Sie sind auf Seite 1von 11

Earth and Planetary Science Letters 373 (2013) 118128

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters


journal homepage: www.elsevier.com/locate/epsl

Hydrothermal seismicity beneath the summit of Lucky Strike volcano,


Mid-Atlantic Ridge
Wayne C. Crawford n, Abhishek Rai 1, Satish C. Singh, Mathilde Cannat, Javier Escartin,
Haiyang Wang, Romuald Daniel, Violaine Combier 2
Institut de Physique du Globe de Paris, Sorbonne Paris Cit, Univ. Paris Diderot, UMR 7154 CNRS, F-75005 Paris, France

art ic l e i nf o a b s t r a c t

Article history: We present two years (July 2007July 2009) of earthquake locations from a local seismological network
Received 12 December 2012 on the Lucky Strike volcano (Mid-Atlantic Ridge), whose summit hosts one of the most active and largest
Received in revised form deep-sea hydrothermal elds known. Two clusters of small (ML o 0.8) but continuous seismicity with
18 April 2013
well-dened lower depth limits extend north and south along-axis from the hydrothermal eld. The
Accepted 19 April 2013
lower limit of the northern cluster (79% of events) is a few hundred meters above the axial magma
Editor: P. Shearer
Available online 30 May 2013 chamber (AMC) reector, whereas the lower limit of the southern cluster (7% of events) is 600 m
shallower. During a 3-month long event swarm in AprilJune 2009, the northern cluster's lower limit
Keywords: deepened by 50100 m and two other event clusters were activated: one to the east of the hydrothermal
Mid-Atlantic Ridge
vent elds (5% of events) and another beneath the elds (2% of events). We interpret the continuous
hydrothermal circulation
background events in the northern and southern clusters as adjustments within a narrow weak axial
marine seismology
axial magma chamber region due to stress created by thermal contraction events at the bottom of the hydrothermal circulation
volcano zone or in the AMC. We interpret the swarm as an episode of hydrothermal penetration toward the AMC
that may have been activated by a broader, AMC-level, contraction event. We propose that a narrow axial
region of lower upper crustal porosity, combined with the sloped top of the AMC, generates along-axis
hydrothermal circulation towards the vent elds. The hydrothermal circulation may in turn modify the
AMC topography through enhanced cooling. The AMC reector's complex topography suggests that the
volcano is in a waning phase in which the AMC has already been signicantly altered since the latest
magma emplacement.
& 2013 Elsevier B.V. All rights reserved.

1. Introduction phases, with the tectonic phases dominating the extension and
creating pathways for uid ow from a stable deep heat source.
The morphology of the Lucky Strike segment of the slow- Currently, high- and low-temperature hydrothermal outows
spreading Mid-Atlantic Ridge, like that of many slow-spreading surround a lava-lled depression at the volcano summit (e.g.,
oceanic ridge segments, suggests a combination of tectonic and Barreyre et al., 2012; Fouquet et al., 1995; Langmuir et al., 1997;
magmatic accretion processes (Phipps Morgan et al., 1987; Humphris et al., 2002; Ondras et al., 2009) and an axial magma
Sempere et al., 1993; Tapponnier and Francheteau, 1978). chamber (AMC) reector lies 3.03.8 km beneath the volcano
A clearly-dened axial valley, bounded by normal faults, is deepest (Combier, 2007; Singh et al., 2006). The AMC reector overlies a
at the segment ends and shallowest at the segment center, where low-velocity zone that spans the lower crust and appears to be cut
it culminates in a volcano that is capped by recent lava ows and short to the east by the deep-cutting eastern axial valley bounding
cut by numerous faults (Ondras et al., 1997, 2009; Parson et al., fault (Seher et al., 2010a; Crawford et al., 2010).
2000; Scheirer et al., 2000). Humphris et al. (2002) proposed that We present here two years of seismicity recorded beneath Lucky
the Lucky Strike volcano undergoes distinct volcanic and tectonic Strike volcano, using two consecutive one-year deployments of ve
ocean-bottom seismometers (OBS) around the volcano (Fig. 1).
The spatial and temporal distribution of seismicity on mid-ocean
n
Correspondence to: Equipe Gosciences Marines, Institut de Physique du Globe ridges provides a window into subsurface tectonic, hydrothermal and
de Paris, 1 rue Jussieu, 75238 Paris Cedex 05, France. Tel.: +33 6 5151 1054. magmatic processes. Microearthquakes can be generated by rupture
E-mail addresses: crawford@ipgp.fr, along faults or by cracking and stresses due to uid circulation
crawford.wayne@gmail.com (W.C. Crawford).
1
Present address: Department of Geosciences, University of Oslo, Norway.
(Lockner and Byerlee, 1977; Gudmundsson et al., 2002; White et al.,
2
Present address: TOTAL, 2 Place Jean Miller, La Dfense 6, 92078 Pairs La 2011) and magmatic ination (Wilcock et al., 2009). Microearthquake
Dfense, France. studies on mid-ocean ridges have revealed hydrothermal circulation

0012-821X/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.epsl.2013.04.028
W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128 119

cracking fronts (Sohn et al., 1998a, 1999; Golden et al., 2003), associated with reactivation of the hydrothermal elds (Langmuir
background uid ow (Stroup et al., 2009), stresses surrounding et al., 1997).
magmatic events (Sohn et al., 1998b; Tolstoy et al., 2008), stresses The summit depression lies within an approximately 0.4 km
caused by ination of the axial magma chamber (Wilcock et al., wide graben that extends to the northern end of the volcano and
2009), the brittleductile transition (Kong et al., 1992) and the shape appears to have split a peak north of the depression in two
of deep-penetrating faults that may be associated with a long-lived (Fig. 1B). We refer to this feature as the central graben. Just south
hydrothermal system (deMartin et al., 2007). Most of the hydro- of the summit depression, the central graben is overprinted by a
thermal studies were made on the fast-spreading East Pacic Rise or more recent peak, which is itself heavily faulted. A wider graben
the intermediate-spreading Juan de Fuca Ridge, where the ridge axis to the west splits the main volcanic edice from a western
and axial magma chamber are approximately linear, and the only volcanic ridge. The freshest lavas observed on submersible dives
study at a slow spreading ridge imaged a fault (deMartin et al., 2007). are within the summit depression and at the northern end of the
The weaker magma supply, stronger tectonism and heavy segmenta- western graben (Ondreas et al., 1997).
tion at slow-spreading centers provide an end-member case to test A seismic reection/refraction study of the Lucky Strike seg-
which aspects of hydrothermal circulation are common to all ment and volcano identied an axial magma chamber (AMC)
spreading ridges and which depend on the spreading rate. reector 3.03.8 km beneath the volcano seaoor (Combier,
2007; Singh et al., 2006). The AMC reector extends 23 km
across-axis and 47 km along-axis. Around the hydrothermal eld
2. Lucky Strike volcano and the south peak, the reector spans the distance between the
western and central grabens, but it narrows signicantly to the
Lucky Strike volcano is the central edice of the 65-km long north and south before disappearing (Fig. 1B). A low-velocity zone
Lucky Strike segment (Fig. 1), which splits a regional topographic (LVZ) spans the crust beneath the reector (Seher et al., 2010a).
high created by the Azores hotspot (Detrick et al., 1995; Cannat The LVZ is wider than the AMC reector but is abruptly cut off near
et al., 1999). The segment's axial valley is approximately 3000 m the east edge of the volcano, near the bottom of the eastern valley
deep at the segment ends and 2000 m deep at the segment center wall's subsurface fault trace (Singh et al., 2006), suggesting that
(excluding the volcano). The full spreading rate is 20.3 mm/yr on the fault may play a role in cooling the lower crust (Seher et al.,
a 1031 azimuth (NS1-NUVEL2 model (Gripp and Gordon, 1990)). 2010a).
The Lucky Strike volcano lies on the eastern half of the axial valley Detailed imagery and in situ observations of the hydrothermal
and rises 400 m above the valley oor. The volcanic edice is heavily eld and surrounding areas (Barreyre et al., 2012) reveal extensive
faulted, with predominantly along-axis orientation. The summit is diffuse venting whose estimated heat ux (1871036 MW) is much
marked by three 100-m high, heavily faulted peaks surrounding a higher than the ux from high-temperature vents (850 MW). Most
depression containing recent lava ows (Fig. 1B). The summit depres- of the hydrothermal outows lie along or near fault scarps,
sion is marked by extensive collapse structures, has almost no suggesting that the faults channel rising hydrothermal uids near
sediments and is surrounded by high- and low-temperature hydro- the seaoor.
thermal vents (Langmuir, 1993; Fouquet et al., 1995; Humphris et al.,
2002; Barreyre et al., 2012). Lavas in the depression cover all but the
most recent faults (Ondras et al., 2009), and abundant relict sulde 3. Data acquisition and processing
deposits and debris suggest a long hydrothermal history (Ondreas
et al., 1997; Humphris et al., 2002). The presence of very young lavas in We deployed ve free-fall ocean-bottom seismometers (OBSs)
the summit depression, with pristine glass showing no alteration, around the summit of Lucky Strike volcano in July 2007, as part of
suggests a recent resurgence of volcanic activity that may have been the MOMAR (MOnitoring the Mid-Atlantic Ridge) program.

