ArticlePDF Available

Abstract and Figures

Between June 2004 and September 2004 a temporary seismic network was installed on the northern insular shelf of Iceland and onshore in north Iceland. The seismic setup aimed at resolving the subsurface structure and, thus, the geodynamical transition from Icelandic crust to typical oceanic crust along Kolbeinsey Ridge. The experiment recorded about 1,000 earthquakes. The region encloses the Tjörnes Fracture Zone containing the Husavik–Flatey strike-slip fault and the extensional seismic Grimsey Lineament. Most of the seismicity occurs in swarms offshore. Preliminary results reveal typical mid-ocean crust north of Grimsey and a heterogeneous structure with major velocity anomalies along the seismic lineaments and north–south trending subsurface features. Complementary bathymetric mapping highlight numerous extrusion features along the Grimsey Lineament and Kolbeinsey Ridge. The seismic dataset promises to deliver new insights into the tectonic framework for earthquakes in an extensional transform zone along the global mid-ocean ridge system.
Content may be subject to copyright.
Abstract Between June 2004 and September 2004 a
temporary seismic network was installed on the
northern insular shelf of Iceland and onshore in north
Iceland. The seismic setup aimed at resolving the
subsurface structure and, thus, the geodynamical
transition from Icelandic crust to typical oceanic crust
along Kolbeinsey Ridge. The experiment recorded
about 1,000 earthquakes. The region encloses the
¨rnes Fracture Zone containing the Husavik–Flatey
strike-slip fault and the extensional seismic Grimsey
Lineament. Most of the seismicity occurs in swarms
offshore. Preliminary results reveal typical mid-ocean
crust north of Grimsey and a heterogeneous structure
with major velocity anomalies along the seismic lin-
eaments and north–south trending subsurface features.
Complementary bathymetric mapping highlight
numerous extrusion features along the Grimsey Line-
ament and Kolbeinsey Ridge. The seismic dataset
promises to deliver new insights into the tectonic
framework for earthquakes in an extensional transform
zone along the global mid-ocean ridge system.
Keywords Broadband seismology ÆMultibeam
bathymetry ÆHot spot-ridge interaction ÆIceland Æ
Ocean bottom seismology ÆTjo
¨rnes fracture zone
The north of Iceland represents in many ways an
extraordinary geological setting. In 1974, Sæmundsson
gave a first good tectonic and geological description of
a region where the Mid-Oceanic Ridge changes from
an offshore to an onshore ridge, separated by a trans-
form zone. However, this transform zone is leaky and
shortly after publishing the geological review of Sæm-
undsson (1974) a subaerial fissure eruption along the
Krafla fissure swarm (Fig. 1) between 1973 and 1983
revealed its volcanic potential. Strong earthquakes
above magnitude 7 make it the primary focus of high-
energetic seismicity in and around Iceland. Whereas
this produced high attention in the scientific commu-
nity its offshore situation in contrast to the onland
transition in the south Iceland Seismic Zone requires
considerably more effort to reveal its detailed tectonic
setup. With our study we try to contribute to this effort.
C. Riedel ÆT. Dahm ÆM. Hensch ÆR. Herber Æ
M. Schnese ÆM. Tho
¨len ÆB. Hofmann ÆS. Winter
Institut fu
¨r Geophysik, Bundesstr. 55,
20146 Hamburg, Germany
A. Tryggvason ÆR. Bo
¨ðvarsson ÆT. Eken
Institutionen fo
¨r Geovetenskaper,
Avdelningen fo
¨r Geofysik, Villava
¨gen 16,
75236 Uppsala, Sweden
B. Brandsdottı
Science Institute, University of Iceland,
¨skju, Sturlugata 7, 101 Reykjavı
´k, Iceland
R. Ste
´fansson ÆK. S. Vogfjord ÆS. Jakobsdottı
´rÆJ. Holmjarn
Physics Department, Meteorological Office of Iceland,
Bustadavegur 9, 150 Reykjavı
´k, Iceland
B. Sigurdsson
Marine Research Institute, Skulagata 4,
121 Reykjavı
´k, Iceland
Present Address:
C. Riedel (&)
Centro de Vulcanologia e Avaliac¸ao dos Riscos
Geologicos, Ponta Delgada, Ac¸ores, Portugal
Mar Geophys Res
DOI 10.1007/s11001-006-9007-0
First results from the North Iceland experiment
C. Riedel ÆA. Tryggvason ÆB. Brandsdottı
´rÆT. Dahm ÆR. Ste´fansson Æ
M. Hensch ÆR. Bo
¨ðvarsson ÆK. S. Vogfjord ÆS. Jakobsdottı
´rÆT. Eken Æ
R. Herber ÆJ. Holmjarn ÆM. Schnese ÆM. Tho
¨len Æ
B. Hofmann ÆB. Sigurdsson ÆS. Winter
Received: 25 April 2005 / Accepted: 7 August 2006
Springer Science+Business Media B.V. 2006
Whereas the Krafla system is still considered part of
the North Volcanic Zone (NVZ) of Iceland (Bjo
et al. 1977), which trends into O
¨rður bay, some
kilometers offshore in the western part of this bay two
islands—the Ma
´nareyjar islands Lagey (Fig. 1) and
´ey—tower above a north–south trending bathy-
metric ridge extending up to 6630 N. This ridge
probably featured a volcanic eruption at the end of the
19th century (1867/1868; see Thoroddsen 1925). The
bathymetry is dominated by north–south trending
troughs and ridges, which following the crest of the
ridges from the south to the north are always offset to
the west before finally joining with the central rift of
Kolbeinsey Ridge at around 67N (see Fig. 1). Kol-
beinsey Ridge shares most characteristics of the typical
Mid-Atlantic Ridge sections and arctic ridges. It is
slow-spreading (Vogt et al. 1980) and produces tho-
leiitic rocks with MgO content of 6–10 wt.% (Devey
et al. 1994), which is often termed a depleted, but still
normal mid-ocean ridge basalt (N-MORB).
20°W 19°W 18°W 17°W 16°W 15°
65° 00'N
65° 30'N
66° 00'N
66° 30'N
67° 00'N
OB31 OB32
Skjalfandi Trough
Fig. 1 Map of bathymetry and topography of North Iceland,
indicating the main historicalvolcanic centers Krafla and Askja,
the permanent stations of the Icelandic network (grey squares)
and the temporary stations (black triangles) of our experiment,
the shot locations (stars) and the two well-known seismic
lineaments Husavik–Flatey fault (HFF) and Grimsey lineament
(GL). Some seismic stations, islands and morphologic deeps
have been marked because they are further discussed in the text.
To allow for further notation, the depth contours offshore start at
200 m depth and are sampled in 100 m. The altitude contours on
land are sampled in 250 m
Mar Geophys Res
The alternating bathymetry north of the Icelandic
coast is a morphological result of en echelon normal
faulting (Sæmundsson 1974; McMaster et al. 1977;
¨gnvaldsson et al. 1998) following shearing between
the spreading regimes of Kolbeinsey Ridge and the
NVZ (Gudmundsson et al. 1993). Within the pull-apart
basins, one of them forming a major bathymetric low-
located just east of Grimsey (see Fig. 1), gaseous
pockmarks are observed (Richter et al. 2003). Their
number increases toward the major permeable faults
as, e.g. the Husavik–Flatey fault (HFF on Fig. 1). Fluid
migration along the faults also feeds high-temperature
hydrothermal fields (e.g. Botz et al. 1999; Riedel et al.
2001), which built hydrothermal smokers of more than
60 m height in Eyjafjordur (Stoffers et al. 1997). Small
volcanic cones of around 300 m base diameter and
smaller (Kuhn et al. revised) have been imaged along
the Grimsey Lineament (GL on Fig. 1), a seismic lin-
eament parallel to and north of the HFF. A major
submarine volcanic structure of around 8 km base
diameter is Storagrunn (Fig. 1) just north of 67N. This
structure has a characteristic shield volcano slope of 2
on average mapped by multibeam bathymetry surveys
by Brandsdottir et al. (2002,2004).
