Geophys. J. Int. (2008)
doi: 10.1111/j.1365-246X.2008.03945.x
Constraints on crustal structure and composition within a
continental suture zone in the Irish Caledonides from shear wave
wide-angle reflection data and lower crustal xenoliths
F. Hauser,1,2 B. M. O’Reilly,1 P. W. Readman,1 J. S. Daly2 and R. Van den Berg3
1 Geophysics
Section, Dublin Institute for Advanced Studies, 5 Merrion Square, Dublin 2, Ireland. E-mail: [email protected]
School of Geological Sciences, University College Dublin, Belfield, Dublin 4, Ireland
3 Department of Geology, Stellenbosch University, Matieland 7602, South Africa
2 UCD
SUMMARY
Shear wave seismic velocities (V s ) when used together with compressional wave velocities
(V p ) are a powerful diagnostic of the chemical composition and mineralogy of the continental
lithosphere. In this paper, we present whole crustal models of V s and V p velocities from two
wide-angle seismic profiles from southwest Ireland, which straddle the Eastern Avalonian
Terrane within the Iapetus Suture Zone. These models are based on traveltime interpretations
of primary and reflected P- and S-wave seismic phases generated by explosive underwater
sources. The quality of the shear wave data is exceptional when compared to similar data
sets recorded elsewhere and allows the distribution of Poisson’s ratio (σ ) within the entire
crust to be accurately determined. Models of Poisson’s ratio, V p and V s are then used to
constrain crustal composition at various depths down to the Moho (ca. 32 km depth) using
published laboratory velocity measurements from rocks of different metamorphic grades and
chemistry. The results indicate that the Irish crust in this region of the Caledonian orogenic
belt is unusually felsic in bulk composition with a σ value of 0.247, which is below the
global average value of 0.265. Variations in σ within the upper ca. 5 km of crust are greatest
(0.20–0.28) and can be correlated with siliciclastic and carbonate sediments in Devonian
and Carboniferous extensional sedimentary basins that were formed during the early part of
the Variscan orogenic cycle. The typically more uniform variation in the mid crust down
to 15 km depth is consistent with a granite/granodiorite or metagreywacke composition of
greenschist to amphibolite facies mineralogy. Within the lower crust the value of σ (0.257–
0.265) is significantly less than the global mean value of 0.281 predicted from an average
crustal petrology model. This suggests a bulk silica content (%SiO 2 ) of ∼64 per cent for
the lower crust, based on an empirical relationship between σ and %SiO 2 , determined from
compilations of laboratory data. These results are consistent with petrophysical, geochemical
and petrological data from an unusually well-preserved lower crustal xenolith suite from the
Irish Midlands, hosted in early Carboniferous volcanic rocks. The xenoliths (mainly partially
melted granulite facies metapelites) evidently represent the bulk composition of the lower
crust, which was largely derived by accretion of sedimentary material, derived from oceanic,
island arc and continental margin sources, during oblique Caledonian collision.
Key words: Composition of the continental crust; Controlled source seismology; Crustal
structure; Europe.
1 I N T RO D U C T I O N
With a few notable exceptions controlled source seismic experiments have rarely reported shear wave velocities as these seismic
phases are usually difficult to recognize with a lower signal-tonoise ratio than P-wave arrivals (Holbrook et al. 1988; Hawman
et al. 1990). Globally, most studies of the shear wave properties
C
2008 The Authors
C 2008 RAS
Journal compilation of the crust have used receiver functions, which rely on various
P to S conversions (and multiples thereof) of teleseismic energy
at the Moho or within the crust to provide information on bulk
(whole crustal) V p /V s variations and hence Poisson’s ratio (Chevrot
& van der Hilst 2000; Landes et al. 2006; Nair et al. 2006). Good
quality controlled source S-wave data, which potentially allows the
determination of variations in Poisson’s ratio within the crust, have
1
GJI Tectonics and geodynamics
Accepted 2008 August 15. Received 2008 June 13; in original form 2008 January 21
2
F. Hauser et al.
rarely been gathered (Holbrook et al. 1988; Morozov et al. 2001;
Mechie et al. 2005). This paper presents results from an analysis of
S waves, converted from P waves at the near surface, and recorded
during the Variscan Network (VARNET) controlled source seismic
experiment.