Fig. 1. Overview maps. (A) Segment scale. The solid contour outlines Lucky Strike volcano. (B) Volcano scale. Solid contours are isodepths outlining the volcano base (1900 m
beneath sea level (mbsl)), the western volcanic ridge (1660 mbsl), the summit depression (1710 mbsl) and its three surrounding peaks (1660 mbsl). Red (thick) dashed line
marks the bounds of the Axial Magma Chamber (AMC) reector (Combier, 2007; Singh et al., 2006). Black dashed lines mark signicant surface faults (Combier, 2007). Black
crosses mark hydrothermal vents and underlying grey circles mark diffuse ow regions (Barreyre et al., 2012). Circles mark OBS sites (blue (outer) 20072008, red (inner)
20082009, letters are site names). The dashed box marks the bounds of Figs. 3 and 5.
120 W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128

We recovered the instruments in August 2008 and redeployed ve Table 1


other OBSs for another year. All of the OBSs had four components: Station corrections for each site (x1 20072008, x2 20082009).
three orthogonal ground motions plus pressure. For each deploy-
Site Minimum 1D model Starting model
ment, four of the instruments were short period (using Mark
Systems L-28 geophones and a HiTech hydrophone) and one was P-delay S-delay P-delay S-delay
broadband (using a Trillium T240 seismometer and a differential
pressure gauge). We picked arrivals on the geophone/seismometer N1 0.01 0.12 0.12 0.33
W1 0.17 0.25 0.01 0.09
channels data, whose frequency bandwidth was 4.525 Hz. C1 0.09 0.03 0.00 0.15
The high frequency limit at 25 Hz is imposed by the sampling rate E1 0.05 0.08 0.08 0.17
of 62.5 samples/s and implemented by a linear phase Finite S1 0.08 0.05 0.11 0.31
Impulse Response lter in the OBS digitizer (Cirrus CS5321/22). N2 0.12 0.02 0.18 0.04
W2 0.09 0.21 0.12 0.11
The low frequency limit at 4.5 Hz is imposed by the geophone and
C2 0.00 0.03 0.09 0.03
we ltered the broadband seismometer data to match the geo- E2 0.04 0.13 0.08 0.04
phone's pass band, using a 2-pole Butterworth high-pass lter at S2 0.02 0.02 0.00 0.10
4.5 Hz. The array aperture was 8 km in 20072008 and 7 km in RMS 0.08 0.12 0.10 0.17
20082009 (Fig. 1).
The OBS clocks were synchronized on deployment and recovery
and a linear drift was assumed in between. The nal clock offset was
veried after the earthquakes were located by plotting the change in the volcano (1900 m below sea level) and the inversion code
time residuals over time. We identied a 1 s nal offset in one of the calculated optimal station corrections for each site. The starting
20082009 instruments (probably due to a misread nal synchroniza- model was based on a velocity prole beneath the volcano summit
tion). We corrected the data and picks for these offsets and veried calculated from seismic refraction data (Seher et al., 2010b).
that they optimized the time residuals. The nal velocity model has lower velocities than the starting
We selected events using a short-term average/long-term model and reduces RMS travel-time residuals from 0.096 to
average algorithm (e.g., Allen, 1982), then manually picked 0.054 s. The deepest velocities return to those in the starting
P- and S-wave arrival times and amplitudes. The data quality model because of the lack of data constraints at these depths. We
was signicantly better in 20082009 than in 20072008, result- were also able to reduce the travel-time residuals to 0.060 s
ing in more picked events and smaller location uncertainties. The without changing the starting model but instead adding much
difference in data quality is probably due to better coupling to the stronger S-wave station delays (Table 1): the hypocenter distribu-
seaoor in the later deployment: the 20072008 deployments tion is similar for the two solutions. We chose to use the minimum
were the rst year-long deployments of these OBSs, and extra 1D model because of its smaller RMS residual and station
oatation was tied to the OBS frames to compensate for the extra corrections.
batteries. In 20082009 the oatation was rigidly attached the OBS As with any seismology experiment using only surface instru-
frames. ments, hypocenters are better constrained laterally than in depth.
We calculated a one-dimensional velocity model (Fig. 2) using Most importantly, any error in the velocity model can create a
the VELEST software and the minimum 1D model procedure systematic error in hypocenter depths that will not be reected in
(Kissling et al., 1994), in which we iteratively searched for the the formal uncertainties. We tested the effect of a range of velocity
model with the smallest deviation from the starting model that models whose hypocenter uncertainty residuals were within 20%
stably minimizes the travel-time residuals. Because interfaces with of our minimum 1D model: they changed the hypocenter depths
large velocity jumps tend to focus hypocenter depths, we used a by up to 1 km without signicantly changing the x/y positions or
large number of thin layers and searched for the smoothest model the clustering.
satisfying the above conditions. We set zero depth at the base of We calculated earthquake local magnitudes using the custom
scale
0 M L log 10A 1:306log 10r 0:0346r2:49;

where A is the amplitude in nm and r is the distance in km.