Between Askja volcano and Grimsey island the
transform zone is monitored by the permanent seismic
South Iceland Lowland (SIL) network (grey squares on
Fig. 1:Ro
¨gnvaldsson et al. 1998) and permanent GPS
stations (Geirsson 2002) from the Meteorological Of-
fice of Iceland. The crustal structure has been analyzed
by local earthquake tomography (LET) of the SIL
seismic travel-time data (Riedel et al. revised). The
results reveal a negative velocity anomaly reaching
down to 12 km depth along the main rupture of the
transform zone, the HFF. A further negative anomaly
between Flatey (Fig. 1) and Grimsey, covering a depth
to 12 km has been suggested to mirror an alignment of
magma bodies or a high heat-flow domain. However, in
the submarine parts the shallow subsurface cannot be
resolved by the permanent seismic network, and
therefore most parts of the GL is not imaged above
8 km.
A main point of discussion is the precise position of
the plate boundary. Gudmundsson et al. (1993) pro-
posed a model of two mode I cracks (i.e. extensional
zones) mirroring the two spreading domains of Kol-
beinsey Ridge and NVZ. In their model the North
American plate and Eurasian plate are separated along
the HFF close to the northern coastline of Iceland.
Taylor et al. (1994) proposed a structural boundary
along the GL, which they call an extensional transform
zone, i.e. along the complete boundary of North
American plate and Eurasian plate spreading occurs in
contrast to the Gudmundsson et al. (1993), where real
strike-slip transform motion occurs. Taylor et al. (1994)
compared oblique spreading regimes and extensional
transform zones worldwide and concluded that the
HFF is acting as a secondary fault, i.e. a Riedel shear,
with respect to the North America–Eurasia transforms
in a similar fashion as the San Andreas fault joins with
the Mendocino Fracture Zone in the Pacific (Hole
et al. 2000).
The purpose of this study is to analyze the transition
from Iceland to Kolbeinsey Ridge by a coupled seismic
and multibeam bathymetry survey. The seismic exper-
iment was aimed to improve hypocenter locations and
LET results by using ocean bottom seismometers
(OBS) and a temporal extension of the station setup on
land. The bathymetric data close some of the gaps in the
multibeam bathymetric map of the Tjo
¨rnes Fracture
Zone and provide a unique view on the bathymetry of
the southern end of Kolbeinsey Ridge.
In a joint effort by Hamburg University, Uppsala
University and the Icelandic Meteorological Office
between June 25 and July 8, 2004, the permanent
seismic network was expanded by 25 additional sta-
tions (see black triangles on Fig. 1). Of these 14 were
seismic ocean-bottom stations (OB30–OB43 see
Tables 1,2) and 11seismic land stations (U2–U12 see
Table 1 Positions of the temporary seismic network stations
Station Latitude Longitude Depth (–)/
Elevation (+) in m
OB31 6701.942N 1840.193W –312
OB32 6659.988N 1759.998W –438
OB33 6700.006N 1729.390W –274
OB34 6644.975N 1900.007W –489
OB36 6645.160N 1810.300W –370
OB38 6629.942N 1859.877W –485
OB39 6630.079N 1830.269W –157
OB40 6635.999N 1740.160W –242
OB41 6629.911N 1700.063W –238
OB42 6617.960N 1814.770W –107
OB43 6618.000N 1715.008W –154
U2 6559.238N 1804.320W 273
U3 6551.360N 1851.090W 275
U4 6545.558N 1751.318W 289
U5 6557.144N 1658.416W 290
U6 6536.060N 1837.512W 309
U7 6543.140N 1712.600W 159
U8 6604.098N 1842.558W 162
U9 6552.494N 1622.326W 321
U10 6534.854N 1742.366W 315
U11 6615.636N 1621.480W 67
U12 6617.790N 1706.942W 22
Mar Geophys Res
Tables 1,2). The ten short-period OBS from GeoPro
and four broadband ocean bottom hydrophones
(OBH) from the University of Hamburg were de-
ployed from R/V Drofn during its final cruise. The 11
seismic land stations from the University of Uppsala
were positioned between 19W and 16W and north of
6530¢N. One of them was installed on La
´gey Island
(see Fig. 1).
With the exception of U12 (Fig. 1) all stations were
up and running during an active source experiment
that was carried out using 16 shots of 22.8–45.6 kg
dynamite in the water column (see white stars on
Fig. 1). Compressors for large volume airguns are not
easily available on the sparsely populated North
Atlantic Island.
Technical details and positions for the seismic sta-
tions are given in Tables 1and 2. During the cruise R/
V Arni Fridriksson A 13, 2004 between September 6
and September 17, 2004, the OBS and OBH were
recovered from the seabed. Three stations (OB30,
OB35 and OB37, marked as white triangles on Fig. 1)
failed to deliver data, one because of a datalogger
drifting into saturation and the other two because
water penetrated the glass spheres, which covered the
seismometers. Due to power failure of the EarthData
logger, the station U11 on La
´gey only recorded
6 weeks of data. However, including the permanent
SIL network, recordings from up to 36 stations are
available for some of the bneraly 1,000 recorded events
during the period of operation.
After recovery of the seismic stations, the multi-
beam sounding equipment of R/V Arni Fridriksson
was used to expand the extensive multibeam survey of
the Tjo
¨rnes Fracture Zone started by Brandsdottir and
others in earlier cruises (Brandsdottir et al. 2002;
Richter et al. 2003). Particularly the region north of
67 N, in Eyjafjordur and surrounding Storagrunn vol-
cano was covered. A fine-scale morphology down to a
lateral resolution of some meters on top of Storagrunn
volcano was obtained. Analysis of multibeam
bathymetry data is particularly dependent on the
acoustic velocity profile in the water column. Therefore
11 conductivity–temperature–depth (CTD) profiles
were acquired during the survey (see Table 3) down to
a maximum depth of about 450 m.
The top and flanks of Storagrunn were dredged, but
during dredging of the suggested recent lava flows east
of the volcano, unfortunately the dredge was lost. The
rocks recovered were all allochthonous material,
probably beach gravel or glacially transported boul-
Data processing
Seismic data
The seismic data set is heterogeneous (Fig. 2). Data
was sampled at different sample rates on different
kinds of instruments (Table 2) and stored in different
formats ranging from miniSeed to internal formats.
The land stations (EarthData loggers) were synchro-
nized every half hour with the Global Positioning sys-
tem (GPS). GPS reception was unproblematic on
northern Iceland and implied that time drifts during
recording were negligible. The ocean bottom registra-
tion, however, need an additional step of preprocess-
ing. They are only communicating twice to a reference
time signal, in our case the GPS time, once when
deployed and once when recovered. In between, a drift
of the internal clock occurs. The seismic time series
data were resampled to account for this drift.
A fundamental difference in the corrections is evi-
dent for the SEND and Sedis III dataloggers (see
Table 2). SEND dataloggers sample a time sequence of
Table 2 The sensors used and their sample rates
Stations Datalogger Sensor Sample rate
OB30–OB33 SEND Geolon MLS OAS Hydrophone (one component: h) 50 Hz
OB34–OB43 GeoPro SEDIS III Hydrophone/Seismometer (four components: Z, N, E and h) 128 Hz
U2 EarthData logger Lennartz LE-1s (three components: Z, N and E) 100 Hz
U3–U12 EarthData logger Lennartz LE-5s (three components: Z, N and E) 100 Hz
Table 3 Map of CTD positions and deployment times and dates
CTD No Latitude N Longitude W Deployment time
298 6613.07 1740.85 21:46:01, 06/09/04
299 6645.17 1730.08 19:48:38, 07/09/04
300 6701.17 1755.65 23:22:47, 07/09/04
301 6653.67 1914.53 20:59:50, 08/09/04
302 6730.24 1859.67 02:19:41, 09/09/04
303 6730.08 1841.82 23:07:27, 09/09/04
304 6701.90 1824.82 21:20:47, 10/09/04
307 6615.84 1651.46 14:13:50, 11/09/04
308 6549.35 1807.76 21:17:12, 12/09/04
309 6553.93 1813.64 23:05:06, 12/09/04
310 6607.68 1829.50 01:12:35, 13/09/04
Mar Geophys Res
an additional internal quartz clock with a much faster
sample rate than the original data to account for the
non-linear internal clock drift. In this way the overall
drift can be associated with the non-linear drift,
whereas the Sedis III datalogger time was corrected
only by compressing the time axis linearly between
deployment and recovery datum using a floating point
sample rate. The maximum clock drift registered was
in the order of 1.7 s over the whole duration of the
experiment, which corresponds to 28 ms/day (i.e.
approximately four samples).
