Geochemistry
Geophysics
Geosystems
3
G
AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES
Published by AGU and the Geochemical Society
Article
Volume 8, Number 2
20 February 2007
Q02008, doi:10.1029/2006GC001359
ISSN: 1525-2027
Click
Here
for
Full
Article
Dynamic models for mantle flow and seismic anisotropy in
the North Atlantic region and comparison with observations
Gabriele Marquart
SRON and Institute of Earth Science, University of Utrecht, Postbus 80.021, 3508 TA Utrecht, Netherlands
([email protected])
Harro Schmeling
Institute of Earth Sciences, Geophysics Section, J. W. Goethe University, Feldbergstrasse 47, D-60323 Frankfurt,
Germany
Ondřej Čadek
Department of Geophysics, Faculty of Mathematics and Physics, Charles University, V Holesovickach 2, 180 00 Praha 8,
Czech Republic
[1] Dynamic mantle flow and temperature models for the North Atlantic based on a regionalized P-wave
and a global S-wave tomography model were derived under the constraint of a maximum fit to the
observed gravity field. For the regional flow model Cartesian geometry, temperature- and depth-dependent
viscosity and a free slip surface were assumed, while the global model assumed a radially dependent
viscosity and kinematic plate velocity boundary conditions. Both models show pronounced upwellings
within the upper mantle beneath the Iceland area and the lower mantle beneath, the regional model
containing a lateral shift associated with a horizontal flow near 660 km depth. The upper mantle
temperature field of the regional model shows two distinct anomalies, one beneath Iceland and the westerly
adjacent regions with a connection to a deep mantle root and an excess temperature of 200°C, and a second
one below 300 km at the Kolbeinsey Ridge with an excess temperature of 120°C. These anomalies do not
appear to be connected. An essentially radial flow pattern is found south of Iceland with ridge parallel flow
along Reykjanes and divergent flow at the Kolbeinsey Ridge. The long-wavelength global model does not
show such details but is characterized by a NE-SW elongated upwelling flow beneath Iceland and a ridge
perpendicular flow within the upper mantle. From the modeled flow fields, lattice preferred orientations
(LPO) of olivine are calculated. For the regional model, azimuthal seismic anisotropy is predicted with fast
directions diverging away from Iceland and the Kolbeinsey ridge. The global model predicts roughly ridge
perpendicular fast directions. Comparison of predicted with observed seismic anisotropy models shows
regions of good agreement north, east, and SE of Iceland, as well as for Iceland. No agreement is found
beneath the Reykjanes Ridge area, leading to the speculation that the fast directions are perpendicular to
the flow due to a change in the LPO generating mechanism. Regarding geochemical findings, the regional
flow model can explain plume-related geochemical signatures observed on the Reykjanes Ridge and
predicts a deep, hot melt zone beneath the Kolbeinsey Ridge without plume tracers.
Components: 14,393 words, 10 figures, 1 table.
Keywords: North Atlantic; Iceland; Reykjanes Ridge; Kolbeinsey Ridge; mantle flow field; seismic anisotropy.
Index Terms: 8121 Tectonophysics: Dynamics: convection currents, and mantle plumes; 8137 Tectonophysics: Hotspots,
large igneous provinces, and flood basalt volcanism; 8180 Tectonophysics: Tomography (6982, 7270).
Received 5 May 2006; Revised 6 September 2006; Accepted 27 October 2006; Published 20 February 2007.
Copyright 2007 by the American Geophysical Union
1 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Marquart, G., H. Schmeling, and O. Čadek (2007), Dynamic models for mantle flow and seismic anisotropy in the North
Atlantic region and comparison with observations, Geochem. Geophys. Geosyst., 8, Q02008, doi:10.1029/2006GC001359.
1. Introduction
[2] This study focuses on the structure of mantle
flow in the North Atlantic region, in particular on the
large-scale region around the Iceland plume where
plume-driven flow and spreading-driven flow from
the Mid-Atlantic Ridge System are superimposed.
On the basis of tomography and gravity data, a
model for mantle dynamics in the North Atlantic
has been derived in a previous study [Marquart and
Schmeling, 2004]. Complemented by another globally based flow model we study some consequences
on geophysical and geochemical observables. Since
mantle flow pattern is expected to give rise to
seismic anisotropy, the flow fields are used to
determine the related elastic deformation tensor of
olivine (based on the formalism of Kaminski and
Ribe [2001, 2002]) and compare the direction of the
fast axis to recent results on fast seismic velocity
directions [e.g., Bjarnason et al., 2002; Li and
Detrick, 2003]. Another constraint on mantle flow
at least along the ridges can be found in the geochemical signature of mid-oceanic ridge basalts
drilled and dredged along the Reykjanes and Kolbeinsey Ridges. Therefore the mantle flow field
derived in this study will also be discussed in the
light of geochemical observations.
[3] Iceland is one of the few locations on Earth
where a mantle plume and a spreading ridge coincide. This is not a permanent setting, since lithospheric plates and their boundaries move with
respect to the mantle, and plumes are believed to
be either fixed in the mantle or move with different
speeds relative to plate boundaries [Steinberger and
O’Connell, 1998]. In the North Atlantic the plume
arrived under central or southeastern Greenland
about 58 Ma ago [Torsvik et al., 2001], presumably
caused rifting and plume-related volcanism in western, northern and eastern Greenland [Skogseid et al.,
1992; White, 1997; Scarrow et al., 2000] and finally
initiated massive volcanism and continental breakup
between the North American and Eurasian Plate.
Thus, in the early phase the plume was located to the
west of the ridge, which may not only have caused
the ridge jump from the extinct Aegir Ridge to the
present Kolbeinsey Ridge north of Iceland (see
Figure 1), but may also have contributed to a higher
plate velocity of the North American Plate compared
to the Eurasian Plate. Today the plate velocity of
Iceland on the North American Plate at 65°N is about
2.6 cm/a in roughly westward direction compared to
1.4 cm/a in SW direction on the Eurasian side;
arrows of motion are shown in Figure 1 (values refer
to the hot spot reference frame based on the HS3Nuvel1a plate motions [Gripp and Gordon, 2002]).
These plate motion vectors indicate a westward
migration of the spreading ridge with approximately
1 cm/a. This migration of the ridge system in respect
to the mantle very likely explains the observation
that after the strong initial volcanic phase in the
North Atlantic Basalt Province volcanic activity
ceased but was renewed about 20 Ma ago. By this
time the ridge axis has migrated close enough to the
decaying plume that hot upwelling plume material
could rise to a shallower level leading to increased
melt production and thus to a renewal of volcanic
activity forming the Iceland plateau [Mihalffy et al.,
2006]. Continuous westward movement of the ridge
system since then has led to the present situation with
the Reykjanes and Kolbeinsey Ridge system already
westward of Iceland, while the on-land spreading
ridge, the neovolcanic zones of Iceland, is anchored
to the mantle fixed plume and forms an indentation
to the east (see Figure 1). The link to the North
Atlantic ridge is provided by the Tjoernes Fracture
Zone in the north, and the ‘‘bookshelf’’-type faults of
the South Iceland Shear Zone. To anchor the rift axis
on Iceland to the hot mantle anomaly beneath, while
the North Atlantic Ridge System moves westward,
repeated eastward ridge jumps on Iceland during the
entire time of the island’s existence have occurred.
[4] The repeated migration of the rift axis is already
a clear indication of the importance of plume flux,
interacting with ridge perpendicular spreading flux
and large-scale plate movement. The complex
flow can be separated into four contributions
(see Figure 2): (1) Plume-related flow can be
expected to be predominately vertical below
200 km and radial outward at shallower depth
(Figure 2a). (2) A ridge parallel flow component
has been proposed on the basis of geochemical
observations (Figure 2b). Variations of rare Earth
elements and trace element concentrations along the
Reykjanes Ridge [Schilling, 1973] indicate that
material from the Iceland plume has been mixed
with more normal MORB type basalts several of
hundreds of kilometers along the Mid-Atlantic
Ridge, but with decreasing degree. This finding
2 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 1. Topography/bathymetry of the North Atlantic shows major tectonic structures (data are based on
ETOPO5 [National Oceanic and Atmospheric Administration, 1988]). The red arrows indicate the plate motion of the
North American and Eurasian Plates in a hot spot reference frame (HS3-Nuvel1a [Gripp and Gordon, 2002]).
has stimulated studies of along ridge channel like
flow transport [e.g., Albers and Christensen, 2001].
(3) Spreading-related flow should follow the divergence of the plates at least in the uppermost mantle
in an extended area on both sides of the ridge
(Figure 2c). (4) Global plate motion related flow
should coincide with the plate motion vectors in
Figure 1 and be characterized by its long wavelengths nature (Figure 2d).
[5] In the following sections, dynamic flow models
based on the P-wave tomography model by
Bijwaard and Spakman [1999] and S-wave
tomography model smean [Becker and Boschi,
2002] will be presented and compared to the four
idealized flow models and to results from seismic
anisotropy and geochemical studies.
2. Dynamic Flow Models for the North
Atlantic Region Based on Seismic
Tomography
[6] If one is interested in mantle flow within a
particular region, either global tomography based
flow models may be used, and one may zoom into
the region under consideration, or both tomography
and flow modeling may be restricted to the particular region. In the first approach far field (i.e.,
global) flow and density structures are consistently
included in the regional flow field. However, due
to computational limitations the flow model is
usually restricted to long-wavelength structures.
Such structures are appropriately represented by
global S-wave tomography models. A further
restriction of such flow models is that they usually
only allow for 1-D radially variable viscosity. In
the second approach, smaller-scale structures may
be included, which are better represented by
shorter-wavelength P-wave tomography. The
regional formulation allows for higher resolution
of the flow modeling and for 3-D variable
viscosity. However, as the far field is not included
care has to be taken when choosing the regional
boundaries and boundary conditions. Boundaries
must not cut through significant tomographic
anomalies. The flow region should be considerably
larger than the region with tomographic anomalies
to allow for large-scale return flows.
3 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 2. Possible idealized mantle flow geometry in the North Atlantic region [after Li and Detrick, 2003].
[7] In this paper we present one flow model based
on each of the above approaches, namely a regional
flow model based on the P-wave model by Bijwaard
and Spakman [1999] regionalized for the North
Atlantic, and a global flow model, based on the
S-wave model smean [Becker and Boschi, 2002].
We then compare the similarities and differences.
2.1. Tomography Models for the Mantle
Beneath the North Atlantic
[8] The North Atlantic has been studied by a
number of regional and global tomography studies.
A global P-wave tomography model with high
resolution within the North Atlantic region has
been developed by Bijwaard and Spakman
[1999]. In this model they have identified a clear
signature of a plume conduit related to Iceland,
rising through the entire mantle. This anomaly
starts at a location at the CMB beneath the southern
tip of Greenland and stretches as a strongly northeastward inclined structure through the lower
mantle and seems to branch off in several anomalies at the mantle transition zone. Bijwaard and
Spakman also observed reduced seismic wave
velocities in the upper mantle below Iceland and
the east Greenland margin, at a depth of 300 km
in a rather narrow zone approximately below the
Kolbeinsey Ridge, and even deeper below the
Greenland shield.
[9] When comparing this tomography model to
others from the same region we have to distinguish
between global and regional models which also
address the controversial topic of whether a plumelike anomaly exists in the lower mantle below
Iceland.