This scale ts the local attenuation of earthquake amplitudes
1 observed in our data and matches the reference local magnitude
St

scale (Richter, 1935; Hutton and Boore, 1987) at a reference


ar
Fi

tin

distance of 1.5 km. We chose this relatively small reference


na

g
l

distance to minimize the effect of the stronger attenuation in


Depth (km)

2
our study area than in the reference model (cf., Bohnenstiehl et al.,
2008).
We rst calculated hypocenters and magnitudes from the
3 manual P- and S-wave picks and amplitude measurements using
the HYPOCENTER program (Lienert et al., 1986; Lienert and
Havskov, 1995). We calculated uncertainties using the standard
4 HYPOCENTER algorithm, which uses the posterior covariance
matrix and a 90% condence interval. The algorithm located
1175 events within our network (358 in 20072008 and 817 in
20082009) and we collected the 809 hypocenters whose
5 maximum location uncertainty was 1.0 km or less (154 events in
2 4 6 8
20072008, 655 in 20082009). The rejected events were gener-
Velocity (km/s) ally small, but two of them had ML 4 0.5: a magnitude 0.6 event
Fig. 2. Velocity model used to locate the earthquakes (solid line) and starting beneath the south peak on 23 August 2008 and a magnitude
model (dashed line, from Seher et al., 2010a, 2010b). 1.7 event south of the south peak on 29 May 2009. The position
W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128 121

preferred to use the HypoDD results, which are based on the same
1D model as the HYPOCENTER locations and which offer much
32

32
17' greater improvement in event locations.

16'
0 1 2 3 4
With only ve stations and no geographical orientation of the
Hypocenter horizontal components, we could not reliably calculate focal
654 locations 1 mechanisms.

8'
371 0

4. Event hypocenters
1
We plot event hypocenters as a function of time (Fig. 4) and
7'
371 space (Fig. 5). Most of the events were between 2 and 3 km
beneath the seaoor and inside of two on-axis clusters: one

Distance along axis (km)


HypoDD, corr+picks: 2007 2008 2009
607 locations 1 Oct Jan Apr Jul Oct Jan Apr Jul

Events/day Cum moment (109 Nm) Cum events


1
8' 600
371 0 0
1
1.5
400

1 200

7' 0
371
1000
800
600
HypoDD, corr only: 400
555 locations 1
SWARM
200
0
8'
371 0 8
6
1 4
2
7'
371 0 1 2 3 4 0
32

32

2
Depth beneath seafloor ref (km)
18'

17'

1
ML

Fig. 3. Comparison of hypocenters provided by different inversion methods. Left: 0


Map view. Right: Along-axis cross-section. (A) Manual picks located using HYPO-
1
CENTER. (B) Manual picks and phase correlations located using HypoDD. (C) Phase
correlations only, located using HypoDD.
Depth bsfr (km)Dist acrossaxis (km)Dist alongaxis (km)

uncertainties of the selected events were 600 7160 m, 1


480 7150 m and 665 7190 m for 20072008 and 5357125 m, 0
465 7 120 m and 650 7190 m for 20082009, for X, Y and Z
1
respectively.
We next relocated hypocenters for the 20082009 data using
the HypoDD program (Waldhauser, 2001) and the cross-
1
correlation between waveforms of different events at the same
station. We used a cross-correlation threshold of 0.8 and rejected 0

correlation-based pick offsets greater than 0.5 s. We could not 1


relocate the 20072008 data because the waveform cross-
0
correlations were too low to be useful. We tested HypoDD with
the cross-correlation data plus the absolute (catalog data) and 1
using only the cross-correlation data (Fig. 3). Running HypoDD on
2
the absolute and cross-correlation data improves the resolution
while retaining the structure indicated by the HYPOCENTER 3
results, with a loss of 7% of the events. Running HypoDD on only
Oct Jan Apr Jul Oct Jan Apr Jul
the cross-correlation data further improves the resolution while
2007 2008 2009
losing 8% more of the events. From here on, we show the HypoDD
locations calculated using only cross-correlation data. 555 of the Fig. 4. Seismicity as a function of time. Grey symbols are results obtained using
HYPOCENTER on 20072009 data, black symbols are results obtained using
655 20082009 events are located, with relative position uncer-
HypoDD on 20082009 data. The circle size scales with the local magnitude. The
tainties of 157 20 m, 20 740 m and 30 755 m in X, Y and Z dashed vertical line marks the division between the 20072008 and the 20082009
respectively, calculated using the Singular Value Decomposition experiments, the grey background marks the 2009 event swarm. (A) Cumulated
method (Waldhauser, 2001). number of events. (B) Cumulated seismic moment. (C) Events per day (3-day bins).
We also tried locating the events using a 3D model based on (D) Local magnitude. (E) Distance along-axis, referenced to the center of the
volcano summit depression. Inset box shows mean uncertainties for HYPOCENTER
the models of Seher et al. (2010a, 2010b) and Arnulf et al. (2011, (grey) and mean uncertainties plus one standard deviation for HypoDD (black). (F)
2012), but the inversions did not signicantly improve the mists Distance across-axis, same reference and uncertainty bars as in F. (G) Depth
or locations compared to the HYPOCENTER results. We therefore beneath the seaoor reference (bsfr: the reference is 1.9 km below sea level).
122 W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128

Depth below seafloor reference (km)


0 1 2 3

A B

Distance along axis (km)


1

ML
-1 -1
0
1
1.5

AMC depth (km beneath seafloor reference)


0
C D
Depth below seafloor reference (km)

3.2

3.4

3.6

3.8

Distance across axis (km)

Fig. 5. Relocated 20082009 hypocenters beneath Lucky Strike volcano. Hypocenters are indicated by circles, scaled by the local magnitude and color-coded by cluster:
green central graben north, blue south peak, purple east, redsummit depression. Large ellipses show absolute location uncertainties (from HYPOCENTER), small
ellipses show relative location uncertainties (from HypoDD). (A) Map view. Thick black contours mark isodepths around the summit depression (1710 m beneath sea level
(mbsl)) and its three surrounding peaks (1660 mbsl). Red (thick) dashed lines mark the bounds of the Axial Magma Chamber (AMC) reector (Combier, 2007; Singh et al.,
2006). Black crosses mark high-T vents, grey circles diffuse ow sites (Barreyre et al., 2012). Dashed lines show the tracks along which the seaoor and AMC depths are
plotted in the cross-sections. (B) Along-axis cross-section. Scale is the same as the map view, depth is referenced to the average seaoor depth, and there is no vertical
exaggeration. The dashed line is the AMC depth on a line crossing beneath the hydrothermal eld. Black crosses mark high-T vents at the seaoor. (C) Across-axis cross-
section, same scales and symbols as in B. (D) Map view of the AMC reector (Combier, 2007), overlain by earthquakes and isodepth contours. The AMC depth is indicated by
the colors and the reection amplitude by the brightness. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this
article.)