After clock drift removal, all data including the
waveforms of the detected events by the stationary SIL
network were converted to GSE format and a database
was constructed in SEISAN (Havskov and Ottemoller
2003). Phase readings from the automatic picking of
SIL were also converted to SEISAN S-files and the
associated waveforms were extracted in a narrow time
window ranging from just before the suggested origin
time to 2 min beyond the suggested last S pick in the
record. Finally, the data were manually repicked on all
stations, the permanent and the temporary ones, using
the SEISAN system. This study uses SIL arrival times
for the ~1,000 events recorded during the time period.
These were manually corrected and extended to picks
on the temporary network for all the data, which were
The seismic data acquired during the dynamite
explosions (Fig. 6) was preprocessed in the same
manner. The time series were rearranged to start at the
GPS time recorded during the explosion and uniformly
resampled to 100 Hz. These resampled time series
were converted to SEG Y format and analyzed using
Seismic Unix (Cohen and Stockwell 2003). For picking
of the first arrival, a Hilbert transform was used on the
data as, e.g. Earle and Shearer (1994), and the Butt-
werworth bandpass filtered amplitude component
(2–4–20–30 Hz) served as the pick dataset. Here, the
real seismic traces are shown.
The land station positions were taken from the GPS
logs of the instruments. The position was logged twice
a day, and the mean positions were used. In general,
2004 728 1527 50.8 L 67.215 -18.680 0.0 UHH 34 0.3
Plot start time: 2004 7 28 15:27 56.122
1183 -44
2 OB31 h
157956 1464
3 OB32 h
268 -471
4 OB38 h
9383 12
5 OB38 E
6305 0
6 OB38 N
9003 14
7 OB38 Z
864 -113726
328 50
9 U02 Z
0 5101520253035
589 55
10 U11 Z
2004 711 0441 50.2 L 66.361 -17.834 0.0 UHH 33 0.3
Plot start time: 2004 7 11 4:41 53.321
1354 -43
6547 4895
2 OB31 h
9793 1859
3 OB32 h
4 OB38 h
3284 12
5 OB38 E
2748 1
6 OB38 N
2123 14
7 OB38 Z
775 -114190
54 56 58 60 2 4 6 8 10 12 14 16
287 -11
9 U11 Z
Time in secondsTime in seconds
Fig. 2 Diagrams show the recordings of the seismic network on
a day with rough weather and sea conditions (left) and on a
quieter weather day (right). The stations have been marked in
Fig. 1, names are printed on the left side followed by component
where Zvertical component, Nand Eare horizontal component
and hrefers to a hydrophone
Mar Geophys Res
2004 8 4 1245 29.0 L
29 30 31 32 33 34
7 km
13 km
22 km
41 km
79 km
102 km
Time in seconds
1 OB38 N
2004 810 513 35.0 L
36 37 38 39 40
1 OB38 N
1 OB38 N
1 OB38 N
1 OB38 N
1 OB38 N
2004 8 1 5 2 9.0 L
2004 8 2 731 11.0 L
12 13 14 15 16
2004 8 1 239 20.0 L
21 22 23 24 25
2004 8 5 1811 15.0 L
15 16 17 18 19 20
Fig. 3 The raw data from the
North component of OB38 is
displayed in this graph, the S
pick has been marked. The
maximum of the S-wave
usually arrives later than the
pick, due to a multiple of
reasons stated in the text. The
delay of the maximum of the
S-wave to the pick varies with
distance. All time windows
displayed are 5 s long
Mar Geophys Res
Reduced Time in s (7.5km/s)
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Offset in km
Reduced Time in s (7.5km/s)
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Offset in km
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
0 10 20 30 40 50 60 80 90 100 110 120 130 140
Reduced Time in s (7.5km/s)
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Offset in km
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Offset in km
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Fig. 4 Reduced time-offset
sections of all recordings and
shots a north of the Grimsey
Lineament and b south of the
Grimsey Lineament both
filtered at 4–20 Hz. Circles
mark the picked onset of the
first arrival. Lines marks the
theoretical model fitted to the
Mar Geophys Res
the horizontal standard deviation was of the order of
3 m, and generally below 4 m vertically. The vertical
positions were also compared to a digital elevation
model with a resolution of 50 m (assuming the hori-
zontal positions were correct). They agreed to within –
4 m + –25 m (1r). The OBS and OBH positions were
determined as the mean position between deployment
and recovery locations. The largest differences are
about 0.1 min in latitude and 0.2 min in longitude, i.e.
about 50–100 m. In an attempt to improve the location
of OBSes from the order of 100 m to the order of 10 m,
in addition to the first arrivals, we tried to pick the
arrival of the water wave on the OBS recordings. This
would be equivalent to a small number of samples on
the time series, but would provide a result independent
of the subsurface seismic velocity model. Although the
water wave phases include the highest amplitudes in
the seismogram, it is often hard to determine the exact
onset on close inspection, so we restricted our pro-
cessing to the analysis above.
Multibeam data
The multibeam data were acquired using the MB sys-
tem software suite (Caress and Chayes 1995)ona
UNIX machine connected to a SIMRAD EM 300
instrument aboard R/V Arni Fridriksson. From time-
to-time, the sound velocity profile used for the inver-
sion of water depth was adapted to the closest spot
66° 00'N
66° 30'N
67° 00'N
P – Wave Velocity in km/s
Depth in km
Northern SIL
Depth in km
Depth in km
Depth in km
Fig. 5 (A) Diagram shows the compressional wave velocity
models used in this study. The solid blue line is the refraction
result north of the Grimsey Lineament (GL), the dashed line
south of this lineament. The same colors have been used for the
following plots (B–D), which display the location of hypocenters
using the according velocity model in 3D ((B) refraction model
north of GL, (C) LET model, and (D) SIL model). Further
details see main text
Mar Geophys Res
where CTD measurements were performed. These
CTDs were acquired using Seabird SBEII plus equip-
ment with an installed thermosalinograph SBE 16.
Further processing is still needed and a complete
bathymetric map will be published elsewhere
(Brandsdottir et al. 2002,2004). The preliminary re-
sults presented here were smoothed and interpolated
by E. Kjartansson (personal communication) at the
University of Iceland. An excerpt around Storagrunn
volcano is shown in Fig. 7.
20°W 19°W 18°W 17°W 16°W
65° 30'N
66° 00'N
66° 30'N
67° 00'N
2 km
4 km
2 km
6 km
Fig. 6 Isoline and grayscale
plot of the quality of locating
by the permanent network.
Darker colors refer to bigger
errors. The lateral error has
been used to number the
lines,small circles mark the
location of a shot by the
network, stars mark the real
position of the shot. The
permanent stations are
marked by gray squares
20°W 19°W 18°W 17°W 16°W 15°
65° 30'N
66° 00'N
66° 30'N
67° 00'N
5 10 15
0 5 10 15
Fig. 7 Isoline and grayscale
plot of the apparent
maximum depth of
hypocenters. This apparent
maximum depth refers to the
depth above which 80% of all
events are located. Each
event is displayed by a small
circle. The radius of this circle
is indicative of its depth. The
maximum depth of events is
found between Grimsey and
Flatey Islands
Mar Geophys Res
First results
Waveforms and picks
The acquired seismic waveforms are very similar
among the broadband instruments. They are subject to
weather and sea conditions. On rough days (see
Fig. 2a) a strong amplitude low-frequency component
(about 0.5 Hz) is visible (here on stations OB31, OB32,
SIG and U11 and less strong on GRI), on quiet days
(see Fig. 2b) seismic signals from the subsurface
dominate the time series (here on stations SIG and
U11 a small amplitude low-frequency component in
the overall signal is still visible due to some environ-
mental effect). The short-period instruments (as an
example OB38 in Fig. 2) are like high-pass filters
implying that no microseismic noise is evident. For
10% of the events their dynamic range is very low,
which is either a consequence of the noise or an elec-
tronic problem of the data logger.