[10] Regional tomography models using dense data
coverage retrieved on Iceland [e.g., Bjarnason et
al., 1996; Wolfe et al., 1997; Allen et al., 1999;
Foulger et al., 2001] show a strong approximately
cylindrical low-velocity anomaly beneath Iceland.
However, due to the limited width of the seismic
station arrays, the outermost parts of these regional
models may not be suitable as a reference for an
undisturbed mantle, but might be part of a larger
anomaly.
[11] Low seismic velocities in the upper mantle
beneath the North Atlantic region are also found in
all global tomography models [e.g., Grand et al.,
1997; Grand, 2002; Masters et al., 2000; Mègnin
and Romanowicz, 2000; Zhao, 2001], but the area
is considerably larger than Iceland and even the
maximum of the low-velocity anomaly is not
centered on Iceland in all cases. Looking to the
upper mantle in more detail, a southward tilted
plume in the upper mantle roughly below Iceland
can be inferred from an along-ridge section of the
somewhat older global tomography model of
Zhang and Tanimoto [1993]. They also see a
4 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
low-velocity anomaly located in deeper parts of the
upper mantle beneath the Kolbeinsey Ridge. A
tilted plume is also found in a N-S section of the
global tomography model by Zhao [2001]. This
model also shows a weak indication of a branching
of the plume head at shallow depth in the direction
of the Reykjanes Ridge. A new global tomography
model with higher resolution by Montelli et al.
[2004] also indicates a seismic slow region below
Iceland, slightly tilted to the south, but their model
can trace the plume at most down to the mid lower
mantle. Concluding, a low-velocity anomaly in the
upper mantle in a large area around Iceland is a
robust feature and therefore also clearly shows up
in the composite S-tomography model smean
[Becker and Boschi, 2002], which we used for
our global flow model.
[12] Apart from tomography, there are a few other
seismic observations indicating that the plume
conduit reaches down into the lower mantle. Shen
et al. [1998, 2002] studied P-S conversions from
the 410 km and 660 km discontinuities and
discovered a maximum thinning of the transition
zone of about 20 km, slightly shifted to the south
in relation to the surface volcanic center. Shen et
al. [2003] also reported an anomalous region of
limited lateral extent at 1050 km depth (which
they also observed below Hawaii and which they
called a discontinuity), which might be regarded
as additional evidence for a deep origin of the
plume.
[13] For the lower mantle, global models show
only a weak low velocity beneath the North
Atlantic, and details of the structure are highly
variable. A low-velocity anomaly in the lower
mantle beneath the North Atlantic similar to the
‘‘plume conduit’’ seen by Bijwaard and Spakman
was only found in the P-wave tomography model
by Zhao [2001]. One reason for the rather strong
differences among tomography models of the lower
mantle is due to the fact that seismic travel time
residuals are often stronger in the upper mantle
compared to the lower mantle. For the BijwaardSpakman data set used here the maximum root
mean square velocity variations for the upper
400 km is 7.0%, for the depth interval 400 km to
1400 km it is 1.5%, and for the lower mantle
down to 2500 km it is only 0.74%. Similar
variations have been observed in other data sets
as well and have been explained by the increase of
the elastic moduli with depth. Since the variations
in the mid lower mantle are so small, the resolved
anomalies are strongly biased by assumptions used
10.1029/2006GC001359
for the inversion routine (e.g., damping). Furthermore, one has to keep in mind that the resolution
of most of the global models is relatively coarse
(e.g., smean is given for spherical harmonics 1–31;
the model by Mègnin and Romanowicz [2000] has a
horizontal resolution between 450 and 850 km).
Bijwaard and Spakman [1999] performed synthetic
resolution tests for their model and claimed an
overall resolution of the order of 300–500 km. As
they used a non-equidistant inversion grid, the
model has a somewhat higher resolution close to
seismic stations where the grid elements are
small. In general, pure body wave models may
have less resolution in the upper mantle away
from seismic stations compared to the lower
mantle.
[14] In summary, at least down to a depth of the
mid lower mantle most tomography models including smean [Grand, 2002; Ritsema et al., 1999;
Masters et al., 2000; Montelli et al., 2004; Becker
and Boschi, 2002] show a clear low-velocity
anomaly beneath the North Atlantic. Whether this
anomaly is related to a plume conduit and can be
traced to greater depth cannot be conclusively
answered on the basis of tomography models, yet.
[15] In addition, a global view of most tomography
models below 1400 km down to the core mantle
boundary, is dominated by two stronger lowvelocity anomalies further south, beneath western
and southern Africa, respectively. These anomalies
are important for global flow models and have to
be kept in mind when interpreting flow structures
in the North Atlantic (see below).
2.2. Mantle Flow Model for the North
Atlantic Based on Regional Modeling
[16] The mantle flow model for the North Atlantic
which is presented in this section is based on a
section of the global P-wave tomography model by
Bijwaard and Spakman [1999]. The section lies
between 49.8°N, 50°W and 85.2°N, 15.4°E with a
resolution of 0.6 degrees and is given in 26 nonequidistant depth layers. This section of the mantle
includes the above mentioned plume conduit
related to Iceland [Bijwaard and Spakman, 1999].
From seismic velocity anomalies within this section internal mantle densities and temperatures
have been inferred, assuming a constant conversion factor of 0.3 between density to seismic
velocity anomalies [Karato, 1993] and a relation
between density and temperature with a depthdependent thermal expansivity as described by
Schmeling et al. [2003]. This leads to a relation
5 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Table 1. Definition of Parameters Used in This Study
Symbol
Parameter
Value and Unit
a
g
m
r0
DrO – S
DrS – P
cv
fO/S
fS/P
g
KT
KT
thermal expansivity
Grueneisenparameter
mantle shear viscosity
mean mantle density
density jump at the olivine spinel transition
density jump at the spinel perovskite transition
specific heat at constant volume
fraction of material transformed from olivine to spinel
fraction of material transformed from spinel to perovskite
gravity acceleration
bulk modulus at constant temperature above 410 km
bulk modulus at constant temperature below 410 km
above 410 km
below 410 km
hydrostatic pressure
dynamical pressure
temperature
flow velocity
seismic P-wave velocity
K1
1.2
Pa s
4250 kg/m3
196 kg/m3
253 kg/m3
1.3 103 J/kg K
%
%
9.87 m/s2
135 109 Pa
200 109 Pa
6
3.5
Pa
Pa
°C, K
cm/a
km/s
dK
dP
dK
dP
P
p
T
u
vP
between seismic velocity and mantle temperature
of the form
@ ln T
1
KT ð P Þ 1
¼ 0:3
¼ 0:3
@ ln vp
aT
r0 cv g T
ð1Þ
The notations are defined in Table 1 together with
the numerical values of the parameters which are in
accordance with a PREM mantle [Stacey, 1992].
All parameters in equation (1) vary with pressure,
but the effect is strongest for the bulk modulus KT;
therefore all other parameters in equation (1) are set
constant, and for KT a relation of the form KT(P) =
KT0 + dK
dP P(z) is used. This approach leads to
buoyancy forces in the deep mantle which are
about a factor 4 smaller compared to the upper
mantle for the same excess temperature.
[17] The temperature anomaly, as derived from the
Bijwaard and Spakman [1999] tomographic model
is shown in Figure 3. The excess temperature of the
structure, defined as the plume conduit by Bijwaard and Spakman, is between 200 and 250K
throughout most of the mantle. Such an excess
temperature is well in agreement with temperature
estimates for the Iceland plume from different
observations and modeling approaches (for example, see the review by Ruedas et al. [2006]). From
geochemical studies on Icelandic lavas, Schilling
[1991] and Nicholson and Latin [1992] estimated a
plume excess temperature around 250K. Similar or
slightly lower excess temperature had been deduced from seismic velocity variations in the
uppermost mantle below Iceland [Allen et al.,
1999; Wolfe et al., 1997; Foulger et al., 2001].
Fully dynamical models of plume rise including
melting in the plume center [Ruedas et al., 2004;
Kreutzmann et al., 2004; Ito et al., 1999; Keen and
Boutilier, 2000] obtained somewhat lower excess
temperatures of 100 to 180K. Excess temperatures
of 200K for the lower mantle are low if compared
to fully dynamical global mantle convection models. Such models suggest that lower mantle excess
temperatures are higher than upper mantle excess
temperatures by a factor of 2–3 [Zhong, 2006].
This suggests that the plume like structures in the
lower mantle as inferred from the Bijwaard and
Spakman tomography model might not be strongly
coupled with the well resolved upper mantle
plume.
[18] Once the temperature field was estimated, it
was used as the driving force for a dynamic model
of mantle flow. As we only have a Cartesian 3-D
variable viscosity convection code available, we
transformed the temperature field to a rectangular
box. Using Mercator projection this box was
defined as a 4000 4000 2880 km box,
representing a region within a sphere bordered by
the latitudes and longitudes as given above. This
box was embedded in the central part of a larger
computational box of 16 000 16 000 2880 km
in order to minimize boundary effects and to
allow for larger-scale return flows. For this region
we solved the Navier-Stokes equation for
incompressible, variable viscosity flow
n h
io
0 ¼ r p þ r m ru þ ðruÞT
þ gðr0 a T þ DrOS fOS þ DrSP fSP Þez
ð2Þ
6 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 3. Three-dimensional view of tomographic data [Bijwaard and Spakman, 1999] converted to temperature
assuming a pressure-dependent thermal expansivity and projected on a Cartesian grid. Notice the uprising low
seismic velocity anomaly, originating at the CMB at a position beneath the southern tip of Greenland. This anomaly
rises, strongly northeastward inclined, through the lower mantle and intersects with the mantle transition zone beneath
the western European margin at the latitude of Great Britain. Through the transition zone itself, no clear continuation
could be identified, but a strong upper mantle anomaly is present beneath Iceland.
Notations are found in Table 1. This equation
contains the effect of olivine phase transitions in
the mantle; for the phase parameters we used
values according to Akaogi et al. [1989] and treated
the phase transitions numerically as described by
Marquart et al. [2000]. The thermal expansivity a
was assumed to be depth-dependent in the same
way as explained in equation (1). The model set up
also included a free slip lower and upper boundary
with a highly viscous lithosphere according to
30 Ma of cooling. An alternative mechanical
boundary condition would be a prescribed
kinematic condition with observed plate velocities.
However, this would imply exerting external forces
to the system in contrast to the stress free surface of
the earth. A draw back of the chosen boundary
condition is the non-plate like behavior of the
model near the surface. A flow model with
prescribed kinematic surface velocities will be
presented in the next section.
[19] Equation (1) was solved by a hybrid spectral/
FD method, described by Marquart et al. [2000]
and Marquart [2001]. For the temperature- and
pressure-dependent mantle viscosity we used an
equation of the type m(T, z) = m0w(z) ecT. The
depth-dependent function w(z) and the constant c
have been determined by fitting the geoid and
gravity data of the EGM96 potential field model
[Lemoine et al., 1998] for the North Atlantic for
a wavelengths range between 400 and 4000 km.
This fitting procedure is the topic of the study by
Marquart and Schmeling [2004] where also the
numerical modeling is explained. In this study
we tested various viscosity functions with gradual
or stepwise increases with depth and calculated
the flow model and the related gravity anomalies
for each of these viscosity distributions. We
found the best fit to the observed gravity anomalies for a model with a stepwise viscosity increase
by a factor 30 from the upper to the lower mantle
and with a moderate dependence on temperature
of 1 order of magnitude for 500°C temperature
variation.