0.51.3 km north of the center of the hydrothermal elds and the although we cannot rule out that it is an artifact of 3D structure
other 0.51 km south of the elds. Because of their much greater within the volcano. The absolute value of these depths is uncertain;
precision, we will focus on the relocated hypocenters. We believe they are clearly above the AMC, but we cannot tell how far because
that they accurately represent the 20082009 seismicity because of the large absolute depth uncertainty.
their cumulative seismic moment (Fig. 4B) is very close to that of Fig. 6 shows events and faults in three across-axis sections:
the HYPOCENTER locations, and we argue that this also represents north, across and south of the hydrothermal eld. In the north
the background 20072008 activity because the distribution of section (Fig. 6A), 95% of the hypocenters are at more than 2.5 km
events located using HYPOCENTER was very similar in 20072008 depth and are not aligned along any structures. The shallower
and 20082009 (Fig. 4EG). northern cluster events are, however, aligned on an along-axis
Fig. 5 shows map and side views of the 20082009 relocated sub-vertical surface (Fig. 7). These small (ML o0) events occurred
events. 79% of the events are in a tight cluster beneath the graben throughout the 20082009 deployment (the 20072008 data lacks
north of the hydrothermal eld, 7% are beneath the peak just south the resolution needed to identify them), although they are slightly
of the eld, 5% are just east of the eld, 2% are beneath the eld and more concentrated during the AprilJune 2009 swarm (Fig. 7D).
the remaining 7% are scattered over the volcano. The north and south These events had very small amplitudes on the seismometer traces
clusters correspond to the two separate groups of events observed and we could not constrain their rst motion directions.
along-axis through both years of the experiment (Fig. 4E). The southern event cluster is 0.50.7 km shallower than the
The northern cluster starts at the north end of the hydrothermal northern cluster (Fig. 5). This cluster can also be divided into two
eld and extends approximately 0.5 km further along-axis. Most of depth ranges, with the deeper events showing no spatial align-
the events are between 2.7 and 3.0 km depth (Fig. 4B and C). ment and the shallower events aligned on an along-axis sub-
Although the absolute hypocenter depths are uncertain, the events vertical surface.
clearly cluster in a 100200 m-thick layer above the AMC reector. Seismic activity increased in both clusters at the end of March
The bottom of this layer is very well dened, deepening northwards 2009, leading to a 3-month long event swarm (Fig. 8).
slightly away from the hydrothermal eld (Fig. 5B). This deepening The increase in activity coincided with the strongest seismic event
trend echoes the deepening of the AMC reector towards the north, in the north cluster (and the second strongest during the two year
W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128 123

Fig. 6. Across-axis cross-sections of seismic events and sub-surface structure. Solid lines faults imaged using seismic reection data (Combier, 2007), dashed lines faults
extrapolated from seaoor observations. Dashed grey line AMC reector. Left: map view with the extent of each across-axis box indicated by the dashed lines. (A) Cross-
section across the central graben north of the hydrothermal eld. (B) Cross-section across the hydrothermal eld, with seaoor hydrothermal discharge zones indicated in
grey. (C) Cross-section across the south peak.

of the experiment): a magnitude 1.6 event on 30 March. This 5. Discussion


event, along with the only other event with ML 4 0.5 within the
northern cluster (a magnitude 0.8 event in October 2008) appears Most of the seismic events beneath the Lucky Strike volcano lie in
to have a different mechanism than the smaller events in the same two clusters, one north and one south of the hydrothermal eld, and
cluster, with opposite phase on the West, Central and East both clusters are aligned with the regional spreading axis and faulting
seismometers (Fig. 9). The rst motions are not the same at the and ssuring direction, beneath the axis dened by the volcano's
exterior seismometers (North, South, East and West), indicating central graben and hydrothermal eld. The cluster events are small
that these are not single-couple events, but they do not appear to (1.2oML o1.6) and the clusters cover only a small portion of the
be normal or thrust events on along-axis faults, either, as the AMC reector, have a clear maximum depth bound and a much
western and eastern sites recorded the same rst motion shorter along-axis extent than the visible volcano faults.
direction. The distribution of events suggests that the clusters are con-
The rate of events gradually increased during the swarm to a trolled both by the AMC and by regional stresses that created
peak level in mid-May. The largest event recorded beneath the surface features. The lateral alignment of the clusters correlates to
volcano, a magnitude 1.7 event south of the volcano's south peak, surface features, whereas their maximum depth bound is probably
occurred on 29 May 2009, preceded by an increase in activity in all controlled by the AMC. Near the bottom of these clusters, where
of the clusters: the earthquake rate in the northern cluster most of the earthquakes occur, there is no evidence for alignment
increases on 17 May and there are event bursts in the southern of events along faults, but the shallower earthquakes are aligned
cluster on 22 May and in the central and eastern clusters on 2729 on steeply dipping planes that strike along-axis.
May 2009. During the swarm, two other clusters appeared in the These characteristics indicate that the clusters sit in a narrow,
previously inactive region between the north and south clusters, relatively weak axial zone that penetrates down near to the AMC.
one beneath the hydrothermal elds and the other just east of the Three different processes are commonly evoked for small earthquakes
elds (Figs. 5 and 8A). Seismicity returns to pre-swarm levels by over mid-ocean ridge magma chambers: thermal contraction at the
early July (Fig. 4C). base of hydrothermal circulation cells (e.g., Sohn et al., 2004; Tolstoy
The maximum depth of earthquakes in the northern cluster et al., 2008), thermal contraction in the AMC (e.g., Sohn et al., 1999) or
increased by 0.050.1 km during the swarm (Fig. 8B). The deepen- ination/deation of the AMC (Wilcock et al., 2009). Ination/deation
ing was gradual and did not correspond to any migration of of the AMC is an unlikely source for the background seismicity
seismic activity along-axis. Although the total deepening is less because of constancy of this activity over 2 yr, the focusing of this
than the absolute hypocenter uncertainty, it is greater than the activity over a very small portion of the AMC, and the lack of measured
relative uncertainty and well within the variance in the distribu- vertical deformation at the seaoor (Ballu et al., 2012). Any of the three
tion of maximum event depths before and after the swarm. processes would generate stress at or below the bottom of the
Absolute pressure gauges at the volcano summit and off-axis, observed clusters: the narrow across-axis range of the clusters well
whose relative precision is on the order of 2 cm, show no evidence above this bottom indicates that they lie in a narrow weakened zone.
for subsidence associated with this event (Ballu et al., 2012). The north cluster events are almost exactly beneath the surface
We nd no evidence for extensional tectonic earthquakes beneath graben, consistent with the fast-spreading model of a sub-graben
the volcano during the 2-yr experiment period. No events have zone of sub-vertical fracturing or faulting created by combined
magnitudes greater than 1.7 and none fall on or near the 30451 extension, diking and magma chamber collapse (e.g., Lagabrielle and
inward-dipping faults that were imaged by active seismic reection Cormier, 1999). The shallower, near-vertically aligned events over both
data (Singh et al., 2006; Combier, 2007) or extrapolated from seaoor clusters may be regions of focused hydrothermal ow along ssures or
fault scarps. The only events that appear to be aligned have near-vertical faulting.
magnitudes less than 0 and their alignment planes are sub-vertical. This narrow, weak axial region is probably a permeable zone that
Water column hydrophone data indicate that Mid-Atlantic Ridge focuses deep hydrothermal circulation. This is clearly the case if the
segments from 15301N host 1050 M 23+ events/yr (Smith et al., deep events are caused by in-rock thermal contraction, but even if the
2003). Indeed, our instruments detect numerous larger events north source is contraction of the AMC, the focusing of events within 1 km of
and south of the volcano (99% of the detected events are outside of the volcano summit eld indicates that this contraction is focused near
our network), suggesting that there is a strong background tectonic the volcano summit. Such a contraction would probably not be
seismicity along the segment axis, but not beneath the volcano. restricted to beneath the hydrothermal elds, because the seismicity
124 W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128