Most P- and S-phases have an impulsive onset,
however the first arrival is not necessarily the same on
all components for the OBS data: a phenomenon that
was already recognized during the TYSEA experiment
(Dahm et al. 2002). This is a consequence of a ringing
effect produced by the low-S velocities in the shallow
underground just beneath the OBS (Thorwart and
Dahm 2003). More P-phases could be picked than
S-phases, but some S-phases are more impulsive than
the first arrival. In rare cases, only after Butterworth
bandpass filtering between 4 and 20 Hz (see above) the
signal was clear enough to provide a sufficient quality
picking (see Fig. 2). This is also indicated by cross-
correlations of the waveforms at one station of various
events within one swarm, which show a maximum
when a bandpass between 4 and 20 Hz is used.
The strongest amplitude arrival for the events is not
necessarily the S-wave onset (cf. OB38 on Fig. 2a).
There can be an amplitude response one order higher
than the S-wave onset, particularly on the hydrophone.
There are several origins for this high-amplitude arri-
val. For large-offset arrivals the SmS and Sn phase
interfere with the S-wave. For smaller offsets and short
time-delays an interference effect in the shallow sub-
surface near the seismometer may cause a delay effect.
Here, the effect is often strongest on the hydrophone,
thus it might be a wave converted into an acoustic wave
at the seabed, i.e. a tertiary or T-wave. Thorwart and
Dahm (2003) noticed that the horizontal and vertically
polarized S-waves might arrive at different times and
produce a similar effect (shallow shear-wave splitting).
An example is given for the North component of OB38
on Fig. 3, where the time delay changes from 1.4 s at
distances of 7 km to around 0.7 s between 40 and
80 km and back to 1.4 s at distances around 100 km.
This is a clear evidence that we do not simply see an S-
to-P conversion, but various effects interfere.
Visual inspection suggests that the picking error for
the P-wave onset on quiet days is in the order of 1–
5 ms, on rough days it increases to 5–10 ms. The
picking error for the S-wave is generally on the order
of 5–20 ms, once the actual onset phase can be estab-
lished. The time difference between onset and maxi-
mum amplitude may be in the order of seconds (cf.
U02 on Fig. 2b). This time-lag effect is typical for the
permanent stations GRI and FLA, where automatic
picking of the S-wave is often not in accordance with
the subsurface model even after correction of the S-
phase pick (see, e.g. GRI on Fig. 2b, where the
amplitude increases before the automatic S-pick).
Refraction seismics
The number and distribution of shots and stations is
not sufficient to evaluate the subsurface structure be-
low a single station or a single shot. Not only after the
analysis of P- and S-picks of the natural seismicity do
we realize that the region in study is very heteroge-
neous and bound to produce all sorts of seismic phe-
nomena. However, we can reconstruct a mean 1D
velocity model by averaging over two seismic sections
including all shots on all stations:
one north of the GL;
and the other south of the GL.
Because all shots were fired north of the HFF, the
different structure north and south of this structure
cannot be compared in a similar way by our survey
results. This trivial approach provides us with seimo-
grams of a median spatial sampling around 1 km
(Fig. 4). The first arrivals were picked (circles in Fig. 6)
and by assuming a simple 1D model, the refracted
arrivals using a ray-approach for homogeneous layers
were fit to the data.
In the reduced time section for the region north of
the GL (Fig. 4a) it is apparent that the resolution in the
20 km closest to the shot is insufficient. A linear trend
in the range of 3.2–3.5 km/s seems to receive the
dominant wave energy, mainly visible in later arrivals
between 4 and 5 s. Latest from 10 km on, travel time
picks indicate a velocity of 5 km/s. Between 20 and
65 km offset, a phase with the velocity of 6.7 km/s is
apparent. Beyond 50 km the best visible trend emerges
with a velocity of 7.2 km/s. Some picks in a distance
range from 80 to 120 km contrain a Moho velocity of
Mar Geophys Res
8.15 km/s. This is the not a unique way to interpret the
data, but this refraction model (blue solid line in Fig. 5)
fits well to other models in the region in study (e.g. the
LET model or the Kolbeinsey Ridge model of Hooft
et al. 2006). The location of the Moho and the seismic
velocity below the Moho remains disputable. We have
simply shown it here, because we want to highlight that
it is not impossible to detect a Moho north of the GL.
Many late arrivals in distance ranges between 40 and
80 km align with the picks on this refractor. To cut a
long story short there is no evidence for a transition
layer but rather a real Moho in north of Iceland.
South of the GL (Fig. 4b) the resolution seems to be
better in the first 20 km, but we are still unable to find a
direct wave and selected the same velocity as the 1D
LET of Riedel et al. (2005). In very decent offsets of
10 km and less, the refracted wave from a 5 km/s
horizon takes over the first break character and from a
distance of 30 km on the character of the picked first
breaks is pretty chaotic. Some of the late arrivals help
to fit a trend of 6.6, 7 and 7.3 km/s. The last phase is
only apparent as the first arrival after 120 km, so we
are dealing with a very slow crust in this case. This
model is shown as dashed blue line in Fig. 5. It would
be possible to align some phases between 110 and
140 km at a refractor velocity of 8.7 km/s, however, it
seems unreasonable that a refractor with such a high
velocity should appear and that it should receive so
much energy, when all the other phases clearly loose
energy beyond distances of 100 km. The restrictions
imposed by our small dataset prevent a more thorough
analysis of this phase. In any case such a high velocity
implies a dip of the refractor should it exist.
Travel times, root-mean-square residuals and 1D
velocity models
The first result of the survey was produced nearly in real
time by Gudmundsson, who improved the automatic
picks of the shots for the SIL stations and located the
events using the SIL velocity model and only the sparse
SIL setup (Fig. 6). The results demonstrate impressively
the performance and restrictions of the permanent
Icelandic network. While a near perfect reconstruction
of the shot epicenter (<2 km lateral error) is available
for the southernmost shots, a clear increase in error is
observed for the shots along 67N latitude and the shot
close to Flatey Island, i.e. station FLA.
As stated above, about 1,000 events have been re-
picked now including the whole temporary setup. They
have been located with Hypoellipse (Klein 1984;
Havskov and Ottemoller 2003). Only phase picks
within 90 km distance from the epicenter were evalu-
ated, because they do not include reflected or refracted
phases from the Moho (see refraction seismics). Only
picks with distances up to 30 km were weighted fully.
No weighting was assigned to the picks according to
individual pick quality. The epicenter was established
and the depth was determined by a grid-search for
minimum root-mean square travel time residual (from
here abbreviated as RMS residual) This procedure was
performed for four different velocity models (results
see Fig. 5), the standard model for the SIL network
based on Bjarnasson et al. (1993), a northern model
from the SIL network (G. Gudmundsson, personal
communication), the best fitting 1D LET model by
Riedel et al. (2005) and a model derived from our
refraction survey. All models (see Fig. 5) neglect the
water layer above the ocean-bottom and surrounding
the island stations.
Single errors for phases are commonly below 0.5 s,
but about 10% of the S-phases for some OBS stations
and SIL stations on the islands, FLA and GRI, cannot
be decreased to an RMS residual below 0.8 s.
The average RMS residual value for the whole
dataset is 0.274 (SIL), 0.255 (Northern SIL), 0.235
(LET) and 0.258 s (Refraction). Generally all models
provide a good solution. The hypocenter distribution is
very similar for the SIL and the Northern SIL model,
but completely different for the LET model. Whereas
events cluster in vertical bands between 5 and 15 km
for the SIL models (Fig. 5), most events occur in a
depth up to 3 km for the LET model (Fig. 7). How-
ever, just southeast of Grimsey a column of hypocen-
ters increases the depth, above which 80% of the
events are located (commonly interpreted as the brit-
tle-ductile transition), to about 15 km just below the
volcano-hydrothermal region (Fig. 7). Depths of the
seismicity cut-off of about 10 km are also characteristic
for the region north of Flatey.