[20] Most dynamic models for the Iceland plume
(as for plumes in general) are fully dynamic in the
sense that the Navier-Stokes equation and the heat
transport equation are solved simultaneously and
an idealized plume rise is studied during its tem7 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 4. Mantle flow field at various depths in the North Atlantic based on the temperature field of Figure 3 and a
viscosity profile with a stepwise increase by a factor 30 from the upper to the lower mantle and a dependence on
temperature of 1 order of magnitude for 500°C. Horizontal flow is shown by arrows (scale in lower figures for upper
(left) and lower (right) mantle) and vertical flow is color coded.
poral evolution [e.g., Ribe et al., 1995; Ito et al.,
1999; Ruedas et al., 2004]. In the approach by
Marquart and Schmeling [2004] a buoyancy field
was determined from a tomography model and the
dynamic response was calculated in a limited area.
Such approaches have been widely used for global
studies [e.g., Richards and Hager, 1984] but hardly
for regional studies and temperature-dependent
viscosity. One exception is the study of Mihalffy
et al. [2006], who dynamically combined global
flow models based on seismic tomography with a
regional convection model of the Iceland plume.
[21] In the following we discuss the mantle flow
field model by Marquart and Schmeling [2004]
(with the best fit to the observed gravity anomalies). In Figure 4, vertical flow is shown in color
coding and horizontal flow by arrows with length
scales different for the upper and lower mantle as
indicated in the two lowermost panels in the lower
right figure (note that the numerical grid was
denser and only every second vector is drawn for
clearer visualization). As the numerical grid was
four times larger as the area under investigation
most of the return flow occurred outside this area.
8 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
In the lower mantle the flow is very slow, less than
1 cm/a, and of long-wavelength nature, mainly
characterized by a broad upwelling. The flow in
the mantle transition zone shows considerably
more detailed small-scale characteristics than in
the mantle below and above, with upwellings in a
broad region around Iceland and beneath the northern ridge system. Above 300 km, the flow field is
dominated by horizontal flow. While Iceland is
situated approximately above the center of maximum vertical flow at the transition zone, it is not in
the center of radially divergent horizontal flux at
shallow depth. Below Iceland, horizontal flow is
mainly directed southward, changing from a N-S
direction in east Iceland to a more NE-SW directed
flow in the western part of the island. Horizontal
flow is mainly along-ridge for the Reykjanes
Ridge, and vertical flow from greater depth occurs
only in the part of this ridge which is adjacent to
Iceland. The Kolbeinsey Ridge, in contrast, is
dominated by divergent flow with strong vertical
flow throughout the upper mantle. In the light of
the idealized flow models shown in Figure 2 one
may conclude that the flow pattern is explained by
a combination of large-scale plume flux and
spreading flux to the north.
[22] The mantle flow field around Iceland has in the
past often been predicted on the basis of numerical
or analogue modeling of idealized plume ridge
interaction. Ribe et al. [1995] and Feighner et al.
[1995] studied the case of a plume, initiated under a
spreading ridge, and found that radial plume flux
normally predominates for slow spreading velocities as in the North Atlantic. Only in case of a very
strong temperature-dependent viscosity along ridge
channel flow superimposing plume flow could be
obtained [Albers and Christensen, 1996].
2.3. Mantle Flow Model for the North
Atlantic Based on Global Modeling
[23] In the dynamic model, described so far, all flow
components of global or very large scale nature have
been neglected, in order to allow more sophisticated
modeling of mantle parameters and to enhance
regional resolution. Furthermore, the model parameters, especially the pressure- and temperaturedependent viscosity law was scaled to fit regional
gravity observations. On the other hand, effects of
the far flow field have not been included. Therefore,
in the following test we modified the approach and
consider a global flow field in spherical harmonics
l = 1 to 30, deduced from a robust tomography
model (smean [Becker and Boschi, 2002]).
10.1029/2006GC001359
[24] We varied the free parameters of the model
(upper to lower mantle viscosity contrast, seismic
velocity to density scaling factor in the upper and
lower mantle, and a layering coefficient characterizing the permeability of upper/lower mantle interface) in order to maximize the fit between the
observed and modeled geoid at low degree
harmonics (l = 2–12). The optimum values of
the model parameters were determined by systematic exploration of the model space. The bestfitting model demands somewhat higher viscosity
increase from the upper to lower mantle of a factor
100 compared to a factor 30 found for the regional
model, and a scaling factor of 0.1 in the upper
mantle and 0.25 in the lower mantle. We note that
in case of the global model, we consider different
boundary conditions at the top boundary and the
upper/lower mantle interface than in the regional
model. The observed plate motion is imposed as a
kinematic boundary condition at the surface and
the permeability at the 660-km depth is modulated
by a free model parameter, called the layering
coefficient, the value of which is determined from
the geoid inversion. The layering coefficient allows
to study the whole range of partially layered flow
models, with a whole mantle flow model (layering
coefficient equal to 0) and a perfectly layered flow
model (1) as end members (for more details, see
Čadek and Fleitout [1999, 2003]). The inversion of
the geoid gives the value of the layering coefficient
close to 0.5 which means that the flow across the
upper/lower mantle boundary is reduced by 50% in
comparison with the whole mantle flow model.
[25] Zooming in to the North Atlantic region the
resulting flow field is shown in Figure 5a with the
net rotation toroidal component of degree 1 being
removed. Also the global model shows a pronounced upwelling flow in the upper as well as
lower mantle. The upwelling region within the
upper mantle has a maximum vertical velocity of
more than 6 cm/a and is elongated in SW-NE
direction, roughly following the North Atlantic
ridge system. The upwelling in the lower mantle
is less elongated in shape and reaches almost 4 cm/
a in the upper part. The horizontal flow in the
upper mantle is characterized by a superposition of
a ridge-perpendicular and a radial divergence flow
component (compare Figures 2a and 2c), on which
a large-scale northerly flow is superimposed. In
contrast, the lower mantle shows a large-scale
southerly flow, which is strongest in the lower part
of the lower mantle. These large-scale upper and
lower mantle flow components seem to correspond
9 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 5. Mantle flow field at various depths in the North Atlantic from the globally derived flow field based on the
tomography model smean and a global fit of the geoid. The optimal model has a viscosity profile with a stepwise
increase by a factor 100 from the upper to the lower mantle and a layering coefficient of about 0.5. Horizontal flow is
shown by arrows (scale below the figures, different for upper and lower mantle), and vertical flow is color coded.
(a) Flow model in spherical harmonics 1 to 30 with the toroidal mode l = 1 removed. (b) High pass filtered flow
model with spherical harmonics between 9 and 30.
to the large-scale upwelling flow induced by the
African superplume.
[26] It is interesting to note that the seismic lowvelocity anomalies identified by the tomography
model smean show a lateral shift between the
upper and lower mantle: As can be seen in
Figure C1 of Becker and Boschi [2002], the upper
mantle anomaly is centered beneath Iceland, while
the (weaker) anomaly in the upper part of the lower
mantle is shifted toward the SE, similarly to the
Bijwaard and Spakman [1999] model. In contrast,
the upwelling flow in the upper and lower mantle
does not show this lateral shift (Figure 5a).
[27] For a better comparison with the regional flow
model, Figure 5b shows a high pass filtered version
of the flow field of the global model (harmonic
degrees 1 to 8 are subtracted). The upwelling flow
structures within the lower and upper mantle are
10 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 5. (continued)
even more isolated compared to the unfiltered case
of Figure 5a, but the flow velocities are damped.
The upwelling flows in the upper part of the lower
mantle and in the upper mantle are clearly interrelated and may be associated with the Iceland
plume. The horizontal component of the high pass
filtered flow (Figure 5b) is considerably different
from the unfiltered field of Figure 5a: the largescale northerly flow within the upper mantle and
southerly flow within the lower mantle vanished.
Without this mantle wind, the horizontal flow
clearly converges and feeds the elongated upwelling region at mid lower mantle depth, and diverges
from a point near western Iceland in the sublitho-
spheric upper mantle. It is interesting to note that
the horizontal feeding flow takes place at 1500 km
depth and not near the core mantle boundary.
2.4. Comparison of Flow Fields Based on
Regional and Global Modeling
[28] We now compare the characteristic features of
the flow fields based on regional and global
modeling (Figure 4 and Figures 5a and 5b).
[29] Starting with the lower mantle, the vertical
upwelling flow structures are remarkably similar,
both strongest in the upper part of the lower
mantle, and situated at almost identical locations.
11 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
However, lateral feeding of the upwelling flow
(i.e., the ‘‘source region of the plume’’) reaches
down to greater depth in the regional model
(compare 1500 and 2540 km depths panels of
Figure 4 and Figure 5b). As expected, the southerly
mantle wind in the global model (Figure 5a, depths
1500 km and 2540 km) is not present in the
regional model. It is also remarkable that despite
significant differences in model assumptions (laterally variable versus stratified viscosity, 30 versus
100 times viscosity jump between upper and lower
mantle, an olivine-spinel-perovskite phase transition versus the assumption of a layering coefficient), both models show an increase of upwelling
velocity between lower and upper mantle by a
similar factor of about 1.5.
[30] It is interesting to note that the regional model
shows a lateral shift between the upper and lower
mantle upwellings, associated with a horizontal
flow component near 660 km depth (Figure 4),
while in the global model this shift is not observed
(Figure 5b). This is surprising, as both tomography
models show such a shift. We attribute this different behavior at least partly to the lateral variation of
the viscosity in the regional model, which more
effectively focuses the upwellings into the hot, low
viscosity regions.
[31] Comparing the upper mantle flow structures
clearly show the different resolution of structures:
in the regional model different upwelling regions
beneath Iceland and beneath the Kolbeinsey ridge
can be distinguished, which seems to be beyond
the resolution of the global model. The different
surface boundary conditions (kinematic plate
velocities versus free slip) are also clearly visible
when comparing the 200km and 100 km depth
panels of Figure 4 and Figure 5b: In the model with
kinematic boundary conditions the high asthenospheric velocities drop to the slower plate velocities on top. This drop is associated with a strong
vertical shear flow, which has a considerable effect
on the LPO (see below). In contrast, in the free slip
model surface velocities are higher, and no vertical
shear flow is visible. Below 100 to 200 km depth
the effect of the difference in surface boundary
conditions becomes small.
[32] It is also interesting to compare our flow field
models with the flow field for the North Atlantic of
Mihalffy et al. [2006]. As this is also a global
flow model it closely resembles the flow field of
Figure 5a, and shows all features discussed above
such as the strong upwelling through the lower and
upper mantle centered beneath Iceland and the
10.1029/2006GC001359
northerly mantle wind induced by the large-scale
upwelling beneath South Africa.
3. Anisotropy in the
North Atlantic Region
[33] In this section seismic anisotropy models
based on seismic observation will be compared to
seismic anisotropy as predicted by the different
flow models.
3.1. Seismic Anisotropy Models
[34] Early studies on seismic anisotropy gave inconsistent results for the North Atlantic regions.