The VARNET project was an international, multidisciplinary
project, designed to examine the role of Caledonian and later
Variscan tectonics in shaping the crustal structure of southwest Ireland (Landes et al. 2003). The S-wave records are unusually good
and allow a well-constrained S-wave velocity structure of the crust
to be determined. In this sense the VARNET S-wave data set is an
important and relatively unique one. Variations in Poisson’s ratio
are derived from the data using the published P-wave models for
crustal velocity structure (Masson et al. 1998; Landes et al. 2000)
as preliminary starting models.
The resulting models for the P- and S-wave structure are used
to derive the distribution of Poisson’s ratio within the crust and to
infer its bulk chemistry. A petrological model for the crust is de-
veloped, using published experimental data (Christensen 1996) on
the elastic properties of rocks and minerals. The petrological model
is then compared and integrated with petrophysical, petrological
and geochemical data from a suite of exceptionally well-preserved
mid-to-lower crustal xenoliths, hosted in lower Carboniferous volcanics from central Ireland (Van den Berg et al. 2005). Finally, the
implications of the results for processes of crustal formation and
tectonics are discussed and set within a larger-scale context.
2 GEOPHYSICAL AND GEOLOGICAL
B A C KG R O U N D
The geological framework of SW-Ireland is dominated by upper
Palaeozoic strata of the tectonically inverted Munster and South
Munster Basin (Fig. 1). In a general plate tectonic context the Munster Basin is part of the early Devonian back-arc rift system that
developed in western and central Europe in response to the oblique
Figure 1. Simplified geological map of southwest Ireland showing the location of the VARNET onshore wide-angle seismic reflection lines A and B. Red
stars are shot point locations. Xenolith locations are within red elliptical area shown in the inset map. Solid black lines depict main faults. KMFZ is the
Killarney-Mallow Fault Zone. Dotted line is the offshore extension of the Clare Basin.
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
convergence between Gondwana and Laurasia (Ziegler 1986). The
basement rocks beneath this upper Palaeozoic succession are part of
the Avalonian palaeo-continent, which was tectonically sutured with
the Laurentian Plate as the Iapetus Ocean closed, during the Ordovician and Silurian (Soper et al. 1992). This process of oceanic plate
subduction and ocean closure was complete by late Silurian/early
Devonian time and involved large amounts of transpressional (sinistral) deformation, producing only small amounts of crustal shortening within the Irish sector of the Caledonian Orogen. The later
phases of Caledonian deformation led to melting of the accreted
crust and the genesis and intrusion of granites into the upper to mid
crust.
The onset of crustal extension and syn-rift subsidence in the
Munster Basin occurred in the Middle Devonian and continued
throughout the Upper Devonian and Lower Carboniferous, with the
deposition of terrestrial sediments (e.g. Higgs & Russell 1981).
Superimposed upon the southern part of the Munster Basin is the
late Devonian to Namurian South Munster Basin (George et al.
1976). The initiation of this smaller marine subbasin approximately
coincides with the widespread Early Carboniferous transgression
of northern Europe (e.g. Williams et al. 1989). Within the South
Munster Basin the Old Red Sandstone fluvial facies is overlain by a
thick sequence of rocks of shallow marine origin, which in turn give
way to thin, deeper water, mudrocks (Naylor et al. 1989). Subsequently, the Munster Basin underwent tectonic inversion, during the
late Carboniferous (end-Variscan) deformation of southern Ireland,
with the intensity of deformation decreasing northwards towards
the Irish Midlands (Price & Todd 1988; Readman et al. 1997).