Morphology and lithology
The overall morphology of the seafloor north of Ice-
land is well known, but it has never been available in
such a resolution. Richter et al. (2003) already high-
lighted the pockmark structures, Brandsdottir et al.
(2002) reported the bathymetric step associated with
the HFF in Skjalfandi Bay. Cone structures in a pull-
apart basin east of Grimsey were already noticed by
Kuhn et al. (revised). The frequency and distribution
of these cones has been described by Brandsdottir et al.
(2002,2004). Typical slope angles of ~30and crater
diameter to base diameter ratios fall well into the field
for tuffcone like structures (Vespermann and
Schmincke 2000). A number of surveys by Brands-
Mar Geophys Res
Fig. 8 Shows the interpolated multibeam bathymetry of (A) Kolbeinsey Ridge and (B) Storagrunn. The excerpts are marked on the central map. A profile and a preliminary
interpretation for Storagrunn between points a and b is found at the bottom right of the figure
Mar Geophys Res
dottir et al. (2002,2004) produced a multibeam bea-
hymetric map of the region.
This survey filled the gap between Holl and Stora-
grunn volcanoes (Fig. 8b). A ridge of cones splits into
two parallel rows of cones close to Storagrunn
enclosing a small basin. Following the apex of the basin
one can trace the center of the huge submarine volcano
just NNW of it. Before the survey the top of Stora-
grunn was unknown and was suggested to consist of a
crater valley. However, we were able to show that it is
composed of a ragged top with only a small crater to
the SSE, which is breeched to the SSW, probably by a
huge flow deposit. The volcano flanks compose a
horse-shoe of high flanks above a bathymetrically
lower segment in the SSE toward the apex of the
neighbouring basin.
Storagrunn was the main target for dredging. How-
ever, two dredge trials only revealed ice-rafted debris
and/or beach gravel. During dredge along the eastern
flank, which appeared to be formed recently, the
dredge was lost in the sea.
A further focus of multibeam work was the Kol-
beinsey ridge region between 19W and 1815¢W south
of 6730¢W. No ridge valley can be oberserved, but it
was observed that the bathymetry becomes shallower
toward the ridge. As well as the ridges of cones close to
Storagrunn (dashed lines on Fig. 7b), cones on Kol-
beinsey Ridge align with a strike of ~50NW (dashed
lines on Fig. 8a), which is subparallel to the more
southerly GL (Fig. 8a).
The newly acquired seismic data fit well to the past
analysis of permanent SIL seismic data in the same re-
gion (Ro
¨gnvaldsson et al. 1998;Riedeletal.2005), i.e. the
1D velocity model inverted by LET provides the smallest
RMS residual. All tested velocity models allow a less
erroneous locating process than the SIL model. The SIL
model was optimized for South Iceland and our results
indicate that the crustal structure from the north of Ice-
land differs from that in the south. However, it is obvious
from Fig. 5, that the decisive factor for a discrimination
of the models is the deep underground (>5 km), because
above the models are very similar. Less erroneous loca-
tions favor a model with a velocity of 7 km/s or faster
velocities from 10 km on.
The new locations cluster at lower depths, which
also means that the brittle-ductile transition is more
shallow than previously thought and the crust is thin-
ner. These observations do not only fit the new velocity
model, but also the old for South Iceland. So far, the
high depths of earthquakes could not be explained
easily. With the new data they disappear and do not
need to be explained.
The velocity model from refraction also agrees well
with these observations. South of the GL (Fig. 4b,
dashed blue line in Fig. 5) no velocity higher than
7.3 km/s can be confirmed as a first arrival. The het-
erogeneity north and south of the lineament as de-
picted by deviations of the phase readings from the
calculated arrival times using the best fit velocity
models is very similar (Fig. 4). The region sampled in
the south is much larger and some of the phases picked
do not seem to be the first break. In contrast to the
refraction model, the better LET 1D model was pro-
duced using events across the whole TFZ region, thus
including information from both sides of the GL and is
thus doomed to be better. Further more, the two
models. LET and Refraction, are very similar up to a
depth of 10 km (Fig. 5). The lower crust seems to
indicate a subtle increase of velocity with depth, which
prevents a first break from a Moho velocity around
8 km/s. In this case a LET will provide a much better
model then refraction with our sparse shot dataset, so
we do not want to discuss our models further.
We can also afford a glimpse further north. Our
refraction model there (Fig.4a, solid blue line in Fig. 5)
highlights a possible Moho in a bout 10 km depth with
a velocity around 8 km/s, further north along the
Southern Kolbeinsey Ridge the velocity model of the
KRISE experiment (Hooft et al. 2006), reaches a
velocity of 6.6–6.8 km/s at 3.5–4 km depth and the
Moho at around 10–12 km depth which compares well
with our 10 km. This might be evidence that there is a
transition here from Icelandic crust underlain by an
unknown, sometimes called transition, layer (Ange-
nheister et al. 1980) to typical oceanic crust underlain
by normal mantle material.
Our analysis shows that the SIL model still provides
a good approximation for the epicenter in routine
purposes. However, the depth of the sources differs
strongly between the SIL and the best LET model.
Most events seem to occur in a shallower subsurface
than previously thought and only few events in the
center of the research area occur deeper down (Fig. 7).
This is in good agreement with the negative flower
structure model (Woodcock and Fisher 1986) proposed
by Riedel et al. (2001) based on fluid geochemistry.
Having in mind, that the crust south of the HFF is not
unusual for Icelandic crust and that the crust north of
the GL is largely N-MORB, we conclude that the crust
in between was and is still formed by a pull-apart
process evoked by differential rifting of Kolbeinsey
Ridge and the NVZ. The active pull-apart structures
Mar Geophys Res
are those where seismicity is most frequent and where
fluids are migrating from the deep subsurface. The
deepest active structures we could image are located
along a line from Flatey to Grimsey in Skjalfandi
Although we might only look at a snapshot in time
of the seismicity, these results are supported by a LET
based on data from the permanent network by Riedel
et al. (2005), which show a low-velocity body in a depth
of 7–12 km between Grimsey and Flatey, where the
unusually deep seismicity is located. The presence of
such a low-velocity body will actually mean that even
the deep events found here are located in shallower
The deepest reaching and most permeable and thus
most probably most active fault of a flower structure is
usually considered to be the principal rupture of the
system (Barton et al. 1995). Here, this would refer to
the GL, where the seimicity aligned at depth reaches
its maximum. According to this hypothesis the rest of
the seismicity is related to the near-surface Riedel
shears at the edge of the pull-apart basin. Thus, the GL
is rather a Grimsey Seismic Zone (GSZ) including a
GL at its northeastern edge. In contrast to a classical
flower structure as proposed by Woodcock and Fisher
(1986), the main lineaments are not strike-slip faults
with a dip-slip components, subparallel to the main
rupture, but mainly dip-slip normal faults. Thus, our
results clearly favor an extensional transform zone
along the GSZ.
The high frequency of volcanic cones along the GL
in contrast to the relatively flat surface along the HFF
also suggests that volcanism and thus magmatic fluid
migration mainly occurs within the GSZ. The align-
ment of cones further north on Kolbeinsey Ridge
(Fig. 8), striking at a similar angle as the GL can only
be produced by two scenarios.
A strong influence of the transform shearing stress
regime to the north, which produces strike-slip fail-
ure, or
– An asymmetric spreading where cones are trans-
ported obliquely away from the ridge center. This
would imply that the Jan Mayen microplate to the
east of Kolbeinsey Ridge is a growing plate, where
more new crust is produced to the north of the plate
to account for the oblique spreading.