The map by Montagner and Tanimoto [1991]
indicates a weak frequency-(depth)-dependent anisotropy with a predominantly NW-SE fast seismic
direction for the entire region. However, another
global study on azimuthal seismic anisotropy using
long-period SS - S times [Woodward and Masters,
1991] did not give indication for upper mantle
anisotropy in the North Atlantic. Yang and Fischer
[1994] analyzed SS arrivals for shear-wave splitting in and found some indication of upper mantle
anisotropy beneath the Atlantic, with the inferred
orientation for the central North Atlantic in
approximate agreement with the global anisotropy
model of Montagner and Tanimoto [1991]. A
newer global study by Ekström [2001] shows for
50s to 150s periods (maximum sensitivity between
80 and 250 km) also a constant WNW-ESE direction for the region around Iceland, but their resolution does neither allow to reveal any details since
azimuthal anisotropy studies on a global scale only
provide the averaged effect over large areas.
[35] More recently an inversion based on Rayleigh
waves was done by Debayle et al. [2005] and by
Pilidou et al. [2005] to infer seismic wave speed
anomalies including anisotropy for the upper
500 km of the mantle. They found also an anisotropy
field in NW-SE direction over most of the North
Atlantic but rotation to nearly E-W slightly north of
the Charlie-Gibbs Fracture Zone (southward of a
latitude of 56°N) and north of 82°N, and a
predominately N-S direction across the Greenland
shield.
[36] Another observation based anisotropy field
has been provided by Jeannot Trampert [Trampert
and Woodhouse, 2003] on the basis of Rayleigh
waves of 40s period with resolution of spherical
harmonic degree of 8. This model generally indi-
12 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
cates more WNW-ESE directions at latitudes south
of Iceland even across the Greenland Shield.
[37] Better lateral resolution than the global models
can be expected from a surface wave dispersion
study of a large cap around the arctic by Levshin et
al. [2001]. For a period of 50s their map indicates a
general W-E to WNW-ESE fast seismic direction
and widely agrees with the global studies, with the
exception that they report a clear N-S fast direction
in the northern part of the Greenland shield. For a
100s period, which is better to compare with our
anisotropy model, their map indicates considerably
more spatial variations, with a NW-SE direction
north of Iceland up to the Knipovich Ridge.
[38] In conclusion still considerable differences are
present between the anisotropy directions in the
North Atlantic published by various authors. One
reason for this might be the sensitivity of the
inversion to the damping factor used, as discussed
by Levshin et al. [2001].
[39] Focussing on Iceland, a number of seismic
anisotropy measurements have been done. Anisotropy observations on Iceland at very shallow (crustal)
depth [Menke et al., 1994] show ridge parallel
anisotropy in the SW. However, one should keep in
mind that these shallow layers cannot be compared to
our model and might be related to crustal accretion
processes. New detailed S-wave anisotropy studies
probing the upper mantle [Bjarnason et al., 2002; Li
and Detrick, 2003; Xue and Allen, 2005] neither
reflect a spreading-parallel nor a radial pattern
across Iceland. Bjarnason et al. [2002] observed a
fast direction in NNW-SSE direction in eastern Iceland and of NNE-SSW direction in western Iceland.
[40] Li and Detrick [2003] studied Rayleigh-wave
anisotropy and observed a few flips in the orientation of the fast axis beneath W-Iceland above a
depth of about 60 km. These different anisotropy
directions mainly in the thick crust are explained as
being due to the combined effect of fossil spreading ridge jumps and a frozen-in crystal orientation
and cannot be linked to the numerical flow model.
At a depth greater than 80 km (better at periods
greater than 60s), Li and Detrick [2003] also found
a NE-SW fast direction in western Iceland, and an
indication for NW-SE direction in eastern Iceland
though the signal is more unclear. They also
reported only weak seismic anisotropy beneath
central Iceland at a depth greater 50 km, which
they interpreted as an indication that buoyant
upwelling may control the orientation of crystals.
Though weak, Li and Detrick [2003] found the
10.1029/2006GC001359
anisotropy direction in this central part to be
consistently rotated by 90° to the directions
below the other parts of Iceland, being nearly
perpendicular to the rift axis. While this strong
variation in azimuthal anisotropy on small scale is
difficult to explain by mantle flow, it might reflect
processes of different kind, such as the presence of
a melting zone at a depth of 80 to 100 km with
water release, which would lead to an alignment of
the b-axis of olivine with the flow direction and the
a-axis perpendicular [Mizukami et al., 2004]. An
alternative model has been proposed by Ruedas
and Schmeling [2006] in which the change in
anisotropy is explained by a dynamically controlled change in deep dyke or channel orientation.
[41] On the basis of the above discussion we
compare the Levshin et al. [2001] anisotropy model
with the predictions of our regional flow model.
Although the resolutions are quite different, it is
still worthwhile to compare our predictions with
the findings for Iceland by Li and Detrick [2003].
On the other hand, it is consistent to compare our
global flow model with a global rather than a
regional anisotropy model. Therefore we choose
the recent model of Debayle et al. [2005] as the
most appropriate mode for comparison.
3.2. Mantle Flow and Seismic Anisotropy
[42] From laboratory and theoretical studies it is
well known that plastic deformation of mantle
rocks may lead to strain induced lattice preferred
orientations (LPO) of olivine crystals (alignment of
the a-axes) which results in seismic anisotropy
[e.g., Nicolas and Christensen, 1987; Mainprice
et al., 2000]. As LPO is a function of the finite
deformation associated with mantle convection,
and the long axis of the finite strain ellipsoid
(fse) often points into the flow direction, seismic
anisotropy has in the past often been interpreted as
a direct image of flow directions [Tanimoto and
Anderson, 1984] and a measure for mantle convection [e.g., Ribe, 1989; Russo and Silver, 1994;
Tommasi, 1998; Blackman and Kendall, 2002;
Becker et al., 2003]. However, this is only a rough
approximation, since the fse may locally strongly
depart from the flow directions, e.g., near stagnation points [McKenzie, 1979], and LPO and the
long axis of the fse only converge after sufficiently
high strain [Kaminski and Ribe, 2002]. In the
models presented below, the differences and similarities between flow direction, the long axis of the
fse and fast axis of LPO will be discussed for a
particular flow case.
13 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
[43] We determined the long direction of the fse as
well as the LPO for an assemblage of olivine grains
with three slip-planes undergoing straining. We
followed the formalism by Kaminski and Ribe
[2001], who determined this direction in a statistical sense, following the assemblage of initially
arbitrarily oriented olivine grains along various
streamlines. For each grain three perpendicular
slip-planes were assumed and each can deform
by intra-crystalline slip [Ribe and Yu, 1991] and
dynamic recrystallization. Dynamic recrystallization depends on the dislocation density on each
slip plane which is a function on the applied stress
and the orientation of the slip plane. Dislocation
creep may lead to grain boundary migration and
formation of stress free sub grains and by that
changing the overall elasticity tensor of the various
aggregate assemblages. The detailed formalism as
well as the various parameters are described in a
number of papers [Kaminski and Ribe, 2001;
Kaminski, 2002; Kaminski and Ribe, 2002] and
will not be repeated here.
[44] The flow field as shown in Figure 4 was
assumed to be steady state (for a time period of
10 Ma). Further assumption for the anisotropy
calculations are: 100% olivine grains, and initial
positions of the olivine assemblages not deeper than
420 km, and to result in a finite strain of at most 10
at the positions where the LPO is determined. The
magnitude of anisotropy may be overestimate
because the mantle also contains other minerals.
3.3. Comparison of Modeled and Observed
Seismic Anisotropy for Regional Models
[45] From the flow model shown in Figure 4 it is
obvious that horizontal flow dominates above
300 km depth, except for the area around Iceland
and along the Kolbeinsey Ridge, where strong
uprising flow is present with about the same
magnitude as horizontal flow. This is consistent
with the observation by Montagner and Tanimoto
[1991], who studied anisotropy on a global scale
and showed that azimuthal anisotropy is important
to depths of 300 km. Therefore we determined
the anisotropy at a depth of 75 to 175 km, which is
just below the lithosphere. The longest axis of the
fse and the locally averaged LPO directions are
given in Figure 6 for a depth of 175 km. The length
of these axes represents the natural strain (ln(a/c)) in
case of fse and the percentage of anisotropy (non
isotropic part of the elasticity tensor) for LPO. The
vertical components of these axes are coded by
color, the horizontal components are shown as bars.
10.1029/2006GC001359
[46] The preferred directions for the fse and the
LPO in Figure 6 are quite different. The long axis
of the fse as well as the fast axis of LPO are
determined by the accumulated strain along the
streamlines above 420 km and, in case of the LPO,
by the ability of olivine slip planes to adjust to the
strain field changing along the streamline. In the
central upwelling region the flow is characterized
by vertical uni-axial compression, leading to horizontal long strain axes with arbitrary directions,
although a tendency of radially diverging fse
directions is visible close to Iceland. In the peripheral area the flow is still directed radially outward,
but it slows down with increasing distance from
the plume. As expected this flow type produces
flow-perpendicular stretching with the long fse
axes tangential to the center of upwelling. This is
clearly visible in the western part of the model. A
further flow component of our model is associated
with a radially outward directed, horizontal shear
flow (at shallow depth the radial flow velocities are
faster than at greater depth). Because dynamic
recrystallization leads to faster adjustment of the
preferred directions in respect of the macroscopic
flow field [Kaminski and Ribe, 2002], the LPO
adjusts faster to this radial shear flow than the fse.
As a result the horizontal flow vectors and the fast
azimuthal directions of the fast axis of olivine point
in similar directions, however, in areas of strong
vertical flow large deviations are possible and even
for some areas outside strong upwelling (e.g.,
south of Iceland) deviations up to 90 degrees are
possible.
[47] The general appearance of the anisotropy field
is characteristic for a large-scale upwelling, with its
center north of Iceland. The area to the north of
Iceland and along the northern ridge system, where
the flow is essential vertical, is characterized by
weak horizontal anisotropy of varying direction. In
the Atlantic south of Iceland, azimuthal anisotropy
changes from NW-SE close to the European continent to NE-SW across the Reykjanes Ridge. The
anisotropy model, presented here, would further
predict a mainly NW-SE direction of fast seismic
velocity in the mantle below the eastern Greenland
shield, and weaker, varying directions in the western and southern part. Variations around an E-W
anisotropy direction are also found in the southerly
adjacent oceanic region.
[48] We now compare our modeled fast LPO
directions to those obtained by Levshin et al.
[2001] for moderate damping. We relate Rayleigh
wave periods of 50s to a depth for the LPO
14 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 6. (bottom) Regional mantle flow model at a depth of 150 km, (middle) long axis of the finite strain
ellipsoid, and (top) LPO of olivine at a depth of 175 km based on the 3-D flow model. vertical components are
indicated by color.
direction of 75 km and 100s to a depth of 125 km
(Figures 7a and 7b). Levshin et al. [2001] found
that a high damping factor leads to a preference of
the long wavelengths parts of the field. Since in our
regional flow model, the maximum wavelength is
restricted to 1600 km, and no global wide largescale flow is present, it is more consistent to
compare our modeled anisotropy directions to
regional anisotropy studies with a moderate damping factor.
[49] For shallow depth (Figure 7a) both, modeled
LPO directions and seismic anisotropy directions
are more irregular and show stronger small-scale
variations. In Figures 7a and 7b we have also
marked in light gray the regions where the modeled
azimuthal directions of the fast LPO and the
15 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 7. Horizontal component of LPO directions of the regional flow model (black bars) and azimuthal
anisotropy (red bars) in the North Atlantic region based on Rayleigh waves after Levshin et al. [2001] for periods of
(a) 50 s and (b) 100 s. Regions where the fit is better than 45° are indicated in light gray.
observed seismic anisotropy directions fit better
than 45°.
