Jökull - 01.01.2010, Page 63

Upptyppingar seismic swarms
of 100 m (Rubin and Gillard, 1998). Also, the two
standard-deviation uncertainties in our relative reloca-
tions average to approximately 60 m. The RMS misfit
to our best-fit plane (i.e., 114 m) is greater than the
location uncertainty and earthquake rupture diameter;
hence, the distribution of our hypocentral locations
is amenable to some brittle fracture directly adjacent
to the dyke, perhaps representing the ambient stress
field that is near to failure. However, we do not ob-
serve widespread brittle deformation away from the
dyke. Additional improvements to the analysis pro-
cedure and station configuration consistently reveal
tighter clustering about the plane. We interpret this
as evidence for minimal and localised brittle defor-
mation surrounding the active dyke plane.
The tight spatial clustering revealed by the ASN
locations also indicates that the seismicity progressed
along the same dip throughout the study period. This
is consistent with recent theoretical and experimen-
tal models of Maccaferri et al. (2010) that suggest
it is energetically favourable for a dyke to continue
to propagate along the same dip in a homogeneous,
visco-elastic medium.
A detailed discussion of focal mechanisms is be-
yond the scope of this paper; however, we note that the
288 events consist of both reverse and normal fault-
ing in an approximately 3:1 ratio. Moreover, an over-
whelming majority of the events have inferred fault
planes that are in alignment with the plane of the dyke
(White et al., 2011).
Although the Upptyppingar seismicity’s tilted ori-
entation, manifestation in the ductile zone of the crust,
and concomitant dominance of high-frequency im-
pulsive P-wave arrivals is unusual, the Lake Tahoe
seismicity exhibited similar anomalous characteristics
(Smith et al., 2004). Linear regression models fit
the seismicity beneath Lake Tahoe to a single planar
structure dipping at 50◦ in visco-elastic crust.
Additional information regarding the presence of
melt may also be obtained from the ratio of P-wave
velocity to S-wave velocity (VP /VS), since the nature
of the host material controls wave propagation rates.
VP /VS ratios are derived independently for each of the
288 events using Wadati diagrams (Wadati, 1933) and
PPICK-determined arrival times. Wadati diagrams
display the time difference between the S-wave and
P-wave arrivals against the arrival time of the P-wave
for each station. For any given event, the slope of
the best-fit line to the plotted station data represents
the ratio of the P-wave to S-wave velocity minus one
(i.e., VP /VS –1).
As depicted in Figure 13, two statistically sig-
nificant populations emerge from the distribution of
VP /VS ratios with hypocentral depth: one at <15.25
km depth with a mean VP /VS ratio of 1.78±0.02 (2σ)
and another at 15.25–18.40 km depth with a mean
VP /VS ratio of 1.80±0.03. Here, statistically signif-
icant refers to a rejection of the null hypothesis that
the two populations are independent random sam-
ples from normal distributions with equal means but
not necessarily equal variances at the 1% significance
level, determined through a two-tailed t-test.
Waves produced by earthquakes at 15.25–18.40
km depth must pass through the overlying region of
lower VP /VS ; hence, it is possible to approximate the
local VP /VS ratio of the deeper region. Assuming a
two-layer model, with the average VP /VS ratio of the
upper layer shallower than 15.25 km being 1.78 and
a wave source at the centroid of the deeper cluster
(∼16.3 km), we estimate the local VP /VS ratio of the
deeper region to be ∼2.09. This is higher than that
expected or measured for typical, solid high magne-
sian gabbroic rocks at mid-depth in the Icelandic crust
(Christensen, 1996; Allen et al., 2002; Bjarnason and
Schmeling, 2009; Eccles et al., 2009); hence, it is con-
sistent with the presence of melt.
SUMMARY AND CONCLUSIONS
We have presented a test case that demonstrates the
benefits of a dense, local seismic network for resolv-
ing fine morphological details within compact seismic
swarms. However, we also show that high quality re-
gional networks, such as the SIL system in Iceland,
are more than sufficient for most interpretational re-
quirements. Furthermore, we conclude that the net-
work size and configuration dominate over the picking
of phase arrivals in determining location accuracy and
precision. This assumes, of course, that a reasonable
effort has been made in picking phase arrivals. The
IMO procedures as well as the techniques presented
JÖKULL No. 60 63