Jökull - 01.01.2010, Qupperneq 47

Reviewed research article
Dense seismic network provides new insight into the 2007
Upptyppingar dyke intrusion
Hilary R. Martens1, Robert S. White1, Janet Key1, Julian Drew1,
Heidi Soosalu1,2 and Steinunn S. Jakobsdóttir3
1Bullard Laboratories, University of Cambridge, Madingley Road, Cambridge CB3 0EZ, UK
2Geological Survey of Estonia, Kadaka tee 82, Tallinn 12618, Estonia
3Icelandic Meteorological Office, Bústaðavegur 9, 150 Reykjavík, Iceland
hilarymartens@gmail.com, rsw1@esc.cam.ac.uk, ajk65@cam.ac.uk, drew2@slb.com, heidi@hi.is, ssj@vedur.is
Abstract — Factors such as network geometry, network size and phase-picking accuracy have significant
effects on the precision of seismic hypocentre locations. In turn, the precision of the hypocentral locations
dictates the degree to which morphological details within seismic swarms may be resolved. The Icelandic
national seismic network (SIL) is designed to monitor seismic activity across large expanses of Iceland in real-
time using automated earthquake detection and location software. Here we examine the performance of the
SIL network relative to a much denser, local network of seismometers deployed around the Askja volcano in
the Northern Volcanic Zone. A subset of earthquakes from the 2007–2008 dyke intrusion beneath Mt. Upptypp-
ingar is used to compare single- and multi-event hypocentral locations. Specifically, we highlight 288, high
signal-to-noise ratio events that occurred during an intensive sequence of earthquakes from 6–24 July 2007,
when the temporary Askja network was active. A careful refinement of phase onsets recorded by our well-
configured, dense network of receivers reveals hypocentres clustered tightly on a planar structure, interpreted
as a dyke dipping at 49◦. The root-mean-square (RMS) misfit to the plane (114 m) is only slightly greater than
the uncertainties in relative locations of the earthquakes themselves, and constitutes a three-fold reduction in
RMS misfit over SIL relative locations. The improved precision, facilitated predominantly by a more favourable
network size and configuration, permits a more detailed analysis of the intrusion.
INTRODUCTION
Between February 2007 and April 2008, intensive
episodes of micro-seismic (local magnitude (Ml)
< 2.2) activity were recorded near Mt. Upptypping-
ar in the northwestern part of the Kverkfjöll volcanic
system in Iceland’s Northern Volcanic Zone. A map
of the study area is shown in Figure 1. The seismicity
occurred primarily in swarms of tens to hundreds of
earthquakes clustered tightly in space (within tens
to hundreds of metres) and time (within minutes to
hours), and delineates a southward dipping planar
structure as shown in Figure 2. The spatial coherency
of the hypocentres, supplemented by surface deforma-
tion recorded by continuous GPS, led Jakobsdóttir et
al. (2008) to propose that the seismicity was caused
by melt intrusion in a dyke.
The Upptyppingar swarms locate at depths of 13–
19 km below sea level, which is well beneath the lo-
cal brittle-ductile boundary at 6–7 km depth (Soosalu
et al., 2010) and above the estimated Moho depth at
25–30 km (Darbyshire et al., 2000). Despite their ori-
gin in ductile crust, however, the earthquakes exhibit
frequency content similar to tectonic earthquakes at
shallower levels. Brittle failure in ductile crust re-
JÖKULL No. 60 47