[50] For both periods a robust feature is a good
agreement of the azimuthal directions in large areas
north, east and SE of Iceland. No coincidence
between the azimuthal directions for both periods
is found south of Iceland in a narrow zone along
the northern end of the Reykjanes Ridge, in the
eastern part of the North Sea, in the oceanic area
south of Greenland, and to the east of the CharlieGibbs Fracture Zone.
[51] Along the Reykjanes Ridge our modeled LPO
directions are mainly ridge parallel, in agreement
with the predicted along ridge shallow flow. The
seismic anisotropy directions show at most in the
southern part of the Reykjanes Ridge a ridge
16 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 8. Horizontal component of LPO directions of the regional flow model (black bars) and azimuthal
anisotropy (red bars) in the area of Iceland based on Rayleigh waves after Li and Detrick [2003] for periods of (a) 50 s
and (b) 67 s.
parallel direction, but closer to Iceland clear ridge
perpendicular directions are found. Roughly ridge
perpendicular directions are found for the ridges
north of Iceland, both, in our modeled LPO directions and the seismic anisotropy.
[52] Even though the regional numerical flow
model does not resolve any lateral variations on
length scales much smaller than 100 km, a comparison to anisotropy observations on Iceland
might be of interest.
[53] Figures 8a and 8b show a close-up of the
modeled horizontal components of the LPO vector
around Iceland at 75 and 125 km depth together
with the azimuthal seismic anisotropy directions
obtained by Li and Detrick [2003] for Rayleigh
wave periods of 50s and 67s. Since Iceland is not
directly in the center of the large-scale diverging
flow, which is about 150 km north of Iceland, the
LPO fast directions change only moderately across
Iceland from NNW-SSE direction on the eastern
side to more NE-SW direction on the western part.
By comparing our fast LPO directions to the
seismic anisotropy directions of Li and Detrick
for 50s (Figure 8a), we found a good agreement
between the two data sets. For 67s of Rayleigh
17 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 9. Horizontal component of LPO directions of the global flow model (black bars) and azimuthal anisotropy
(red bars) in the North Atlantic region based on Rayleigh waves after Debayle et al. [2005].
wave periods however, the eastern part of Iceland
behaves quite differently. However, by performing
a sensitivity study, Li and Detrick stated that their
results in the eastern part of Iceland are not robust
while the NNW-SSE directions in the western part
are reliable.
[54] The travel time differences between the fast
and the slow S-wave which they observed is larger
in the eastern part of Iceland and had been interpreted as due to a 100–200 km thick anisotropic
layer under E-Iceland. Though the fast directions
they observed are very similar to the fast olivine
directions modeled here and may very well be
explained by a slight rotation, they defined two
different anisotropic domains with a divide to be
located about 100 km west of the neo-volcanic
zones and interpreted this finding as an indication
of a former spreading center still conserved due to
uncompleted crystal reorientation.
3.4. Comparison of Modeled and Observed
Seismic Anisotropy for the Models
[55] On the basis of the globally derived flow field
we also determined seismic anisotropy directions
and compared them to global observation based
anisotropy directions [Debayle et al. [2005]. Their
inversion is based on surface wave observations
and is done for seismic anomalies including anisotropy. While it would be consistent to use their
data on seismic anomalies also for the tomography
model, this can hardly be done, since their data set
is rather shallow (down to 500 km) and not suitable
for estimation of deep mantle density anomalies.
While the model parameters are chosen (or better
found) in a way to give the best fit to the longwavelength geoid, the fit to the anisotropy is
estimated for the North Atlantic region only. Our
best model is shown in Figure 9, where the black
bars indicate the modeled anisotropy and the red
bars indicate the observations. The overall fit is
33.7°, which is not too bad, since inspecting
Figure 9 in more detail, it becomes obvious that
the misfit is mainly attributed to the Greenland
Shield Area and the northern part of Fennoscandia
and the adjacent Barents Sea. For the Greenland
shield the azimuthal anisotropic in the upper
200 km might not be related to present mantle
flow, nor are the density and viscosity conditions
for this thick shield properly modeled in our
numerical approach.
[56] As for the regional model a good fit is found in
a broad region of the North Atlantic north, east and
SE of Iceland, as well as for Iceland itself. In the
area NE of Iceland directions are roughly perpendicular to the ridge system. As in the regional flow
model (Figures 7a and 7b), a significant region
associated with the Reykjanes Ridge area south
west of Iceland shows a pronounced misfit.
3.5. Discussion of the Anisotropy Results
[57] Although the regional and global flow models
show satisfactory agreement in their flow fields
18 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
(see section 2.4), their predictions for LPO are
more different. While both approaches are characterized by circular or elongated regions of upwelling and diverging flow beneath Iceland, only the
regional model shows divergent LPO directions,
while the global model is more characterized by a
roughly ridge-perpendicular LPO direction. We
attribute this difference to (1) the higher sensitivity
of the LPO to strain rates rather than flow directions, in particular to the vertical shear strain rate at
shallow depth associated with the imposed kinematic boundary condition which is absent in
the regional free slip model (compare flow fields
shown in Figures 5 and 7), and (2) with the NE-SW
elongation of the upwelling region present in the
global model. In the Atlantic region northeast of
Iceland most seismic anisotropy models roughly
agree on a ridge perpendicular fast direction, which
we found for both, our global and our regional
model.
[58] The comparisons shown in sections 3.3 and 3.4
reveal a few robust features: Fast directions of the
seismic anisotropy models and both flow models
agree well in large sections to the north, east, and SE
of Iceland. We therefore conclude that in these
regions the observed seismic anisotropy is caused
by LPO with fast a-axes oriented in NW-SE directions, induced by the North Atlantic upwelling flow
associated with the Iceland plume and the thermal
anomalies beneath the Kolbeinsey Ridge.
[59] Another robust feature seems to be the pronounced misfit for the Greenland shield and SW of
Iceland including the Reykjanes Ridge (compare
Figures 7a, 7b, and 9, as well as in other comparisons not shown here). One explanation might be
that in these regions the mechanisms producing the
observed azimuthal anisotropy may be different
from the one used in our model. For the Greenland
shield seismic anisotropy down to the base of the
thick lithosphere at about 200 km depth may not
exclusively be generated to the ongoing mantle
flow but may be inherit from older dynamic
processes. For the Reykjanes Ridge and closely
adjacent regions the presence of water may be
important. In the presence of water olivine aggregates have been found to exhibit B-type anisotropy,
in which the fast a-axes are oriented perpendicular
to the shear flow within the shear plane [Kneller et
al., 2005]. Thus, if applied to our regional flow
model (Figures 7a and 7b), fast axes in such water
bearing mantle regions would have to be flipped by
90°. Because for the Kolbeinsey Ridge the agreement between observed and modeled anisotropy
10.1029/2006GC001359
does not require such a flip, we might speculate
that we have B-type anisotropy south of Iceland
and A-type anisotropy north of it. This would
suggest that high degree melting beneath the Kolbeinsey ridge would have extracted all the water,
while it might be still present in the mantle beneath
the Reykjanes ridge. We will address the issue of
different temperatures within the mantle regions
north and south of Iceland in the following section.
4. Comparison to Geochemical
Observations
[60] It might be of some interest to relate our
findings on temperature distribution and mantle
flow to some geochemical data obtained by drilling
or dragging along the North Atlantic Mid-Ocean
Ridge north of the Charlie-Gibbs-Fracture Zone
[e.g., Hart et al., 1973; Schilling, 1973; Mertz et
al., 1991; Devey et al., 1994; Hanan et al., 2000].
[61] First we summarize our main results for the
regional temperature and flow model (Figure 4 and
Figure 10) and then relate them to geochemical
observations: We found a strong temperature
anomaly below Iceland and the northwesterly adjacent oceanic region (Figure 10, 100–630 km)
which drives an upwelling flow and which seems
to be weakly connected to a deeply rooted lower
mantle anomaly southeast of Iceland (light upper
part of the lower mantle anomaly visible in
Figure 3). In our regional flow model the upwelling
region of the lower mantle feeds material to the
upper mantle by an obliquely northwest oriented
flow (see the horizontal flow component at 680 km
depth in Figure 4). A second pronounced temperature anomaly is located in the upper mantle at a
depth of about 200–300 km below the Kolbeinsey
and southern Mohns Ridge (Figure 10). This
anomaly has no direct deep seated root, although
it might be weakly connected to the first anomaly
and its lower mantle feeding system (Figure 4).
(It should be noted that the flow field represents
only a snap shot of a time-dependent flow, and
statements about feeding and ‘‘flow connections’’
must be handled with care). A super adiabatic
temperature of 200 K was estimated using equation (1) for the Iceland centered anomaly and a value
of 120 K for the Kolbeinsey-Mohns Ridge anomaly. Temperature anomalies along the Reykjanes
Ridge exist too, but are shallower than 100 km and
the super adiabatic temperature increase is less than
100 K. These temperature anomalies should lead to
excess melting, thick crust and shallow bathymetry.
A thick crust in excess of 24 km has not only been
19 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
10.1029/2006GC001359
Figure 10. Temperature anomaly in the upper mantle beneath the North Atlantic derived from the tomography
model of Bijwaard and Spakman [1999] and the seismic velocity temperature relation of equation 1.
20 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
repeatedly reported for Iceland [e.g., Darbyshire et
al., 2000] but is also well established for the
adjacent ridges. Both the Kolbeinsey and the
Reykjanes Ridge show some morphological features which are unusual for slow spreading ridges:
overlapping spreading centers, absence of a median
valley and a ridge topography untypical for the bulk
part of the Mid-Atlantic Ridge System [Sinha et al.,
1998; Kodaira et al., 1997]. All these features
indicate a high production rate of magma [Kodaira
et al., 1997] and a crustal thickness in excess of
about 9 km [Detrick et al., 1995]. For the Reykjanes
Ridge at 61°4400N, Searle et al. [1998] found a
thickness of the crust of 11.2 km and Kodaira et
al. [1997] observed a crustal thickness at the
Kolbeinsey ridge of 10 km at 70°2000N, which
corresponds with a water depth of not more than
500 m [Moeller, 2002].
[62] On the basis of Hf and Nd isotopes ratios and
trace element data from various mid-ocean ridge
basalts, Salters [1996] found indications that the
onset of melting for the Kolbeinsey Ridge is deeper
than for normal oceanic ridges. Low Na2O and
high FeO contents have been observed for Kolbeinsey Ridge basalts [Moeller, 2002; Humler and
Besse, 2002], being a clear hint for a high degree of
partial melting over a large depth range of a
depleted source. Corresponding temperature estimates of 1460°C for Kolbeinsey Ridge basalts at
melting [Humler and Besse, 2002] are in fairly
good agreement with the temperature of 1440°C
below the Kolbeinsey Ridge at 250 km depth,
which we estimated for our flow model assuming
an adiabatic background temperature of 1320°C at
that depth. Extensive melting and an anomalously
high temperature along the Kolbeinsey Ridge is
also supported by the observation of a high depletion in rare earth elements (REEs) [Schilling et al.,
1983; Haase and Devey, 1994; Devey et al., 1994].