Fig. 7. Shallow events above the north cluster. Events less than 2.5 km beneath the seaoor, north of the hydrothermal eld and beneath the central graben are marked in
black, others are marked in grey. (A) Map view. The dashed box outlines the horizontal range we used to select events. (B) Across-axis cross-section. (C) Along-axis cross-
section. (D) Distribution of these events with time (3-day bins).

cuts off rapidly rather than gradually to the north and south. Our assumed that hydrothermal uids ow preferentially through high
preferred source for the background events (those outside of the permeability zones: along-axis dikes emplacement and collapse zones
swarm period) is thermal contraction of the rock near the bottom of at magma-rich ridges and across-axis faults at magma-poor ridges.
the hydrothermal cells. This interpretation is consistent with the sharp However, numerical modeling indicates that the signicant changes in
along-axis limits to the seismicity and particularly with the lack of uid properties beneath high-ux vent elds can play an important
events beneath the hydrothermal eld, because numerical models role in organizing ow (Coumou et al., 2009). Applying these models
predict low basal heat ow at the intersection of two along-axis to the case of the Lucky Strike volcano, Fontaine et al. (2012) found
hydrothermal cells (e.g., Fontaine et al., 2012). The shallower events in that a narrow permeability zone was not sufcient to generate along-
the clusters could be small fractures in response to stresses generated axis hydrothermal ow, but that the combination of this permeable
by the deeper thermal contraction events. zone with the shallowing of the AMC towards the vent elds could
The scale and direction of hydrothermal circulation is still an open generate km-long along-axis ow. In addition to providing evidence
question at mid-ocean ridges. Along-axis hydrothermal circulation has for this ow, the earthquake clusters suggest that the physical
been invoked on fast- and magma-rich intermediate-spreading ridges structure of the Lucky Strike volcano very near to its summit may
(e.g., Sohn et al., 1997; Tolstoy et al., 2008), whereas across-axis ow currently be similar to that found at fast-spreading ridges.
has been invoked at intermediate- and slow spreading rate ridges Fig. 10 presents our model for the hydrothermal circulation, as
(e.g., deMartin et al., 2007; Johnson et al., 2010). It has generally been constrained by the event locations, surface features and subsurface
W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128 125

Cumulative Number of Events chemical differences could also be interpreted as resulting from two-
500 50 phase behavior of a similar source uid (Pester et al., 2012).
A SWARM The 2009 event swarm probably represents a period of enhanced
400 1.7 40

Other Clusters
downward propagation of thermal cracking into newly cooled rock,
North Cluster

which could be associated with a contraction event in the AMC. The


300 uth 30
So increase in activity in both the north and south clusters and the
200 rth 20 activation of the central and eastern clusters suggest that the swarm
No was triggered by a change in stress whose scale is larger than either of
0.8 st l
100 1.6 Ea C entra 10 the circulation cells, for example a broad AMC contraction event, but
the lack of measurable vertical motion at the seaoor is inconsistent
0
Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul with any collapse on the order of the 50100 m deepening of events
during the swarm. The swarm begins concurrently with the second
Depth beneath seafloor ref (km)

2.6 B largest event recorded within the network: a magnitude 1.6 event
near the bottom of the northern cluster on 30 March 2009 (Figs. 2D
2.7 and 5B). First motions for this event are more consistent with across-
axis than along-axis normal slip (dilation to the north, west and east;
2.8 compression nearby and to the south). Although along-axis faulting is
0.8 usually assumed for AMC events, across-axis slip could be created by
differential northsouth contraction of the AMC. The magnitude
2.9 1.6 event in the northern cluster, could fracture recently brittle rock
1.6
at the bottom of the hydrothermal circulation cells, allowing the
3 development and penetration of new cracks (Lister, 1974), while
Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul reactivating the overlying fracture network along the axial graben.
This event could have been triggered by a deeper contraction event, or
Fig. 8. Cumulative sums of 20082009 events, and depths of north cluster events. it could itself have triggered the swarm by generating stresses and
The grey background marks the 2009 event swarm. (A) Cumulative number of uid ow that affect the other clusters.
events for the 4 clusters. The times of the two largest north cluster events (ML 0.8
and 1.6) and the largest volcano event (ML 1.7) are marked by stars. (B) Detail of
The lack of background seismic activity beneath the hydro-
northern cluster event depths with time, with the two largest north cluster events thermal eld is consistent with numerical modeling that suggests
marked by stars. basal heat ux is low beneath the upow intersection of two
along-axis hydrothermal cells (Fontaine et al., 2012). Increased
activity in this zone during the 2009 event swarm, suggests either
an increase in basal heat ux due to perturbation or augmentation
of the ow eld, or local adjustment to increased stress during
the swarm.
Although the AMC does not appear to affect the across-axis
distribution of earthquakes, it probably permits the development
of the along-axis ow cell and organizes the ow direction.
Numerical modeling suggests that the topography of both the
bottom and tops of hydrothermal systems affects the upow, with
topographic highs on the heat source generating upow (Fontaine
et al., 2008, 2011) and seaoor highs attracting the upow
(Bani-Hassan et al., 2012). In a cross-section along the zone of
interpreted hydrothermal circulation, the AMC is shallowest
Fig. 9. Vertical channel P-wave arrival waveforms for north cluster events. Traces
beneath the hydrothermal eld (Fig. 4B) and should focus upow
are named by the position of the station relative to the hydrothermal elds: (N) there. This would be reinforced by the surface topography, which
orth, (S)outh, (E)ast, (W)est and (C)entral. Left: The two largest events, super- is also shallowest around the eld.
imposed. Right: Stacked traces for ML o 0.5 events with correlation greater than The AMC topography is more complex than a simple shallow-
0.8: the number of events stacked depends on the station and is shown at the left
ing beneath the volcano summit. The AMC reector is shallowest
end of the trace. The timing between stations is not the same for each event, we
stacked them at the highest correlation value and then shifted the traces to 1 km west of the summit depression, in a region with little
correspond to the timing observed on the large events. seismicity and no observed hydrothermal vents or diffuse outow
(Fig. 11). Hydrothermal circulation may be the primary cause for
imaged faults. Most of the hydrothermal circulation is in two the complicated shape of the AMC, pushing down the AMC
along-axis cells, approaching the summit depression from the beneath the circulation cells and leaving it nearer to its initial
south and the north. The north cell penetrates much closer to the emplacement depth to the west.
AMC than does the south cell and contains ten times more events, Taking into account these observations, we speculate that the
suggesting that most of the hydrothermal heat ux comes from AMC may currently be in the waning phase of a magmatic cycle,
there. The cells could also exploit adjacent extensional faults or the probably one of several that built the volcanic edice and
porous peaks to either side of the central graben to bring seawater periodically reactivate and/or reset the hydrothermal eld. Several
from further off-axis. studies indicate that Mid-Atlantic Ridge hydrothermal systems are
The two main hydrothermal cells (north to south from the central periodically reactivated and that their active periods are signi-
graben north of the eld, south to north from the south peak) and cantly shorter than the reactivation time (e.g., Lalou et al., 1998;
their different penetration depths could provide different source uids Humphris and Cann, 2000). Geochemical studies of the Snake Pit,
to the surface vents. Fluid geochemistry data at the Lucky Strike vents TAG and other MAR hydrothermal elds suggest activation periods
do show two families of high-T vents, with the southeastern vents of hundreds of years separated by quiescent periods of thousands
having signicantly lower Cl/Mg and higher 86Sr/87Sr than the north- of years (Lalou et al., 1990, 1993, 1995, 1998; Cherkashov et al.,
eastern and western vents (Charlou et al., 2000). However, these 2010). At Lucky Strike, we lack such geochronological constraints,
126 W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128

km
8
3.