Analysis of the P-wave energy recorded along the wide-angle
seismic lines of the VARNET project resolved the deeper structure
of the crust and mantle in this area, close to the surface trace of the
Iapetus Suture Zone (Masson et al. 1998; Landes et al. 2000). The
VARNET results are in broad agreement with the older COOLE I
onshore profile of Lowe & Jacob (1989) that crosses the eastern end
of the Munster Basin, however, extends further towards the north
coast of Ireland. Throughout Ireland the crust is three layered (each
layer approximately 10 km thick) and the overall crustal thickness
is about 30 km with a 2–3 km thick Moho transition zone, observed
on the longer-range (>200 km) seismic refraction profiles (Jacob
et al. 1985; Lowe & Jacob 1989).
3
for S-wave and P-wave seismograms, respectively. The horizontal
components of the geophones were oriented north south and east
west during the field experiment and since the profiles run almost
north to south (Fig. 1) it was not necessary to rotate them into radial
and transverse components. In the following descriptions the term
‘traveltime’ refers to the reduced traveltime of the seismic sections,
‘offset’ is the distance from the source to the receivers on a seismogram section, and ‘distance’ is measured from the southernmost
shot points on both lines A and B.
Two major shear-wave phases can be correlated on all record
sections. The first is S g (a refracted arrival through the upper crust)
and is strong on the outermost shot points along both profiles
(Figs 2b and 5b), but weaker and sometimes within the signalto-noise level for central shot points. It has an apparent velocity
of about 3.5 km s−1 and can, on the better sections, be correlated
to over 100 km offset. Arrivals with lower apparent velocities than
3.46 km s−1 are also observed and are assigned to the shallower
sedimentary layers.
The second major phase, which can be observed on all record
sections, is S m S (an S-wave reflection from the crust/mantle boundary. This phase is observable from offsets greater than 70 km to
the extremities of the profiles (Fig. 2b). The outermost shot points
show a strong and clear S m S phase, while on some of the innermost
shot points it is only recognizable by an increase in energy that is,
however, coherent over several traces (Fig. 3b).
In addition, several intracrustal reflections can be correlated;
however, because of the lower signal-to-noise ratio they are not
observed on all record sections. These reflections from the middle
and the lower crust are easier to correlate in the wide-angle offset range (i.e. 150–200 km) along line A (Figs 2b and 4b). In the
subcritical distance range amplitudes are usually small, indicating a
small velocity contrast across these interfaces. This is in contrast to
the P-wave record sections, where Landes et al. (2000) and Masson
et al. (1998) correlated up to five intracrustal reflectors along lines
A and B, respectively.
Although an S-wave refraction through the upper mantle (S n )
is likely, the amplitude of this phase rarely exceeds the noise level,
making it difficult to recognize (compare Figs 2b and 4b). Moreover,
no reflected S-wave from within the upper mantle, corresponding
to the P-wave reflection observed by Landes et al. (2000) could be
correlated.
3 S E I S M I C D ATA A N D P R O C E S S I N G
The controlled source experiment recorded strong P-wave seismic
energy along two high-resolution wide-angle seismic profiles (lines
A and B in Fig. 1) in two separate deployments. Along line A, nine
in-line shots were recorded over a length of ca. 200 km, while along
line B, which is ca. 140 km long, 13 in-line shots were recorded.
Where the water depths were great enough, a 25–50 kg explosive
source was fired at an optimum water depth of 65 m (Jacob 1975), in
the seas offshore Ireland, or in water-filled disused quarries on land,
and generated clear converted S-wave energy at the sediment/water
interface. These converted S-wave phases were observed on all
three geophone components from instruments along the two VARNET lines. More details of the controlled source experiment can
be found elsewhere (e.g. Masson et al. 1998; Landes et al. 2003).
Some examples of the P-wave and excellently converted S-wave
seismograms from lines A and B are shown in Figs 2–7.