Since REEs are incompatible with mantle rocks,
they become extracted with the first melts produced, thus further melting only ‘‘sees’’ an increasingly depleted source. Kolbeinsey Ridge basalts
are even more depleted in the light REEs than
Reykjanes Ridge basalts. This extreme depletion
has even been interpreted as being produced by the
highest degrees of partial melting found anywhere
in the world’s oceans [Klein and Langmuir, 1987].
[63] Another important question addresses the
material supply from the lower to the upper mantle
beneath Iceland and the amount of deep mantle
material injected into the ridge system. Our flow
model suggests that at least in the vicinity of
10.1029/2006GC001359
Iceland the deep mantle upwelling would lead to
deep mantle tracers exposed in extruded basalts.
On Iceland, a plateau-shaped maximum of 3He/4He
is well established [e.g., Breddam et al., 2000], and
commonly regarded as an indication for at least
some deep mantle material in the melting source.
The detailed compositions of basalts on Iceland,
however, have been found to be very heterogeneous indicating that beside a common MORB
source a number of non-MORB sources are
involved [e.g., Hanan and Schilling, 1997; Hanan
et al., 2000; Stracke et al., 2003; Thirlwall, 1995;
Fitton et al., 1997, 2003; Kempton et al., 2000].
Our flow model supports the common geochemical
view that the latter sources might at least partly
originate from the lower mantle.
[64] Since in the shallow upper mantle an essentially radial flow is predicted around Iceland and to
the south of it, ridge parallel flow is a consequence
along the Reykjanes Ridge. Thus we expect a
decreasing influence of deep mantle material tracers away from Iceland along Reykjanes Ridge
rather than a sharp transition. This is in agreement
with the steady decrease of several geochemical
observables [see Murton et al., 2002] along the
ridge away from Iceland, being characteristic for
ocean island basalts or deep mantle influence, such
as the La/Sm ratio [Schilling, 1975], or the ratio of
3
He/4He and 22Ne/21Ne [Dixon et al., 2000].
[65] For the Kolbeinsey Ridge our model predicts
divergent flow driven by a separate upper mantle
heat source with at most only a weak connection to
the deep mantle source to the south. Thus we
expect a geochemical signature for Kolbeinsey
basalts typical for extensive MORB melting, but
deep mantle tracers should be essentially absent.
Very much in line with our model Mertz et al.
[1991] could not find MORB-plume mixing indications north of the Tjoernes Fracture Zone (TFZ)
along the Kolbeinsey Ridge. In contrast to basalts
from the Reykjanes Ridge south of Iceland, close
to the TFZ a sharp decrease in the La/Sm ratio is
observed.
[66] A number of studies [Macpherson et al., 1997;
Botz et al., 1999; Breddam et al., 2000] investigated
noble gas concentrations in samples from the
Kolbeinsey Ridge and generally observed remarkably constant isotope ratios for He, Ne, and Ar
along the ridge with the exception of a maximum
in 3He/4He ratio at the Tjoernes Fracture Zone.
However, while Ne and Ar isotope ratios do not
indicate a deep mantle source, 3He/4He values
are significantly greater than typically found in
21 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
N-MORB. The obviously high degree of melting
and deep melt source along Kolbeinsey Ridge has
been interpreted in the past as due to the proximity
of the thermal anomaly associated with the Iceland
mantle plume. However, the absence of latitudinal
variations for most geochemical trace elements
do not indicate a gradual plume-MORB mixing.
Schilling et al. [1999] presented a review about Pb,
Nd, Sr, and He systematics along the Kolbeinsey
Ridge and claimed to see some plume mixing
across the TFZ and defined the boundary of the
plume influence about 300 km north of Iceland,
which would even fit better to our flow model (see
Figure 4).
[67] The superadiabatic temperature of about 120K
at 400 km and above which we estimated from the
Bijwaard and Spakman [1999] tomography model
for the Kolbeinsey and the southern Mohns Ridge is
in agreement with a deep melt source with a high
degree of melting. The related flow model shows
nearly uniform upwelling along the Kolbeinsey
Ridge at depth and ridge perpendicular flow at
shallow depth which fits to the observation of hardly
any along ridge variations in basaltic composition
and noble gas isotope ratios. However, whether the
relatively high 3He/4He ratio found along Kolbeinsey
is indeed a tracer of deep mantle material and may be
explained by a lateral connection to the Icelandic
anomaly remains an open question.
5. Discussion and Conclusion
[68] Dynamic models of temperature anomalies and
mantle flow in the North Atlantic region between
the Charlie-Gibbs and Jan-Mayen Fracture Zones
based on the regionalized P-wave tomography
model of Bijwaard and Spakman [1999] and on
the global, long-wavelength S-wave tomography
model smean [Becker and Boschi, 2002] were
presented and compared. The regional temperature
model predicts an anomaly throughout the upper
mantle below Iceland and beneath the westerly
adjacent section of the mantle and vanishes at the
Greenland margin and a second anomaly between
400 and 100 km depth below the Kolbeinsey and
partly Mohns Ridge. The long-wavelength model
smean does not resolve these features, but shows a
strong and broad low-velocity anomaly in the upper
mantle of the North Atlantic centered at Iceland.
From the regional flow model we deduce that the
resulting mantle flow has a strong upwelling component beneath Iceland of more than 4 cm/a,
however, beneath Iceland the flow is oblique in
S or SW direction, turning horizontally at shallow
10.1029/2006GC001359
depth into ridge parallel direction along the
Reykjanes Ridge. In contrast, the Kolbeinsey Ridge
is subjected to strong vertical flow diverging ridge
perpendicular above 200 km. The global flow
model is dominated by a NE-SW elongated upwelling region associated with a ridge perpendicular
component all over the North Atlantic. If the
idealized flow models on Figure 2 are recalled,
we conclude that the flow in the asthenosphere
beneath the North Atlantic can be characterized
by a large-scale radial plume flux around Iceland,
superimposed with a ridge parallel component in
the SW-parts of Iceland and the Reykjanes Ridge,
and a larger-scale ridge perpendicular component
being a consequence of plate divergence and the
large-scale NE-SW elongated upwelling flow in the
upper mantle (see Figure 5).
[69] Pipe-like flow of plume material along the
Reykjanes Ridge rift axis south of Iceland may
also occur in addition to the flow model presented.
Such a flow type has been proposed on the basis
of the observation of geochemical signature and
V-shaped pattern in bathymetry and gravity anomalies south of Iceland first by Vogt [1971] and been
modeled by Albers and Christensen [2001] and Ito
[2001], but the model approach used here cannot
resolve such small-scale features. Along ridge
channel flow demands a localized strong viscosity
reduction and will thus be largely decoupled from
the (upper) mantle wide circulation. Therefore
along ridge channel flow might simply add to the
flow field shown in Figure 4.
[70] A strong seismic low-velocity anomaly in the
upper mantle directly below Iceland is seen in the
large-scale smean [Becker and Boschi, 2002] and in
the Bijwaard and Spakman [1999] model and even
clearer in regional tomography models. Wolfe et al.
[1997] used the data obtained in the ICEMELT
experiment [Bjarnason et al., 1996] and detected
an approximately circular vertical structure between
100 and 400 km centered beneath Vatnajökull, and
estimated its radius to about 150 km and the excess
temperature to be 200K which is in very good
agreement with the estimates obtained in our study
(see Figure 3 and Figure 10). However, this anomaly of limited size, though hot, is not large enough to
drive the bulk of the upper mantle flow. In the view
of our flow models, the plume below the North
Atlantic rooting down (see Figure 4) into the lower
mantle is a large feature of the upper mantle possibly centered to the west of Iceland, whereby the
anomaly below Iceland is a hot, but smaller side
branch. Rapid vertical flow within the sub Iceland
22 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
anomaly is likely, but beyond the resolution of our
model. The upper mantle anomaly is likely to be fed
by a lower mantle upwelling centered east of Iceland, implying a significant horizontal flow components near 660 km depth as seen in the regional flow
model.
[71] In summary the following conclusions can be
drawn from our dynamic flow models:
[72] 1. Low seismic velocity anomalies in the
upper mantle and upper part of the lower mantle
beneath Iceland as seen by the regionalized P-wave
model of Bijwaard and Spakman [1999] and the
global S-wave model smean [Becker and Boschi,
2002] are associated with regional-scale upwelling
flows at least down to mid lower mantle depth.
[ 73 ] 2. The upwellings within the upper and
lower mantle are probably laterally shifted, resulting in a pronounced horizontal flow component
near 660 km depth.
[74] 3. At least two distinct upwelling flow regions
are identified by the regional flow model, one
situated beneath Iceland extending to NW of Iceland and one beneath the Kolbeinsey ridge.
[75] 4. Modeled seismic anisotropy based on flow
induced LPO is sensitive to dynamic flow characteristics such as kinematic versus free slip boundary conditions: While the free slip flow model is
dominated by seismically fast directions diverging
from Iceland and the Kolbeinsey Ridge, the kinematic boundary condition model shows predominantly ridge-perpendicular fast directions for most
of the area.
[76] 5. A satisfying fit to observed seismic anisotropy directions is achieved in several but not all
regions. Robust features include the following:
(1) a good fit in a broad region north, east, and
SE of Iceland indicates that the NW-SE anisotropy
observed in this area is induced by the Iceland plume
(i.e., the large-scale North Atlantic upwelling flow),
(2) a good fit beneath Iceland indicates the possible
importance of a SW oriented flow beneath western
Iceland, and (3) a robust misfit between the regional
models and observations in the Reykjanes ridge area
leads to the speculation that in this region B-type
anisotropy with the fast axis perpendicular to
the flow direction may be dominating, possibly
indicating the presence of water.
Acknowledgments
[77] Thanks are extended to Wim Spakman for giving us
access to his tomography model data and to Edouard Kaminski
10.1029/2006GC001359
for making his code DRex available to us. Our colleagues in
the ‘‘Frankfurt-Mainz Iceland Plume Working Group,’’especially Wolfgang Jacoby and Peter Mihalffy, made essential
contributions to this work through their interest and discussions. Thomas Ruedas’s superb Iceland review facilitated our
literature search considerably. Numerous comments of two
anonymous reviewers helped to considerably improve the
manuscript. The original study was supported by a research
grant from the Deutsche Forschungsgemeinschaft; the Space
Research Organization of the Netherlands (SRON) and the
MAGMA grant supported by the European Commission made
it possible to finish this manuscript.
References
Akaogi, M., E. Ito, and A. Navrotsky (1989), Olivine-modified
spinel-spinel transition in the system Mg2SiO4 – Fe2SiO4:
Calorimetric measurements, thermochemical calculations,
and geophysical application, J. Geophys. Res., 94, 15,671 –
15,685.
Albers, M., and U. R. Christensen (1996), The excess temperature of plumes rising from the core-mantle boundary, Geophys. Res. Lett., 23, 3567 – 3570.
Albers, M., and U. R. Christensen (2001), Channeling of
plume flow beneath mid-ocean ridges, Earth Planet. Sci.
Lett., 187, 207 – 220.
Allen, R. M., et al. (1999), The thin hot plume beneath Iceland,
Geophys. J. Int., 137(1), 51 – 63.
Becker, T., and L. Boschi (2002), A comparison of tomographic and geodynamic mantle models, Geochem. Geophys.
