Jökull - 01.01.2015, Page 55

Seismicity beneath Þeistareykir, NE-Iceland
0
4
8
12
16
20
D
ep
th
b
.s
.l.
[k
m
]
1 2 3 4 5 6 7
Seismic velocity [km/s]
VpVs
Tested profiles
SIL model
Krafla regional
VELEST model
0
4
8
12
16
20
D
ep
th
b
.s
.l.
[k
m
]
−17˚00' −16˚50' −16˚40'
Longitudea) b)
Figure 3. a) Regional seismic SIL velocity model used for automated event localization (Stefánsson et al., 1993)
and 1D P-wave velocity model obtained with VELEST using our data with Vp/Vs=1.78. b) Final hypocenter
locations (crosses) of selected earthquakes around Þeistareykir. Horizontal and vertical bars show the loca-
tion error estimated by calculating the hypocenter locations with either the SIL or VELEST velocity model.
Colored scatter points mark the sampled probability density function locations. Dashed lines represent the
68% error contours and the black dot the hypocenter with the highest location likelihood. – a) Þrjú hraða-
snið P- og S-bylgna. Gráskyggt svæði afmarkar þau hraðasnið sem prófuð voru til skjálftastaðsetninga. SIL
hraðasniðið (grænt) er notað af Veðurstofu Íslands. Til samanburðar er hraðasnið fyrir Kröflu (blátt) skv.
Schuler og fl. (2015) og VELEST hraðasniðið sem var notað. b) Nákvæmi í staðsetningum jarðskjálfta við
Þeistareyki. Krossar tákna mismun á skjálftastaðsetningum með VELEST og SIL hraðalíkönunum en brotalína
68% skekkjumörk 5000 staðsetninga sem reiknaðar voru með þéttifallsnálgun samkvæmt NonLinLoc forritinu.
Table 1. VELEST velocity model for Þeistareykir best fitting our limited dataset. Note that the 1D model is not
well constrained in the uppermost part (< 2 km) and below about 6 km depth below sea level (b.s.l.). An aver-
age Vp/Vs ratio of 1.78 was estimated from Wadati diagrams using the discussed 153 earthquakes. – VELEST
hraðalíkanið sem notað var til staðsetningar jarðskjálftanna. Líkanið er nákvæmast á 2–4 km dýpi.
Depth b.s.l. (km) Vp (km/s)
-0.70 1.63
0.00 3.35
1.00 4.64
2.00 4.90
3.00 6.16
4.00 6.51
6.00 6.67
9.00 6.85
20.00 7.30
tribution of recorded earthquakes. This is mainly the
case in the shallow part (< 2 km) and below 5–6 km
depth b.s.l.. A significant difference between our best
velocity model and the SIL model is observed at 2–5
km depth b.s.l., where higher velocities in our model
led to more robust solutions.
We further refined the locations using the travel-
time differences code hypoDD2.1 (Waldhauser and
Ellsworth, 2000) with their relative locations overlain
in Figure 1. Confirmation of correctly picked P- and
S-wave arrivals of the deep events was obtained by
constructing record sections (e.g. Figure 2).
JÖKULL No. 65, 2015 55