field dy-
namic triggering as during the June 2000 events (Ant-
onioli et al., 2006), faults within the SISZ and RPRZ
are subjected to both dynamic and static triggering
(Árnadóttir et al., 2004). An increase in aftershock
seismicity on May 31–June 1, June 2–4, 6–7 and 8–9
is thus most likely caused by short-term static stress
buildup on adjacent faults due to abrupt changes in
upper crustal pore pressure. A relatively fast visco-
elastic response of the lower crust may also play a
role. Pore-pressure oscillations can affect two-phase
flow within geothermal areas. Some of the triggered
seismicity originated within geothermal areas north
of Hveragerði (Grensdalur) and at the junction of the
SISZ with the southern Hengill Rift Zone.
A clear cyclicity in the number of aftershocks with
a 24 hour periodicity is obvious in the CMM located
events. No tidal forcing effect is observed. In order to
establish the exact period of the cycle a Fourier Trans-
form of the histogram was produced showing a peak
of 1.001 days which means that the cycle of increase
and decrease in aftershock frequency is, to within a
minute or so, precisely one day. The wind records
show the wind force being generally lower during the
night hours. The permanent stations used by SIL are
buried in vaults and thus not as sensitive to wind and
cultural noise as the surface installed temporary sta-
tions which were also located adjacent to some of the
major roads in the region (Figure 5). The 24 hour cy-
cle, with about ten times fewer events detected during
the daytime, is most likely due to higher cultural back-
ground noise during the day, causing fewer events to
be detected by the CMM program’s threshold SNR
value.
36 JÖKULL No. 60