Geosyst., 3(1), 1003, doi:10.1029/2001GC000168.
Becker, T. W., J. B. Kellogg, G. Ekström, and R. J. O’Connell
(2003), Comparison of azimuthal seismic anisotropy from
surface waves and finite strain from global mantle-circulation models, Geophys. J. Int., 155, 696 – 714.
Bijwaard, H., and W. Spakman (1999), Tomographic evidence
for a narrow mantle plume below Iceland, Earth Planet. Sci.
Lett., 166, 121 – 126.
Bjarnason, I. T., C. J. Wolfe, S. C. Solomon, and
G. surname>Gudmundsson (1996), Initial results from the
ICEMELT experiment: Body wave delay times and shearwave splitting across Iceland, Geophys. Res. Lett., 23(5),
459 – 462.
Bjarnason, I. T., P. G. Silver, G. Rümpker, and S. C. Solomon
(2002), Shear wave splitting across the Iceland hot spot:
Results from the ICEMELT experiment, J. Geophys. Res.,
107(B12), 2382, doi:10.1029/2001JB000916.
Blackman, D. K., and J.-M. Kendall (2002), Seismic anisotropy in the upper mantle: 2. Predictions for current plate
boundary flow models, Geochem. Geophys. Geosyst., 3(9),
8602, doi:10.1029/2001GC000247.
Botz, R., et al. (1999), Origin of trace gases in submarine
hydrothermal vents of the Kolbeinsey Ridge, north Iceland,
Earth Planet. Sci. Lett., 171, 83 – 93.
Breddam, K., M. D. Kurz, and M. Storey (2000), Mapping out
the conduit of the Iceland mantle plume with helium isotopes, Earth Planet. Sci. Lett., 176, 45 – 55.
Čadek, O., and L. Fleitout (1999), A global geoid model with
imposed plate velocity and partial layering, J. Geophys. Res.,
104(B12), 29,055 – 29,075.
Čadek, O., and L. Fleitout (2003), Effect of lateral viscosity
variations in the top 300 km on the geoid and dynamic
topography, Geophys. J. Int., 152, 566 – 580.
Darbyshire, F. A., R. S. White, and K. F. Priestley (2000),
23 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
Structure of the crust and uppermost mantle of Iceland from
a combined seismic and gravity study, Earth Planet. Sci.
Lett., 181, 409 – 428.
Debayle, E., B. Kennett, and K. Priestley (2005), Global
azimuthal seismic anisotropy and the unique plate-motion
deformation of Australia, Nature, 433, 509 – 512,
doi:10.1038/nature03247.
Detrick, R. S., H. D. Needham, and V. Renard (1995), Gravity
anomalies and crustal thickness variations along the MidAtlantic Ridge between 33°N and 40°N, J. Geophys. Res.,
100(B3), 3767 – 3787.
Devey, C. W., C.-D. Garbe-Schoenberg, P. Stoffers, C. Chauvel,
and D. F. Mertz (1994), Geochemical effects of dynamic
melting beneath ridges: Reconciling major and trace element
variations in Kolbeinsey (and global) mid-oceanic ridge
basalt, J. Geophys. Res., 99, 9077 – 9095.
Dixon, E. T., M. Honda, I. McDougall, I. H. Campbell,
and I. Sigurdsson (2000), Preservation of near-solar neon
isotopic ratios in Icelandic basalts, Earth Planet. Sci.
Lett., 180, 309 – 324.
Ekström, G. (2001), Mapping azimuthal anisotropy of intermediate-period surface waves (abstract), Eos Trans. AGU,
82(47), Abstract S51E-06.
Feighner, M. A., L. H. Kellogg, and B. J. Travis (1995),
Numerical modeling of chemically buoyant mantle plumes
at spreading ridges, Geophys. Res. Lett., 22, 715 – 718.
Fitton, J. G., A. D. Saunders, M. J. Norry, B. S. Hardarson, and
R. N. Taylor (1997), Thermal and chemical structure of the
Iceland plume, Earth Planet. Sci. Lett., 153, 197 – 208.
Fitton, J. G., A. D. Saunders, P. D. Kempton, and B. S. Hardarson
(2003), Does depleted mantle form an intrinsic part of the
Iceland plume?, Geochem. Geophys. Geosyst., 4(3), 1032,
doi:10.1029/2002GC000424.
Foulger, G. R., et al. (2001), Seismic tomography shows that
upwelling beneath Iceland is confined to the upper mantle,
Geophys. J. Int., 146(2), 504 – 530.
Grand, S. P. (2002), Mantle shear-wave tomography and the
fate of the subducted slabs, Philos. Trans. R. Soc. London,
Ser. A, 360, 2475 – 2491.
Grand, S. P., R. D. van der Hilst, and S. Widiyantoro (1997),
Global seismic tomography: A snapshot of convection in the
Earth, GSA Today, 7, 1 – 7.
Gripp, A. E., and R. G. Gordon (2002), Young tracks of hotspots and current plate velocities, Geophys. J. Int., 150,
321 – 361.
Haase, K., and C. W. Devey (1994), The petrology and geochemistry of Vesteris Seamount, Norwegian Basin—An intraplate
volcano of non-plume origin, J. Petrol., 35, 295 – 328.
Hanan, B. B., and J.-G. Schilling (1997), The dynamic evolution of the Iceland mantle plume: The lead isotope perspective, Earth Planet. Sci. Lett., 151, 43 – 60.
Hanan, B. B., J. Blichert-Toft, R. Kingsley, and J. Schilling
(2000), Depleted Iceland mantle plume geochemical signature: Artifact of multicomponent mixing?, Geochem. Geophys. Geosyst., 1(4), doi:10.1029/1999GC000009.
Hart, S. R., J.-G. Schilling, and J. L. Powell (1973), Basalts
from Iceland and along Reykjanes Ridge: Sr isotope geochemistry, Nature, 256, 104 – 107.
Humler, E., and J. Besse (2002), A correlation between midoceanic-ridge basalt chemistry and distances to continents,
Nature, 419, 607 – 609.
Ito, G. (2001), Reykjanes ‘‘V’’-shaped ridges originating from
a pulsing and dehydrating mantle plume, Nature, 411, 681 –
684.
Ito, G., Y. Shen, G. Hirth, and C. J. Wolfe (1999), Mantle flow,
melting and dehydration of the Iceland mantle plume, Earth
10.1029/2006GC001359
Planet. Sci. Lett., 165, 81 – 96.
Kaminski, E. (2002), The influence of water on the development
of lattice preferred orientation in olivine aggregates, Geophys.
Res. Lett., 29(12), 1576, doi:10.1029/2002GL014710.
Kaminski, E., and N. M. Ribe (2001), A kinematic model for
recrystallization and texture development in olivine polycrystals, Earth Planet. Sci. Lett., 189, 253 – 267.
Kaminski, E., and N. M. Ribe (2002), Timescales for the evolution of seismic anisotropy in mantle flow, Geochem. Geophys. Geosyst., 3(8), 1051, doi:10.1029/2001GC000222.
Karato, S. (1993), Importance of anelasticity in the interpretation of seismic tomography, Geophys. Res. Lett., 20, 1623 –
1626.
Keen, C. E., and R. R. Boutilier (2000), Interaction of rifting
and horizontal plume sheets at volcanic margins, J. Geophys.
Res., 105, 13,375 – 13,387.
Kempton, P. D., J. G. Fitton, A. D. Saunders, G. M. Nowell,
R. N. Taylor, B. S. Hardarson, and G. Pearson (2000), The
Iceland plume in space and time: A Sr-Nd-Pb-Hf study of the
North Atlantic rifted margin, Earth Planet. Sci. Lett., 177,
255 – 271.
Klein, E. M., and C. H. Langmuir (1987), Global correlations
of ocean ridge basalt chemistry with axial depth and crustal
thickness, J. Geophys. Res., 92, 8089 – 8115.
Kneller, E. A., P. E. van Keken, S. Karato, and J. Park (2005),
B-type olivine fabric in the mantle wedge: Insights from
high-resolution non-Newtonian subduction zone models,
Earth Planet. Sci. Lett., 237(3 – 4), 781 – 797.
Kodaira, S., R. Mjelde, K. Gunnarsson, H. Shiobara, and
H. Shimamura (1997), Crustal structure of the Kolbeinsey
Ridge, North Atlantic, obtained by use of ocean bottom
seismographs, J. Geophys. Res., 102, 3131 – 3151.
Kreutzmann, A., H. Schmeling, A. Junge, T. Ruedas,
G. Marquart, and I. T. Bjarnason (2004), Temperature and
melting of a ridge-centered plume with application to Iceland,
part II: Predictions for electromagnetic and seismic observables, Geophys. J. Int., 159, 1097 – 1111.
Lemoine, F. G., et al. (1998), The development of the joint
NASA GSFC and NIMA geopotential model EGM96,
NASA/TP-1998-206861, NASA Goddard Space Flight Cent.,
Greenbelt, Md.
Levshin, A. L., M. H. Ritzwoller, M. P. Barmin, A. Villasenor,
and C. A. Padgett (2001), New constraints on the arctic crust
and uppermost mantle: Surface wave group velocities, Pn
and Sn, Phys. Earth Planet. Inter., 123, 185 – 204.
Li, A., and R. S. Detrick (2003), Azimuthal anisotropy and
phase velocity beneath Iceland: Implications for plume-ridge
interaction, Earth Planet. Sci. Lett., 214, 153 – 165.
Macpherson, C. G., D. R. Hilton, D. F. Mertz, and T. Dunai
(1997), Tracing plume, MORB and atmosphere contribution
to Kolbeinsey Ridge Basalts: He, Ne, and Ar isotope evidence (abstract), paper presented at Seventh Goldschmidt
Conference, Lunar and Planet. Inst., Tucson, Ariz.
Mainprice, D., G. Barruol, and W. Ben Ismail (2000), The
seismic anisotropy of the Earth’s mantle: From single crystal
to polycrystal, in Earth’s Deep Interior: Mineral Physics and
Tomography From the Atomic to the Global Scale, Geophys.
Monogr. Ser., vol. 117, edited by S.-I. Karato et al., pp. 237 –
264, AGU, Washington, D. C.
Marquart, G. (2001), On the geometry of mantle flow beneath
drifting lithospheric plates, Geophys. J. Int., 144, 356 – 372.
Marquart, G., and H. Schmeling (2004), A dynamic model for
the Iceland Plume and the North Atlantic based on tomography and gravity data, Geophys. J. Int., 159, 40 – 52.
Marquart, G., H. Schmeling, G. Ito, and B. Schott (2000),
Conditions for plumes to penetrate the mantle phase bound24 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
aries, J. Geophys. Res., 105, 5679 – 5694.
Masters, G., G. Laske, H. Bolton, and A. Dziewonski (2000),
The relative behavior of shear velocity, bulk sound speed,
and compressional velocity in the mantle: Implications for
chemical and thermal structure, in Earth’s Deep Interior:
Mineral Physics and Tomography From the Atomic to
the Global Scale, Geophys. Monogr. Ser., vol. 117, edited
by S.-I. Karato et al., pp. 63 – 87, AGU, Washington, D. C.
McKenzie, D. (1979), Finite deformation during fluid flow,
Geophys. J. R. Astron. Soc., 58, 689 – 715.
