Jökull - 01.01.2010, Síða 60

Martens et al.
6543210
19
18
17
16
15
14
13
Distance along azimuth 175 [km]
D
e
p
th
[k
m
]
06Jul07
08Jul07
10Jul07
12Jul07
14Jul07
16Jul07
18Jul07
20Jul07
22Jul07
24Jul07
Figure 10. SIL-system double-difference relative locations fit to a plane using linear regression. A dashed, black
line depicts the best-fit plane and the view is looking along the strike direction (Table 2). The RMS misfit to
the plane is 293 m. – Upptök jarðskjálfta samkvæmt afstæðum staðsetningum úr SIL-kerfinu og besta nálgun
við tvívíðan flöt. Svarta strikalínan sýnir halla flatarins. Horft er eftir strikstefnunni (85◦). Frávik skjálftanna
frá fletinum (RMS) er nú 293 m (tafla 2).
DISCUSSION
It is well known that factors such as network size and
station distribution affect the accuracy of hypocen-
tre locations (e.g., Bai et al., 2006; Bondár et al.,
2004). Table 3 summarises the disparities between the
ASN and SIL network sizes and geometrical config-
urations. The azimuthal gap provides a quantitative
measurement of the extent to which events have oc-
curred within the network. Smaller azimuthal gaps
indicate denser networks with a preferable geometri-
cal configuration.
Mean horizontal and vertical absolute location er-
rors determined from all 288 single-event locations
and as output by HypoInverse-2000 are included for
comparison. Consistent with the double-difference
relative location uncertainties, ASN single-event er-
rors (±0.3 km and ±0.5 km for horizontal and verti-
cal errors, respectively) are improved by 40–50% over
SIL single-event errors (±0.5 km and ±1.0 km). Sta-
tion distribution likely plays the main role as a com-
parison of locations based on IMO picks versus our
picks for the same six stations yields similar results
(i.e., ±0.8 km and ±1.0 km for the horizontal and ver-
tical errors, respectively, for the same six stations but
different time picks; see also Figure 7).
Particularly after relative relocation, SIL does a
remarkable job considering the sparseness of the net-
work. The orientation of the dyke plane is clear
and migration patterns within swarms can be dis-
tinguished. One reason for the high quality perfor-
mance of the SIL network relative to the denser, local
ASN network is that the SIL system is not as sparse
60 JÖKULL No. 60