All seismograms were processed with the SeismicHandler program package of Stammler (1994). After several trials to improve
the data quality, a band-pass filter of 2–9 Hz and 4–12 Hz was used
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation 4 M O D E L L I N G P RO C E D U R E S
The procedure for the 2-D traveltime modelling of the S-wave data,
placed more emphasis on forward modelling, and was similar to
that used by Masson et al. (1998) and Landes et al. (2000) for the
interpretation of the P-wave data along lines A and B, respectively.
The interpretation is based on forward and inverse ray tracing techniques and uses the top-to-bottom approach as described in Zelt
(1999). This is because the very wide shot spacing (20–80 km) is
inappropriate for a tomographic approach, using first arrivals, since
the dense ray coverage required is not achieved.
Initially, the published P-wave models of Landes et al. (2000)
for line A and Masson et al. (1998) for line B were used as starting models to derive the S-wave velocity structure. This involved
assigning V p /V s ratios to the published P-wave models in order to
predict the arrival times of the converted S-wave energy. The range
in assigned V p /V s ratios encompassed the range of experimentally
determined values for all major rock types within the crystalline
4
F. Hauser et al.
(a)
PmP
P 6P
P5P
P4P
P3P
Pg
Pn
(b)
S 6S
S5S
SmS
S4S
S3S
Sg
Sn
Figure 2. (a) P-wave record section from shot point 4 along line A (for location, see Fig. 1). The phase labelled P g is a diving wave penetrating to ca. 10
km depth and sampling the first three layers in the model with velocities ranging from 5.3 to 6.0 km s−1 . Phases labelled P 3 P to P 6 P are reflections from
intracrustal interfaces (shown in Fig. 8a). P m P is the reflection from the Moho and P n the diving wave in the upper mantle. The superimposed curves are the
calculated traveltimes from the final velocity model. Traveltime picks are indicated by the small horizontal bars. (b) S-wave record section (EW-component)
from shot point 4 along line A. The phase labelled S g is a diving wave penetrating the first three layers in the model with velocities ranging from 3.0 to 3.55
km s−1 . Phases labelled S 3 S to S 6 S are reflections from intracrustal interfaces. S m S is the reflection from the Moho and the most dominant phase in the entire
record section. S n (the diving wave in the upper mantle) is extremely weak but correlateable between 140 and 170 km offset. The superimposed curves are the
calculated traveltimes from the final velocity model. Traveltime picks are indicated by small horizontal bars.
crust at the appropriate temperatures and pressures (Christensen &
Mooney 1995; Christensen 1996).
At first, arriving S-wave energy from the upper 10 km of crust
(i.e. the S g phase) was considered. However, the calculated arrival
times necessary to fit the observed ones required unrealistically
high S-wave velocity gradients and V p /V s ratios that were greatly
inconsistent with the range of values permitted by the laboratory
experiments (Christensen 1996).
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
5
(a)
PmP
P6P
P 5P
P4 P
P 3P
Pg
(b)
SmS S S
S5S
6
S4S
S3S
Sg
Figure 3. (a) P-wave record section from shot point 8 along line A. The phases and traveltime picks are labelled as in Fig. 2(a). The P n phase is not apparent as
a first arrival because the source-receiver offsets do not exceed the crossover distance. The P m P phase becomes strong at offsets exceeding 80 km. (b) S-wave
record section (EW-component) from shot point 8 along line A. The phases and traveltime picks are labelled as in Fig. 2(b). The signal-to-noise ratio is lower
but again the S m S phase dominates the record section from about 70 km onwards.
This required a modified modelling strategy where both P- and
converted S-wave energy were simultaneously modelled for both
the P g and S g phases. All three components were used for correlation; however, only the component with the best signal-to-noise
ratio was used for picking traveltimes. Since the S-wave converted
energy for the first arrival is remarkably good; however, of lower fre
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation quency content than the P-wave energy, the derived S-wave velocity
variation is smoother than the P-wave variation.