1.6 km

4.2 km
4.2 km

3 km North Rift
summit Depression
south peak

Fig. 10. Model of hydrothermal circulation beneath the Lucky Strike segment. Left: Along-axis cross-section showing along-axis circulation correlated with maximum
seismicity depths. Black arrows show vertical and along-axis ow, grey arrows represent ow from off-axis along faults. Grey portions at the seaoor indicate hydrothermal
diffuse hydrothermal ow, black asterisks are high-T vents. Right: Cross-sections showing the downwelling and upwelling zones. Lines show faults and the AMC as in
previous plots, shaded areas indicate regions of higher permeability and arrows suggest ow pathways.

long-term evolution of seismicity and the AMC reector, would be


needed to determine whether there is or recently was deep hydro-
thermal circulation outside of the currently active axial region.
1000
6. Conclusions
Along-axis distance (m)

Seismological events recorded over the 2 yr from July 2007 to July


2009 beneath Lucky Strike volcano outline two along-axis hydro-
0 thermal cells feeding the summit vent eld. Events are clustered into
two groups of continuous activity around the volcano's hydrothermal
vent eld: one beneath the graben north of the eld and another
3200 beneath the peak just south of the eld.
The events are focused within a narrow axial region aligned with
AMC Depth (m bsfr)

1000 3400 the volcano's vent eld and central graben that have a clear depth
limit over the AMC reector. We interpret this region as a perme-
3600
able zone of sub-vertical fracturing and/or faulting, analogous to
those found beneath fast-spreading ridge grabens. The source of
3800
2000 these events is probably thermal contraction of rocks at the bottom
of the hydrothermal cells or contraction of the AMC: we favor the
1000 0 1000 former explanation because of the tight across- and along-axis
Across-axis distance (m) bounds on the events. The lack of events directly beneath the vent
Fig. 11. Comparison of the axial magma chamber reector topography, deep eld is consistent with the lower basal heat ux that is seen in
hydrothermal circulation cells and surface features. Hatchered boxes deep hydro- numerical models at the upwelling intersection of two along axis
thermal circulation cells identied in this paper, thick contoursisodepth contours hydrothermal cells. This along-axis ow depends both on the
for the summit depression and the three surrounding peaks. Shaded con- existence of a narrow, highly permeable axial zone and on an
tours AMC reector depth beneath the seaoor reference (bsfr).
AMC that shallows towards the vent elds.
A 3-month long event swarm in AprilJune 2009 results in a 50
100 m deepening of the northern circulation cell. The swarm affects
but it is possible that the reactivations are caused by new both the northern and southern event clusters and activates seismi-
injections of magma beneath the volcano, which may be ulti- city between them. A magnitude 1.6 event, which occurred near the
mately linked to the formation of individual volcanic cones that beginning of the event swarm and which was located at the bottom
are rifting at varying degrees. Backwards modeling of the AMC of the northern cluster, may have fractured the underlying rock,
shape as a function of hydrothermal circulation might provide an opening it to downward propagation of hydrothermal circulation
idea of the age of the last magma injection and a prediction for the through water penetration and further fracturing.
decadal evolution of the hydrothermal elds. The present-day shape of the AMC, which is shallowest 1 km
The reector is also depressed beneath the three peaks surround- west of the depression, may be the result of cooling being the most
ing the summit depression. These peaks are much more porous than efcient on-axis, pushing down the AMC there more than
the surrounding seaoor (Arnulf et al., 2011) and could make efcient elsewhere.
pathways for hydrothermal uid circulation, but there is no evidence
that this porosity extends down to near the AMC (the Arnulf et al.
(2011) study only constrains the upper 0.7 km of the crust). The deep Acknowledgments
zones beneath the peaks could also have been cooled by uids
traveling along faults further off-axis, such as the western graben's We thank the captain and crew of the research vessels N/O
western bounding fault (Singh et al., 2006). More data, such as the Suroit and Atalante for their help with the OBS deployments and
W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128 127