In either case we find the influence of the transform
regime still at a latitude of 6730¢N, far away from the
Icelandic coastline. This implies that the crustal struc-
ture is changing smoothly toward the north and no
sudden shift occurs at the GL.
We observe a transition from Kolbeinsey mid-ocean
ridge crust to Icelandic crust along the GL, the
northern limit of a GSZ. Whereas a mantle refraction
could be identified for a region north of the GL, there
is no confirmed mantle refractor south of it. The crust
for this northern region is thinner than in Iceland and
maybe even thinner than in the adjacent Kolbeinsey
Ridge, though only slightly. South of the GL, there is a
gradual increase in compressional wave velocity with
depth starting at 4 km depth, implying that even at an
offset of 140 km from the shot, there is no first break
with a mantle velocity. A later arrival cannot be ruled
out but could not be observed due to a low signal to
noise ratio. In between the GL and the HFF, a main
area of seismicity is located between Grimsey and
Flatey Island running from north to south. Along this
narrow line, the deepest hypocenters were recorded,
whereas above it hypocenters spread out across the
entire region in study, mostly clustering in north–south
trending lineaments close to the surface. The Skjalf-
andi trough/basin complex resembles a classical nega-
tive flower structure as in the Woodcock and Fisher
(1986) model. No petrological evidence for existing
volcanism could be found from dredging, but multi-
beam bathymetry highlights some rough structures that
appear to be recent. The opening of the flower struc-
ture produces a sharp contrast at the HFF with an
origin in the stress regime of the GSZ, which can be
traced up to 6730¢N and implies a gradual change in
crustal structure toward Kolbeinsey Ridge.
Acknowledgments We thank Captains Ragnar G. D. Her-
mansson, and Gudmundur Bjarnasson and their crews for
excellent support during the marine surveys. We ackowledge
financial support by the Deutsche Forschungsgemeinschaft
(DFG) under grants Da 478/13-1 and Ri 1220/2-1, the Swedish
Vetenskapsradet, the Icelandic Research organization and the
Leitstelle fu
¨r mittelgrosse Forschungsschiffe for taking over the
costs for the scientific cruises. Further more we are grateful for
the access to instruments of the Geophysical Instrument Pool
Potsdam, and for the shot permission and for the research
permission to the Icelandic Research organization (RANNIS).
C. Riedel also wants to thank Reinhard Wineberger at the
german embassy in Reykjavik for unbureaucratic and fast
Angenheister G, et al. (1980) Reykjanes Ridge Iceland Seismic
Experiment (RRISP 77). Z Geophys 47:228–238
Barton CA, Zoback MD, Moos D (1995) Fluid flow along
potentially active faults in crystalline rock. Geology 23:683–
Mar Geophys Res
Bjarnasson IT, Menke W, Flovenz OG, Caress D (1993)
Tomographic image of the Mid-Atlantic plate boundary in
southwestern Iceland. J Geophys Res 98:6607–6622
¨rnsson A, Saemundsson K, Einarsson P, Tryggvason E,
¨nvold K (1977) Current rifting episode in north Iceland.
Nature 266:318–323
Botz R, Winckler G, Bayer R, Schmitt M, Schmidt M, Garbe-
Schonberg D, Stoffers P, Kristjansson JK (1999) Origin of
trace gases in submarine hydrothermal vents of the Kol-
beinsey Ridge. Earth Planet Sci Lett 171:83–93
Brandsdottir B, Detrick R, Driscoll N, Kent G (2002) Pilot study
of the Tjo
¨rnes fracture zone, offshore northern Iceland,
using high-resolution multichannel seismic reflection profil-
ing and CHIRP sonar. In: Proceedings of the AGU fall
Meeting, San Francisco, 2001
´r B, Richter B, Riedel C, Dahm T, Helgadottı
Kjartansson E, Detrick R, Magnusson A, A
´sgrimsson AL,
´lsson BH, Karson J, Saemundsson K, Mayer L, Calder B,
Driscoll N (2004) Tectonic details of the Tjo
¨rnes Fracture
Zone, an onshore-offshore ridge-transform in N-Iceland.
Eos Trans AGU, 85(47):F1071
Caress DW, Chayes DN (1995) New software for processing
sidescan data from sidescan-capable multibeam sonars. In:
Proceedings of the IEEE Oceans 95 conference, San Diego,
Cohen JK, Stockwell JW Jr (2003) CWP/SU: Seismic Unix re-
lease 37: a free package for seismic research and processing.
Center for Wave Phenomena, Colorado School of Mines
Dahm T, Thorwart M, Flueh ER, Braun T, Herber R, Favali P,
Beranzoli L, D’Anna G, Frugoni E, Smiriglio G (2002)
Ocean bottom seismometers deployed in Turrhenian Sea.
EOS Trans 83:309–314
Devey C, Garbe-Schonberg CD, Stoffers P, Chauvel C, Mertz
DF (1994) Geochemical effects of dynamic melting beneath
ridges: reconciling major and trace element variations in
Kolbeinsey (and global) mid-ocean ridge basalts. J Geophys
Res 99(B5):9077–9095
Earle PS, Shearer PM (1994) Characterization of global seis-
mograms using an automatic picking algorithm. Bull Seis-
mol Soc Am 84(2):366–376
Gudmundsson A, Brynjolfsson S, Jonsson MT (1993) Structural
analysis of a transform fault-rift zone junction in North
Iceland. Tectonophysics 220:205–221
Geirsson H (2002) Continuous GPS measurements in Iceland
1999–2002. MSc thesis, University of Iceland
Havskov J, Ottemoller L (2003) SEISAN: the earthquake anal-
ysis software, Version 8.0 preliminary, University of Bergen,
Hooft E, Brandsdottir B, Mjelde R, Shimamura H, Murai Y
(2006) Asymmetric plume-ridge interaction around Iceland:
the Kolbeinsey Ridge Iceland Seismic Experiment. Geo-
chem Geophys Geosyst 7:Q05015; doi:10.1029/
Hole J, Beaudoin BC, Klemperer SL (2000) Vertical extent of
the newborn San Andreas fault at the mendocino triple
junction. Geology 28:1111–1114
Klein FW (1984) User’s guide to HYPOINVERSE, a program
for VAX and PC 350 computers to solve for earthquake
locations, USGS open file report-84-000
Kuhn T, Riedel C, Herzig PM, Theilen F (revised) Shallow 3d
structure of the Grimsey Lineament offshore North Iceland.
Mar Geol
McMaster RL, Schilling J-GE, Pinet PR (1977) Plate boundary
within Tjo
¨rnes fracture zone on northern Iceland’s insular
margin. Nature 269:663–668
Richter B, Driscoll N, Detrick R, Fornari D, Brandsdottir B
(2003) Recently discovered near-shore gas-charged sedi-
ments and pockmarks, northern Iceland. In: Proceedings of
the AGU fall. Cite abstracts as: Eos Trans AGU, 84(46),
Fall Meet Suppl Abstract AGUFMOS51B0854R
Riedel C, Schmidt M, Botz R, Theilen F (2001) The grimsey
hydrothermal field offshore North Iceland: crustal structure,
faulting and related gas venting. Earth Planet Sci Lett
Riedel C, Tryggvason A, Dahm T, Stefansson R, Bo
¨dvarsson R,
Gudmundsson GB (2005) The seismic velocity structure
north of Iceland. J Seismol 5:383–404
¨gnvaldsson ST, Gudmundsson A, Slunga R (1998) Seismo-
tectonic analyses of the Tjo
¨rnes fracture zone, an active
transform fault in north Iceland. J Geophys Res 103:30117–
Sæmundsson K (1974) Evolution of the axial rifting zone in
northern Iceland and the Tjo
¨rnes fracture zone. GSA Bull
Stoffers P, Botz R, Garbe-Schonberg G, Hannington M, Hauzel
B, Herzig P, Hissmann K, Huber R, Kristjansson JK, Pe-
tursdottir SK, Schauer J, Schmitt M, Zimmerer M (1997)
R/V Poseidon, Cruise report 229a. Institut fur Meereskunde,
Universitaet Kiel
Taylor B, Crook K, Sinton J (1994) Extensional transform zones
and oblique spreading centers. J Geophys Res
Thoroddsen T (1925) Die Geschichte der islandischen Vulkane,
K. Dan. Vidensk. Skr., 9. Raekke, Naturvidensk. Mat. Afd.,
8, Copenhagen
Thorwart M, Dahm T (2003) Evaluation of ringing-phases on
ocean-bottom seismometer, EGS-AGU-EUG joint assem-
bly. Nice, France, 6–11 April 2003
Vespermann D, Schmincke H-U (2000) Scoria cones and tuff
rings. In: Sigurdsson H, Houghton B, McNutt S, Rymer H,
Stix J (eds) Encyclopedia opf volcanoes. Academic Press,
New York, pp683–694
Vogt P, Johnson GL, Kristjansson L (1980) Morphology and
magnetic anomalies north of Iceland. J Geophys 47:767–780
Woodcock NH, Fisher M (1986) Strike-slip duplexes. J Struct
Geol 8:725–735
Mar Geophys Res
... The North ICeland Experiment (NICE) was a temporary deployment of on-land and offshore seismological instruments to record data simultaneosly with the SIL network during the summer of 2004. The main purpose was to resolve the subsurface structure of the TFZ and study the transition from the Icelandic crust to more typical oceanic crust near the southern end of the Kolbeinsey Ridge (Riedel et al., 2006). In addition, to the seismological deployment, bathymetric mapping was performed, improving the resolution of previously available data. ...