Once a satisfactory model for the upper crust had been found the
mid to lower crust was modelled using mainly secondary reflected
phases and occasionally clear head waves. Again, because of the
lower frequency S-waves, and the prolongation and reverberations
6
F. Hauser et al.
(a)
PmP P P
6
P5P
P4P
P3P
Pg
Pn
(b)
SmS S6S S S
5
S4S
S3S
Sg
Sn
Figure 4. (a) P-wave record section from shot point 11 along line A, with phase labelling and traveltime picks as in Fig. 3(a). The P g phase is very weak
and almost horizontal implying a surface velocity of 6.0 km s−1 with a very low vertical gradient. The P m P phase is again very strong at the largest source
to receiver offsets. The diving wave through the upper mantle (P n ) becomes a first arrival at about 125 km offset with an apparent velocity of 8 km s−1 . (b)
S-wave record section (EW-component) from shot point 11 along line A. The S g phase is somewhat stronger than the corresponding P g phase suggesting
higher S-wave velocity gradients. The S m S is the dominant crustal phase between 60 and 180 km. A weak S n phase is present at 130–160 km offset.
of the P-wave coda, the reflected S-phases are harder to recognize
(see Figs 2–7). This means that the final S-wave model for the
mid to lower crust is smoother along both VARNET lines. Finally,
velocities within the subcrust were determined using the P n and
S n phases, if possible. Since line A is longer (200 km) than line B
(140 km), both Moho depth and upper mantle velocities are better
constrained on line A, where more P m P and S m S post-critical energy
was recorded (Figs 2 and 4).
The resulting joint P- and S-wave model for the upper crust differs
in detail from the original P-wave velocity models (Masson et al.
1998; Landes et al. 2000); however, the overall velocity structure for
the middle to lower crust is still very similar. The largest differences
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
7
(a)
PmP
P 6P
P5P
P4P
P 3P
Pg
(b)
SmS
S6S
S5S
S4S
S3S
Sg
Figure 5. (a) P-wave record section from shot point 33 along line B, with labelling as in Fig. 4 (for location, see Fig. 1). The phase labelled P g is a diving
wave sampling the first three layers in the model with velocities ranging from 5.0 to 6.0 km s−1 . Phases labelled P 3 P to P 6 P are reflections from intracrustal
interfaces. Phase P 6 P is the reflection from the top of the interface which subdivides the lower crust into two layers, similar to the lower crustal structure found
on line A. P m P is the reflection from the Moho. No first arrival P n phase is visible as the line is too short. The superimposed curves are the calculated traveltimes
from the final velocity model. (b) S-wave record section (NS-component) from shot point 33 along line B. S g is a diving wave penetrating the first three layers
in the model with velocities ranging from 3.0 to 3.6 km s−1 . Phases labelled S 3 S to S 6 S are reflections from intracrustal interfaces. S 6 S is observable beyond
75 km and S m S is the reflection from the Moho and the most dominant phase. The superimposed curves are the calculated traveltimes from the final velocity
model. Traveltime picks are indicated by small horizontal bars.
along line A occur in the upper crust, where the velocity structure
and model geometry is now (this study) much simpler. Along line
B the basic geometry of the upper 10 km of crust remains the same
as in the model of Masson et al. (1998).
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Within the middle crust, only slight variations to the original published models were required by the additional S-wave constraints.
The largest differences are along the southern part of line A, where
the geometry of the middle crust is simpler than in the published
8
F. Hauser et al.
(a)
P6P
PmP
PmP
P5P
P4P
P3P
Pg
Pg
(b)
SmS
S6S
S5S
S5S
S6S
SmS
S4S
S3S
Sg
Sg
Figure 6. (a) P-wave record section from shot point 19 along line B. The phases and traveltime picks are labelled as in Fig. 5(a). The signal-to-noise ratio
is low and the reflected energy is all pre-critical. (b) S-wave record section (E–W component) from shot point 19 along line B. The phases are labelled as in
Fig 5(b). The signal-to-noise ratio is low for all reflected phases.
model of Landes et al. (2000). In the lower crust along line A the geometry remains the same; however, the additional S-wave data now
require lower vertical and lateral velocity gradients with similar
mean P-wave velocities to the original model.