recoveries. We thank the Monitoring of the Mid-Atlantic Ridge Humphris, S.E., Fornari, D.J., Scheirer, D.S., German, C.R., Parson, L.M., 2002.
(MoMAR) and European Multidisciplinary Seaoor Observatory Geotectonic setting of hydrothermal activity on the summit of Lucky Strike
Seamount (37117'N, Mid-Atlantic Ridge). Geochem. Geophys. Geosyst. 3 (8),
(EMSO) initiatives for supporting this research. We also thank http://dx.doi.org/10.1029/2001GC000284.
Fabrice Fontaine for fruitful discussions and Rob Sohn and Del- Hutton, L.K., Boore, D.M., 1987. The M_L scale in southern California. Bull. Seismol.
Wayne Bohnenstiehl for detailed and insightful reviews. Soc. Am. 77 (6), 20742094.
Johnson, H.P., Tivey, M.A., Bjorklund, T.A., Salmi, M.S., 2010. Hydrothermal circula-
tion within the Endeavour Segment, Juan de Fuca Ridge. Geochem. Geophys.
Geosyst. 11 (5), http://dx.doi.org/10.1029/2009GC002957.
References Kissling, E., Ellsworth, W.L., Eberhart-Phillips, D., Kradolfer, U., 1994. Initial
reference models in local earthquake tomography. J. Geophys. Res. 99 (B10),
Allen, R., 1982. Automatic phase pickers: their present use and future prospects. 1963519646.
Bull. Seismol. Soc. Am. 72 (6), S225S242. Kong, L.S.L., Solomon, S.C., Purdy, G.M., 1992. Microearthquake characteristics of a
Arnulf, A.F., Singh, S.C., Harding, A.J., Kent, G.M., Crawford, W., 2011. Strong seismic mid-ocean ridge along-axis high. J. Geophys. Res. 97 (2), 16591685.
heterogeneity in layer 2A near hydrothermal vents at the Mid-Atlantic Ridge. Lagabrielle, Y., Cormier, M.-H., 1999. Formation of large summit troughs along the
Geophys. Res. Lett. 38 (13), http://dx.doi.org/10.1029/2011GL047753. East Pacic Rise as collapse calderas: an evolutionary model. J. Geophys. Res.
Arnulf, A.F., Harding, A.J., Singh, S.C., Kent, G.M., Crawford, W., 2012. Fine-scale 104 (B6), 1297112988.
velocity structure of upper oceanic crust from full waveform inversion of Lalou, C., Thompson, G., Arnold, M., et al., 1990. Geochronology of TAG and Snakepit
downward continued seismic reection data at the Lucky Strike Volcano, Mid- hydrothermal elds, mid-Atlantic Ridge: witness to a long and complex
Atlantic Ridge. Geophys. Res. Lett. 39 (8), http://dx.doi.org/10.1029/ hydrothermal history. Earth Planet. Sci. Lett. 97, 113128.
2012GL051064. Lalou, C., Reyss, J.-L., Brichet, E., et al., 1993. New age data for Mid-Atlantic Ridge
Ballu, V., de Viron, O., Crawford, W.C., Cannat, M., Escartin, J., 2012. Long-term hydrothermal sites: TAG and Snakepit chronology revisited. J. Geophys. Res. 98
observations of seaoor pressure variations at Lucky Strike volcano, Mid- (B6), 97059713.
Atlantic Ridge. Fall Meeting, AGU, Abstract, OS13B-1730. Lalou, C., Reyss, J.L., Brichet, E., Rona, P.A., Thompson, G., 1995. Hydrothermal
Bani-Hassan, N., Iyer, K., Rpke, L.H., Borgia, A., 2012. Controls of bathymetric relief activity on a 105-year scale at a slow-spreading ridge, TAG hydrothermal eld,
on hydrothermal uid ow at mid-ocean ridges. Geochem. Geophys. Geosyst. Mid-Atlantic Ridge, 261N. J. Geophys. Res. 100 (B9), 1785517862.
13 (5), http://dx.doi.org/10.1029/2012GC004041. Lalou, C., Reyss, J.L., Brichet, E., 1998. Age of sub-bottom sulde samples at the TAG
Barreyre, T., Escartn, J., Garcia, R., et al., 2012. Structure, temporal evolution, and active mound. In: Herzig, P.M., Humphris, S.E., Miller, D.J., Zierenberg, R.A.
heat ux estimates from the Lucky Strike deep-sea hydrothermal eld derived (Eds.), TAG: Drilling An Active Hydrothermal System on a Sediment-Free
from seaoor image mosaics. Geochem. Geophys. Geosyst. 13, http://dx.doi.org/ Slow-Spreading Ridge. Ocean Drilling Program, College Station, TX, pp. 111118.
10.1029/2011GC003990. Langmuir, C.H., 1993. Geological setting and characteristics of the Lucky Strike vent
Bohnenstiehl, D.R., Waldhauser, F., Tolstoy, M., 2008. Frequency-magnitude dis- eld at 37117'N on the Mid-Atlantic Ridge. Eos Trans. Am. Geophys. Union
tribution of microearthquakes beneath the 91 50'N region of the East Pacic 74, 99.
Rise, October 2003 through April 2004. Geochem. Geophys. Geosyst. 9 (10), Langmuir, C., Humphris, S., Fornari, D.J., et al., 1997. Hydrothermal vents near a
http://dx.doi.org/10.1029/2008GC002128. mantle hot spot; the Lucky Strike vent eld at 37 degrees N on the Mid-Atlantic
Cannat, M., Briais, A., Deplus, C., et al., 1999. Mid-Atlantic RidgeAzores hotspot Ridge. Earth Planet. Sci. Lett. 148, 12.
interactions; along-axis migration of a hotspot-derived event of enhanced Lienert, B.R., Havskov, J., 1995. A computer program for locating earthquakes both
magmatism 10 to 4 Ma ago. Earth Planet. Sci. Lett. 173 (3), 257269. locally and globally. Seismol. Res. Lett. 66 (5), 2636.
Charlou, J.L., Donval, J.P., Douville, E., et al., 2000. Compared geochemical signatures Lienert, B.R., Berg, E.W., Frazer, N.L., 1986. HYPOCENTER: an earthquake location
and the evolution of Menez Gwen (37 degrees 50'N) and Lucky Strike (37 method using centered, scaled, and adaptively. Bull. Seismol. Soc. Am. 76 (3),
degrees 17'N) hydrothermal uids, south of the Azores Triple Junction on the 771783.
Mid-Atlantic Ridge. Chem. Geol. 171 (12), 4975. Lister, C.R.B., 1974. On the penetration of water into hot rock. Geophys. J. R. Astron.
Cherkashov, G., Poroshina, I., Stepanova, T., et al., 2010. Seaoor massive suldes Soc. 39, 465509.
from the Equatorial Mid-Atlantic Ridge: new discoveries and perspectives. Mar. Lockner, D., Byerlee, J.D., 1977. Hydrofracture in Weber sandstone at high conning
Georesour. Geotechnol. 28, 222239. pressure and differential stress. J. Geophys. Res. 82 (14), 20182026.
Combier, V., 2007. Mid-Ocean Ridge Processes: Insights from 3D Reection Seismics Ondras, H., Fouquet, Y., Voisset, M., Radford-Knoery, J., 1997. Detailed study of
at the 91N OSC on the East Pacic Rise, and the Lucky Strike Volcano on the three contiguous segments of the Mid-Atlantic Ridge, south of the Azores (371N
Mid-Atlantic Ridge. Doctoral thesis. Institut de Physique du Globe de Paris. to 38130'N), using acoustic imaging coupled with submersible observations.
Coumou, D., Driesner, T., Geiger, S., Paluszny, A., Heinrich, C.A., 2009. High- Mar. Geophys. Res. 19, 231255.
resolution three-dimensional simulations of mid-ocean ridge hydrothermal Ondras, H., Cannat, M., Fouquet, Y., et al., 2009. Recent volcanic events and the
systems. J. Geophys. Res. 114 (B7), http://dx.doi.org/10.1029/2008JB006121. distribution of hydrothermal venting at the Lucky Strike hydrothermal eld,
Crawford, W.C., Singh, S.C., Seher, T., et al., 2010. Crustal structure, magma chamber Mid-Atlantic Ridge. Geochem. Geophys. Geosyst. 10, Q02006, http://dx.doi.org/
and faulting beneath the Lucky Strike hydrothermal vent eld, Diversity of 10.1029/2008GC002171.
Hydrothermal Systems on Slow Spreading Ocean Ridges. AGU pp. 113132. Parson, L.M., Grcia, E., Coller, D., German, C.R., Needham, D., 2000. Second-order
deMartin, B.J., Sohn, R.A.R., Canales, J.P., Humphris, S.E., 2007. Kinematics and segmentation; the relationship between volcanism and tectonism at the MAR,
geometry of active detachment faulting beneath the Trans-Atlantic Geotraverse 381N35140'N. Earth Planet. Sci. Lett. 178, 231251.
(TAG) hydrothermal eld on the Mid-Atlantic Ridge. Geology 35 (8), 711714. Pester, N.J., Reeves, E.P., Rough, M.E., et al., 2012. Subseaoor phase equilibria in
Detrick, R.S., Needham, H.D., Renard, V., 1995. Gravity anomalies and crustal high-temperature hydrothermal uids of the Lucky Strike Seamount (Mid-
thickness variations along the Mid-Atlantic ridge between 331N and 401N. Atlantic Ridge, 37117?N). Geochim. Cosmochim. Acta 90, 303322, http://dx.
J. Geophys. Res. 100 (B3), 37673787. doi.org/10.1016/j.gca.2012.05.018.
Fontaine, F.J., Cannat, M., Escartin, J., 2008. Hydrothermal circulation at slow- Phipps Morgan, J., Parmentier, E.M., Lin, J., 1987. Mechanisms for the origin of mid-
spreading mid-ocean ridges: the role of along-axis variations in axial litho- ocean ridge axial topography: implications for the thermal and mechanical
spheric thickness. Geology 36 (10), 759, http://dx.doi.org/10.1130/G24885A.1. structure of accreting plate boundaries. J. Geophys. Res. 92 (12), 1282312836.
Fontaine, F.J., Olive, J.-A., Cannat, M., Escartin, J., Perol, T., 2011. Hydrothermally- Richter, C.F., 1935. An instrumental earthquake magnitude scale. Bull. Seismol. Soc.
induced melt lens cooling and segmentation along the axis of fast-and inter- Am. 25 (1), 132.
mediate-spreading centers. Geophys. Res. Lett. 38, http://dx.doi.org/10.1029/ Scheirer, D.S., Fornari, D.J., Humphris, S.E., Lerner, S., 2000. High-resolution seaoor
2011GL047798. mappling using the DSL-120 sonar system: quantitative assessment of sidescan
Fontaine, F.J., Cannat, M., Escartin, J., Crawford, W.C., Singh, S.C., 2012. Physical and phase-bathymetry data from the Lucky Strike segment of the Mid-Atlantic
inter-relationships between hydrothermal activity, faulting and magmatic Ridge. Mar. Geophys. Res. 21, 121142.
processes at the center of a slow-spreading, magma-rich mid-ocean ridge Seher, T., Crawford, W.C., Singh, S.C., et al., 2010a. Crustal velocity structure of the
segment: a case study of the Lucky Strike segment (MAR, 37103-37?N). 2012 Lucky Strike segment of the Mid-Atlantic Ridge (371N) from seismic refraction
Fall Meeting, AGU, Abstract OS12A-04. measurements. J. Geophys. Res. 115 (B03103), http://dx.doi.org/10.1029/
Fouquet, Y., Ondreas, H., Charlou, J.L., et al., 1995. Atlantic lava lakes and hot vents. 2009JB06650.
Nature 377 (6546), 201. Seher, T., Singh, S., Crawford, W., Escartin, J., 2010b. Upper crustal velocity structure
Golden, C.E., Webb, S.C., Sohn, R.A., 2003. Hydrothermal microearthquake swarms beneath the central Lucky Strike segment from seismic refraction measure-
beneath active vents at Middle Valley, northern Juan de Fuca Ridge. J. Geophys. ments. Geochem. Geophys. Geosyst. 11 (5), Q05001, http://dx.doi.org/10.1029/
Res. 108 (B1), http://dx.doi.org/10.1029/2001JB000226. 2009GC002894.
Gripp, A.E., Gordon, R.G., 1990. Current plate velocities relative to the hotspots Sempere, J.-C., Lin, J., Brown, H.S., Schouten, H., Purdy, G.M., 1993. Segmentation
incorporating the NUVEL-1 global plate motion model. Geophys. Res. Lett. 17 and morphotectonic variations along a slow-spreading centerthe Mid-
(8), 11091112. Atlantic Ridge (24100'N30140'N). Mar. Geophys. Res. 15 (3), 153200.
Gudmundsson, A., Fjeldskaar, I., Brenner, S.L., 2002. Propagation pathways and uid Singh, S.C., Crawford, W.C., Carton, H., et al., 2006. Discovery of a magma chamber
transport of hydrofractures in jointed and layered rocks in geothermal elds. and faults beneath a Mid-Atlantic Ridge hydrothermal eld. Nature 442 (7106),
J. Volcanol. Geotherm. Res. 116 (34), 257278. 10291032.
Humphris, S.E., Cann, J.R., 2000. Constraints on the energy and chemical balances of Smith, D.K., Escartin, J., Cannat, M., et al., 2003. Spatial and temporal distribution of
the modern TAG and ancient Cyprus seaoor sulde deposits. J. Geophys. Res. seismicity along the northern Mid-Atlantic Ridge (15 degrees 35 degrees N).
105 (B12), 2347728488. J. Geophys. Res. 108 (B3), http://dx.doi.org/10.1029/2002JB001964.
128 W.C. Crawford et al. / Earth and Planetary Science Letters 373 (2013) 118128