... Sixteen explosions of 22.8-45.6 kg dynamite were fired in the water column and recorded (Riedel et al., 2006). They provided ray-paths in areas of the northern TFZ that are seismically quiet. ...
... A waveform database was created with data from the NICE and the SIL networks recorded during the simultaneous deployment, in order to facilitate a joint analysis and phase picking. Continuous records of the NICE stations, after correction of the OBS data for clock drift (Riedel et al., 2006), were converted from the GSE (Global Seismic Exchange) to the SEISAN format (Havskov & Ottemöller, 1999). Independently, waveforms in the SIL catalog of earthquakes located in Northern Iceland were converted from SIL to SEISAN format. ...
Full-text available
Local earthquake tomography has been carried out in the Tjörnes Fracture Zone. This transform region connects the Mid‐Atlantic Ridge to the Northern Volcanic Zone in Iceland in a mostly offshore area. The challenge to record seismic information in this area was the motivation for the North ICeland Experiment (NICE). Fourteen ocean‐bottom seismometers and eleven on‐land stations were installed in the project and operated simultaneously with the permanent Icelandic seismic network (SIL) during summer 2004. Data from the experiment were used to estimate P‐ and S‐wave crustal velocities. Also, the gravity anomaly was derived for comparison with the tomographic results. Upper‐crustal velocities are found to be relatively low in the offshore region. In particular, low velocities are mapped along the Húsavík‐Flatey Fault, where a more confined negative gravity anomaly and a sedimentary basin are found. Low velocities are also mapped along the Grímsey Oblique Rift and in a zone connecting these two main lineaments. The northern half of the aseismic Grímsey Shoal appears as a fast anomaly. Furthermore, localized high‐velocity anomalies are found beneath northern Tröllaskagi and Flateyjarskagi Peninsulas, where bedrock dates from Upper and Middle Miocene (10–15 Ma). Regions of low Vp/Vs ratios are mapped at depth along the main lineaments. Low velocities along the lineaments are interpreted as due to fracturing extending into the middle crust, while high velocities in the upper crust beneath Tertiary formations are associated with relic volcanoes. Low Vp/Vs variations along the lineaments are interpreted as due to the presence of supercritical fluids.
... Placing faults in this confined region on basis of measurements have important implications, as the location of the most recent large earthquake in the region, the 1910 w 7.1 earthquake, is largely unknown but given its reported effects in the northern region, it has been placed in the centre of the TFZ (see the approximate location of the large event, between mid-HFFZ and Grímsey island, in Fig. 1). Then, the North Iceland Experiment project saw the use of fourteen ocean-bottom seismometers in addition to a land-based densification of seismic stations over a few months in 2004 (Riedel et al. 2006). This enabled the detailed analysis of microseismicity on the basis of empirical travel times and local tomography modeling Abril et al. 2021). ...
Full-text available
Proceedings of the NorthQuake 2022 workshop on Earthquakes in North Iceland, held in Húsavík, North Iceland, 19-23 October, 2022.
... 1). Then, the North Iceland Experiment project saw the use of fourteen ocean-bottom seismometers in addition to a land-based densification of seismic stations over a few months in 2004(Riedel et al. 2006). This enabled the detailed analysis of ...
Full-text available
A strategy is proposed to incorporate effects of 3-D velocity variations on earthquake locations using empirical traveltimes (ETTs). Traveltime residuals are interpolated from those predicted using a 1-D velocity reference, mapped on to the hypocentres of corresponding earthquakes for each station in a network. First, station corrections are computed by averaging the residuals over a fixed scale. Then, summary-ray variograms are used to estimate uncertainty and that, in turn, is used to tune a local fit to neighbouring residuals to refine the corrections. Resulting 3-D traveltime estimates are then used as a description of the forward problem in a nonlinear grid-search relocation. Data are weighted according to the estimated uncertainty. Data from the Icelandic Southern Iceland Lowlands (SIL) national seismic network are used to test the strategy. ETTs are estimated for P- and S-waves at 65 stations in the SIL network, based on four million arrival time readings of 300.000 events registered between 1990 and 2012. ETTs are strongly correlated for the two wave types. The spatial variations of the predicted corrections are consistently comparable or somewhat less for S-waves than P-waves. This feature suggests variations of the VP/VS ratio in the Icelandic crust. Error estimates are also strongly correlated for the two wave types and between nearby stations. Relocations are tested by comparison with explosions and small populations of well-located events within denser subnetworks. Relocations result in modestly enhanced clustering of explosions and earthquakes and significantly improved depth estimates. Estimates of the random relocation error are statistically better behaved than those of the SIL system. They are in general reduced, as is expected since 3-D heterogeneity is now partly taken into account.
Full-text available
Tectonic earthquake swarms (TES) often coincide with aseismic slip and sometimes precede damaging earthquakes. In spite of recent progress in understanding the significance and properties of TES at plate boundaries, their mechanics and scaling are still largely uncertain. Here we evaluate several TES that occurred during the past 20 years on a transform plate boundary in North Iceland. We show that the swarms complement each other spatially with later swarms discouraged from fault segments activated by earlier swarms, which suggests efficient strain release and aseismic slip. The fault area illuminated by earthquakes during swarms may be more representative of the total moment release than the cumulative moment of the swarm earthquakes. We use these findings and other published results from a variety of tectonic settings to discuss general scaling properties for TES. The results indicate that the importance of TES in releasing tectonic strain at plate boundaries may have been underestimated.
The Tjörnes Fracture Zone separates Iceland's North Volcanic Zone from Kolbeinsey-Ridge (Greenland-Sea). Seismicity mainly occurs in swarms, often 100 and more per day, with similar waveforms and frequently offshore.We analysed three earthquake swarms between June and September 2004, recorded by 35 stations, i.e. the permanent Icelandic SIL network and a temporary setup of land and ocean-bottom-seismometers, and 5 swarms of the years 1994-1997 only using the sparser SIL network. Events occurring in the same swarm often show similar waveforms at the same station. We cross-correlated these time series by using a new approach of three component cross-correlation in order to relocate the hypocenters relative within the swarms and to precisely determine the direction and velocity of migration. Our method delivered good relocations with small spatial and temporal errors. This allows the interpretation and characterisation of the observed earthquake swarms. We try to classify observed migration velocities by comparing them to typical fluid- and crack-propagation velocities and determine focal-mechanisms and orientation of the best fitting plane through the hypocenter distribution.We separate the investigated events into two types of earthquake swarms, supposedly dike-induced and hydrothermally- or gas-induced swarms, by pointing out typical characteristics of both types and by comparing them to similar events of other volcanic regions. Based on different migration velocities, we will discuss possible mechanisms and their triggers of all single clusters within a swarm. Hypothetic models will be established, trying to explain the processes during the swarm episodes and to derive possible pre-eruptive patterns from the character of seismicity.