Mègnin, C., and B. Romanowicz (2000), The shear velocity
structure of the mantle from the inversion of body, surface,
and higher modes waveforms, Geophys. J. Int., 143, 709 –
728.
Menke, W., B. Brandsdottir, S. Jakobsdottir, and R. Stefansson
(1994), Seismic anisotropy in the crust at the mid-Atlantic
plate boundary in south-west Iceland, Geophys. J. Int., 119,
783 – 790.
Mertz, D. F., C. W. Devey, W. Todt, P. Stoffers, and A. W.
Hofmann (1991), Sr-Nd-Pb isotope evidence against plumeasthenosphere mixing north of Iceland, Earth Planet. Sci.
Lett., 107, 243 – 255.
Mihalffy, P., B. Steinberger, and H. Schmeling (2006), The
effect of the large-scale mantle flow field on the Iceland
hotspot track, Tectonophysics, in press.
Mizukami, T., S. R. Wallis, and J. Yamamoto (2004), Natural
examples of olivine lattice preferred orientation patterns with
a flow-normal a-axis maximum, Nature, 427, 432 – 436.
Moeller, H. (2002), Magma genisis and mantle sources at the
Mid-Atlantic Ridge east of Ascension Island, dissertation,
Math. Naturwiss. Fak., Christian-Albrechts-Univ., Kiel,
Germany.
Montagner, J.-P., and T. Tanimoto (1991), Global upper mantle
tomography of seismic velocities and anisotropy, J. Geophys.
Res., 96, 20,337 – 20,351.
Montelli, R., G. Nolet, F. A. Dahlen, G. Masters, E. R.
Engdahl, and S.-H. Hung (2004), Finite-frequency tomography reveals a variety of plumes in the mantle, Science,
303, 338 – 343.
Murton, B. J., R. N. Taylor, and M. F. Thirlwall (2002), Plumeridge interaction: A geochemical perspective from the Reykjanes Ridge, J. Petrol., 43, 1987 – 2012.
National Oceanic and Atmospheric Administration (1988),
Data Announcement 88-MGG-02: Digital relief of the surface of the Earth, technical report, Natl. Geophys. Data
Cent., Boulder, Colo.
Nicolas, A., and N. L. Christensen (1987), Formation of anisotropy in upper mantle peridotites: A review, in Composition, Structure and Dynamics of the LithosphereAsthenosphere System, Geodyn. Ser., vol. 16, pp. 111 – 123,
edited by K. Fuchs and C. Froidevaux, AGU, Washington,
D. C.
Nicholson, H., and D. Latin (1992), Olivine tholeiites from
Krafla, Iceland: Evidence for variations in melt fractions
within a plume, J. Petrol., 33, 1105 – 1124.
Pilidou, S., K. Priestley, E. Debayle, and O. Gudmundson
(2005), Rayleigh wave tomography in the North Atlantic:
High resolution images of the Iceland, Azores and Eifel
mantle plume, Lithos, 79, 453 – 474.
Ribe, N. M. (1989), Seismic anisotropy and mantle flow,
J. Geophys. Res., 94, 4213 – 4223.
Ribe, N. M., and Y. Yu (1991), A theory for plastic deformation and textural evolution of olivine polycrystals, J. Geophys. Res., 96, 8325 – 8335.
Ribe, N. M., U. R. Christensen, and J. Theissing (1995), The
dynamics of plume-ridge interaction, 1. Ridge-centered
10.1029/2006GC001359
plumes, Earth Planet. Sci. Lett., 134, 155 – 168.
Richards, M. A., and B. H. Hager (1984), Geoid anomalies in a
dynamic Earth, J. Geophys. Res., 89, 5987 – 6002.
Ritsema, J., H. J. van Heijst, and J. H. Woodhouse (1999),
Complex shear wave velocity structure imaged beneath
Africa and Iceland, Science, 286, 1925 – 1928.
Ruedas, T., and H. Schmeling (2006), The orientation of
tensile dikes in a ridge-centered plume, Tectonophysics, in
press.
Ruedas, T., H. Schmeling, G. Marquart, A. Kreutzmann, and
A. Junge (2004), Dynamics and melting of a ridge-centered
plume with application to Iceland, part I: Evolution and crust
production, Geophys. J. Int., 158, 729 – 743.
Ruedas, T., G. Marquart, and H. Schmeling (2006), Iceland:
The current picture of a ridge-centred mantle plume, in
Mantle Plumes: A Multidisciplinary Approach, edited by
J. Ritter and U. R. Christensen, in press.
Russo, R. M., and P. G. Silver (1994), Trench-parallel flow
beneath the Nazca plate from seismic anisotropy, Science,
263, 1105 – 1111.
Salters, V. J. M. (1996), The generation of mid-ocean ridge
basalts from the Hf and Nd isotope perspective, Earth
Planet. Sci. Lett., 14, 109 – 123.
Scarrow, J. H., J. M. Curran, and A. C. Kerr (2000), Major
element records of variable plume involvement in the north
Atlantic province tertiary flood basalts, J. Petrol., 41, 1155 –
1176.
Schilling, J.-G. (1973), Iceland mantle plume: Geochemical
evidence along the Reykjanes Ridge, Nature, 242, 565 –
571.
Schilling, J.-G. (1975), Rare-earth variations across normal
segments of the Reykjanes Ridge 60° – 53°N, Mid-Atlantic
Ridge 29°S, and East Pacific Rise 2°19°S, and evidence on
the composition of the underlying low velocity layer,
J. Geophys. Res., 87, 5593 – 5610.
Schilling, J.-G. (1991), Fluxes and excess temperature of mantle plumes inferred from their interaction with migrating
mid-ocean ridges, Nature, 313, 187 – 191.
Schilling, J.-G., M. Zajac, R. Evans, T. Johnston, W. White,
J. D. Devine, and R. Kingsley (1983), Petrologic and geochemical variations along the Mid-Atlantic Ridge from
29°N to 73°N, Am. J. Sci., 283, 510 – 586.
Schilling, J.-G., R. Kingsley, D. Fontignie, R. Poreda, and
S. Xue (1999), Dispersion of the Jan Mayen and Iceland
mantle plume in the Arctic: A He-Pb-Nd-Sr isotope tracer
study of basalts from the Kolbeinsey, Mohns, and Knipovich Ridges, J. Geophys. Res., 104, 10,543 – 10,569.
Schmeling, H., G. Marquart, and T. Ruedas (2003), Pressureand temperature-dependent thermal expansivity and the
effect on mantle convection and surface observables,
Geophys. J. Int., 154, 224 – 229.
Searle, R. C., J. A. Keeton, R. B. Owens, R. S. White,
R. Mecklenburgh, B. Parsons, and S. M. Lee (1998), The
Reykjanes Ridge, from multibeam bathymetry, gravity, and
magnetic investigations, Earth Planet. Sci. Lett., 160, 463 –
478.
Shen, Y., S. C. Solomon, I. T. Bjarnason, and C. J. Wolfe
(1998), Seismic evidence for a lower mantle origin of the
Iceland plume, Nature, 395, 62 – 65.
Shen, Y., et al. (2002), Seismic evidence for a tilted mantle
plume and north-south flow beneath Iceland, Earth Planet.
Sci. Lett., 197, 261 – 272.
Shen, Y., C. J. Wolfe, and S. C. Solomon (2003), Seismological evidence for a mid-mantle discontinuity beneath Hawaii
and Iceland, Earth Planet. Sci. Lett, 214, 143 – 151.
Sinha, M. C., S. C. Constable, C. Peirce, A. White,
25 of 26
Geochemistry
Geophysics
Geosystems
3
G
marquart et al.: flow and anisotropy models
G. Heinson, L. M. MacGregor, and D. A. Narvin (1998),
Magmatic processes at slow spreading ridges: Implications
of the RAMESSES experiment at 57°4500N on the MidAtlantic Ridge, Geophys. J. Int., 135, 731 – 745.
Skogseid, J., T. Pedersen, O. Eldholm, and B. T. Larsen
(1992), Tectonism and magmatism during NE Atlantic continental break up: The Voring Margin, in Magmatism and
the Causes of Continental Break-Up, edited by B. C. Storey,
T. Alabaster, and R. J. Pankhurst, Geol. Soc. Spec. Publ., 68,
305 – 320.
Stacey, F. D. (1992), Physics of the Earth, 3rd ed., Brookfield,
Brisbane, Queensland, Australia.
Steinberger, B., and R. J. O’Connell (1998), Advection of
plumes in mantle flow: Implications on hotspot motion,
mantle viscosity and plume distribution, Geophys. J. Int.,
132, 412 – 434, doi:10.1046/j.1365-246x.1998.00447.x.
Stracke, A., A. Zindler, V. J. M. Salters, D. McKenzie,
J. Blichert-Toft, F. Albarède, and K. Grönvold (2003),
Theistareykir revisited, Geochem. Geophys. Geosyst., 4(2),
8507, doi:10.1029/2001GC000201.
Tanimoto, T., and D. L. Anderson (1984), Mapping convection
in the mantle, Geophys. Res. Lett., 11, 287 – 290.
Thirlwall, M. F. (1995), Generation of the Pb isotopic characteristics of the Iceland plume, J. Geol. Soc. London, 152,
991 – 996.
Tommasi, A. (1998), Forward modeling of the development of
seismic anisotropy of the upper mantle, Earth Planet. Sci.
Lett., 160, 1 – 13.
Torsvik, T. H., J. Mosar, and E. A. Eide (2001), CretaceousTertiary geodynamics: A North Atlantic exercise, Geophys.
J. Int., 146, 850 – 866.
10.1029/2006GC001359
Trampert, J., and J. H. Woodhouse (2003), Global anisotropic
phase velocity maps for fundamental mode surface waves
between 40 and 150s, Geophys. J. Int., 154, 154 – 165.
Vogt, P. R. (1971), Asthenosphere motion recorded by the
ocean floor south of Iceland, Earth Planet. Sci. Lett., 13,
153 – 160.
White, R. S. (1997), Rift-plume interactions in the North
Atlantic, Philos. Trans. R. Soc. London, Ser. A, 355(1723),
319 – 339.
Wolfe, C. J., I. T. Bjarnason, J. VanDecar, and S. Solomon
(1997), Seismic structure of the Iceland mantle plume, Nature, 385(5727), 245 – 247.
Woodward, R. L., and G. Masters (1991), Global upper mantle
structure from long-period differential travel-times, J. Geophys. Res., 96, 6351 – 6377.
Xue, M., and R. M. Allen (2005), Asthenospheric channeling
of the Icelandic upwelling: Evidence from seismic anisotropy, Earth Planet. Sci. Lett., 235, 167 – 182.
Yang, X., and K. M. Fischer (1994), Constraints on North
Atlantic upper mantle anisotropy from S and SS phases,
Geophys. Res. Lett., 21, 309 – 312.
Zhang, Y.-S., and T. Tanimoto (1993), High-resolution global
upper mantle structure and plate tectonics, J. Geophys. Res.,
98, 9793 – 9823.
Zhao, D. (2001), Seismic structure and origin of hotspots and
mantle plumes, Earth Planet. Sci. Lett., 192(3), 251 – 265.
Zhong, S. (2006), Constraints on thermochemical convection
of the mantle from plume heat flux, plume excess temperature, and upper mantle temperature, J. Geophys. Res., 111,
B04409, doi:10.1029/2005JB003972.
26 of 26