Sohn, R.A., Webb, S.C., Hildebrand, J.A., Cornuelle, B.D., 1997. Three-dimensional Stroup, D.F., Tolstoy, M., Crone, T.J., et al., 2009. Systematic along-axis tidal
tomographic velocity structure of upper crust, Coaxial segment, Juan de Fuca triggering of microearthquakes observed at 91500N East Pacic Rise. Geophys.
Ridge: implications for on-axis evolution and hydrothermal circulation. Res. Lett. 36, L18302, http://dx.doi.org/10.1029/2009GL039493.
J. Geophys. Res. 102 (B8), 1767917695. Tapponnier, P., Francheteau, J., 1978. Necking of the lithosphere and the mechanics
Sohn, R.A., Fornari, D.J., Von Damm, K.L., Hildebrand, J.A., Webb, S.C., 1998a. Seismic of slowly accreting plate boundaries. J. Geophys. Res. 83 (8), 39553970.
and hydrothermal evidence for a cracking event on the East Pacic Rise crest at Tolstoy, M., Waldhauser, F., Bohnenstiehl, D.R., Weekly, R.T., Kim, W.Y., 2008.
9 degrees 50'N. Nature 396 (6707), 159161. Seismic identication of along-axis hydrothermal ow on the East Pacic Rise.
Sohn, R.A., Hildebrand, J.A., Webb, S.C., 1998b. Postrifting seismicity and a model for Nature 451 (7175), 181184, http://dx.doi.org/10.1038/nature06424.
the 1993 diking event on the CoAxial segment, Juan de Fuca Ridge. J. Geophys. Waldhauser, F., 2001. hypoDDA Program to Compute Double-Difference Hypo-
Res. 103 (5), 98679877. center Locations, USGS Open File Report 010113.
Sohn, R.A., Hildebrand, J.A., Webb, S.C., 1999. A microearthquake survey of the high- White, R.S., Drew, J., Martens, H.R., et al., 2011. Dynamics of dyke intrusion in the
temperature vent elds on the volcanically active East Pacic Rise (9150'N). J.
mid-crust of Iceland. Earth Planet. Sci. Lett. 304 (34), 300312, http://dx.doi.
Geophys. Res. 104 (B11), 2536725378.
org/10.1016/j.epsl.2011.02.038.
Sohn, R.A., Barclay, A.H., Webb, S.C., 2004. Microearthquake patterns following the
Wilcock, W.S.D., Hooft, E.E.E., Toomey, D.R., et al., 2009. The role of magma injection in
1998 eruption of Axial Volcano, Juan de Fuca Ridge: mechanical relaxation and
localizing black-smoker activity. Nat. Geosci. 509513, http://dx.doi.org/10.1038/
thermal strain. J. Geophys. Res. 109 (B01101), http://dx.doi.org/10.1029/
2003JB002499. NGEO550.

Das könnte Ihnen auch gefallen