A long-range seismic refraction experiment is described which was realized in 1977 along an 800km long line across Iceland and along the sutheastern flank of Reykjanes Ridge. The main purpose of the experiment was to resolve the structure of the crust and upper mantle to greater depth than previously possible and to study the transition frm the oceanic to the Icelandic structure. Shots, both on Iceland and at sea, were recorded by up to 90 stations from Iceland, Germany, and the Soviet Union and by seven ocean-bottom starions from Germany and Canada. In addition, detailed marine seismic investigations of the Reykjanes Ridge have been carried out from board RV METEOR.-Authors
Extract In seismology, a wealth of data-processing programs are available, and a seismic observatory typically makes use of several. A common problem is the lack of a proper database structure which prevents effective use of the data, particularly with data from different sources. A primary goal of SeisAn is to organize data from all kinds of seismic stations into a simple database and to provide most of the tools needed for routine processing. The SeisAn database is a way of organizing data by using the directory and file structure, and is not a relational data system with general access techniques such as SQL, which usually is meant by the term “database.” In addition, the intention is to facilitate research tasks by integrating additional programs to work directly on the database. A second goal is that SeisAn must work in an identical manner under Sun (Solaris-1 and -2), Linux, and MS-Windows, and...
In July, 2001 we conducted a high--resolution geophysical survey of the Tjörnes Fracture Zone (TFZ), offshore Northern Iceland, using two high--resolution seismic profiling systems -- a SUBSCAN Chirp/Side Scan Sonar and a portable High--Resolution Multichannel Seismic System. This was the first time these two systems had been used together and this nested approach proved to be very successful. Our principal scientific objectives were:\ (1) to define the current location and history of strike--slip movement along the offshore portion of the Húsavík --Flatey (HFF) fault; and (2) to estimate the amount of extension across the Eyjafjardaráll graben (the southern extension of the Kolbeinsey Ridge), Skjálfandi Bay, and Öxarfjördur basin in order to determine how strain has been accommodated in the TFZ over the past 6\ Ma. Within the glacially--eroded and sediment--filled graben on the N--Iceland shelf, the Chirp typically penetrated up to 30--40\ m subbottom; HiRes MCS data was able to extend this stratigraphy to 300--500 m depth. The uppermost Holocene section was characterized by a 10--25\ m thick transparent layer of silts and muds overlying a highly reflective, jökulhlaup--derived, turbidite sequence 5--10\ m thick of late glacial or early post--glacial age inferred to be deposited by jökulhaups, which buries an older reflective surface interpreted as glacial diamictites. Several thin, high--amplitude reflectors occurring within the transparent layer may represent tephra layers identified in nearby cores . The thickest 'transparent' Holocene section is in the Öxarfjördur basin with a progressively thinner Holocene section in the Skjálfandi and Eyjafjardaráll basins. Numerous faults were mapped in both the MCS and Chirp data, some of which extend to the seafloor, suggesting these graben are actively extending. The HFF is marked by a prominent 10--15\ m high, south facing scarp between Flatey Island and ~17o36'W where it intersects a N--S trending basement ridge. East of this point the HFF fault appears to step to the north, where its expression is much more subdued. We were unable to find the offshore extension expression of the faults that have been mapped on--land into the town of Húsavík. Side scan sonar data collected along the HHF west of 17o36'W show numerous, lineated shallow depressions in the surficial sediments on the upthrown block near the fault which may be caused by focused fluid expulsion along the fault.
Technical Report
The Icelandic Meteorological Office operates a network of continuous GPS stations (ISGPS). The network was initiated as a collaborative project in 1999, to monitor crustal movements in active tectonic and volcanic areas in Iceland. There are presently 17 continuous GPS stations in Iceland, of which 14 belong to the ISGPS network, two are IGS stations and one is operated by the National Land Survey of Iceland. The design of the ISGPS network is aimed towards simplicity, robustness and cost-efficiency. The number of electric components in the field is minimized and we use a stainless steel quadripod monument design to achieve high monument stability. Data from the ISGPS stations are automatically downloaded and processed on a daily basis. We use the Bernese V4.2 software to process the data. The data are initially processed using predicted satellite orbits, and then reprocessed with CODE final orbits. The time series from most ISGPS stations are dominated by motion caused by plate spreading, in general agreement with the NUVEL-1A plate motion model. Discrepancies are observed at stations which are within the plate boundary deformation zone or close to volcanic deformation sources. Transient signals attributed to an eruption in Hekla in February 2000, are observed. The nearest station, located 50 km from Hekla, recorded 7 mm horizontal motion towards Hekla during the eruption. Time series from stations located near Katla volcano indicate there is a slow pressure increase beneath the volcano. Two magnitude MW=6.5 and MW=6.4 earthquakes spaced 17 km apart occurred on June 17 and June 21, 2000, in the South Iceland Seismic Zone. Although most of the ISGPS stations were not located close to the epicenters, a clear deformation signal was detected at all operational stations. The coseismic displacements for the June 21 event fit well to a source model based on network GPS measurements. The observed displacements for the June 17 event include deformation from triggered events on Reykjanes Peninsula. The ISGPS network has proven a valuable tool to monitor crustal deformation and timing deformation events.
The Tjornes Fracture Zone (TFZ) links the northern rift zone (NVZ) in Iceland with the Kolbeinsey Ridge north of Iceland. The TFZ was initiated during the Miocene (about 7 Ma), following an eastward jump of the spreading axis in northern Iceland. A roughly 150 km long (EW) and 50 km wide (NS) deformation zone has since developed incorporating both right-lateral movement along WNW-trending strike-slip faults and oblique extension (105°) within three major N-S trending grabens (from west to east the Eyjafjardaráll, Skjálfandi and Oxarfjordur basins). Recently collected EM300 and RESON8101 multibeam bathymetric data, and CHIRP subbottom data combined with onshore mapping have enhanced our understanding of the rift-transform interactions within the TFZ. The transform motion is incorporated within two seismically active WNW trending zones, the Grímsey Seismic Zone (GSZ) and the Húsavík-Flatey fault (HFF), spaced ~40 km apart along the margins of the extensional basins. Being the propagating continuation of the NVZ offshore the GSZ has both the characteristics of an oblique rift zone and a transform whereas the HFF is more akin to oceanic transform faults. Four left-stepping, en-echelon, NS-striking rift segments (volcanic systems) exist along the GSZ. Large GSZ earthquakes, however, seem to be mainly associated with lateral strike-slip faulting along WNW-striking fault planes. Fissure swarms transecting the offshore volcanic systems also indicate right-lateral strike-slip motion parallel to the spreading direction. The HFF has an overall strike of N65°W and can be traced continuously onshore and offshore along its 75-80 km length, between the NVZ, across Skjálfandi and into Eyjafjardaráll. Four pull-apart basins occur along the fault, the largest at the intersection with Eyjafjardaráll, the southward but magma-starved, continuation of the KR. Tertiary dikes, parallel to the HFF indicate it has been a leaky transtensional feature. The southwestern margin of the fault is characterized by NE-striking lavas which dip steeply (30-50°) towards Eyjafjardaráll. The lavas are dissected by en echelon arrays of conjugate strike-slip faults intersecting the HFF fault at angles of N20°-30°W and N20°E. Some can be traced onto land where they exhibit complicated flower patterns. Destructive earthquakes occurred on the HFF in 1755, 1867 and 1872. The 1867 events were most likely associated with rift-transform interaction within the Mánáreyjar volcanic system, similar to the 1975-1989 Krafla rifting episode, when a lateral intrusion event triggered a M6.5 strike-slip earthquake at the junction of the Krafla fissure swarm and the GSZ. Although transform motion within the TFZ is currently taken up by two parallel fault systems the Tjornes microplate will merge with the North American plate as continued northward propagation of the divergent plate boundary gradually deactivates the extensional basins and HFF.