The major difference in the structure and velocity of the lower
crust is present along line B. Here the original one-layer lower crust
of Masson et al. (1998) is replaced by a two-layer one, similar in
structure to that on line A. This is based on close inspection of Pand S-wave seismograms at larger shot to receiver offsets, which
define bursts of reflected energy (e.g. defined by P 6 P and S 6 S phases
in Figs 5a and b) at an arrival time consistent with the existence of
this additional layer on line B.
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
9
(a)
PmP
P6P
P5P
P4P
P3P
Pg
(b)
SmS
S 6S
S5S
S4S
S3 S
Sg
Figure 7. (a) P-wave record section from shot point 16 along line B, with labelling as in Fig. 6. The P g phase is initially strong but becomes weak at distances
exceeding 65 km offset, indicating a decrease in velocity gradient associated with the low velocity zone in the upper crust at 80 km along the profile. Phase
P 6 P is again correlateable at distances greater than 75 km. (b) S-wave record section (E–W component) from shot point 16 along line B. The S g phase is weak
relative to the ambient noise. The S 6 S phase is visible at 70–110 km offset.
4.1 Model velocity uncertainties
Estimates of the value of Poisson’s ratio within different parts of the
crust are very sensitive to error estimates for both P- and S-wave
velocities (Hawman et al. 1990; Christensen 1996). An evaluation
of model uncertainties is, therefore, critical before inferring vari-
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation ations in Poisson’s ratio and interpreting them in terms of rock
composition.
The strategy used to access model velocity uncertainties is
based on a combination of forward and inverse ray tracing techniques, as described for example in Zelt & Smith (1992), Zelt &
Forsyth (1994) and Zelt (1999). During the inversion procedure, the
10
F. Hauser et al.
South
(a) 5
(b)
North
Line A
South
34
2
8
13
North
14
Line B
Figure 8. (a) S-wave ray coverage for line A. Only rays from the central and end shotpoints (5, 2, 8 and 13 in Fig. 1) are shown for clarity. (b) S-wave ray
coverage for line B, with the end shotpoints (14 and 34 in Fig. 1) indicated.
absolute parameter uncertainties and the resolution can be estimated. The match between computed and observed traveltimes
(T rms ) is measured using the normalized form of the misfit parameter χ 2 as defined by Bevington (1969). The ideal situation is
met when χ 2 ∼ 1. This means that the computed and observed
traveltimes match to within their assigned uncertainties. If χ 2 1
the observed data have been over-fitted and the model will include
some structure, which are not supported by the data, or the assigned
phase picking errors may be too pessimistic. If, however, χ 2 1,
the observed data have sampled small heterogeneities that cannot
be resolved (Zelt 1999). Nevertheless, in many cases inversion results that yield values of χ 2 > 1 can be accepted, if the resolution
parameter is high enough and if rays can be traced to most receiver
stations (Figs 8a and b).
Table 1 summarizes the results for the simultaneous inversion
of all P- and S-wave velocity nodes, for both lines A and B, with
the total number of traveltime picks used for each of the lines as
well as the final T rms and χ 2 values. However, in order to evaluate
the P and S model sensitivities to velocity perturbations within the
final model at different depths, velocity nodes were incremented by
set values and the model resolution parameters (T rms and χ 2 ) were
calculated. The results of this type of analysis are shown in Tables
2, 3, 4 and 5 for lines A and B.
Based on this type of analysis the uncertainties in model velocities
of the upper 10 km of crust (i.e. the P g and S g phase branches)
are estimated to be better than ±0.05 km s−1 . Perturbing the model
velocities within the upper 10 km of crust by ±0.05 km s−1 increases
T rms and χ 2 to values greater than the picking uncertainties for
the final models (see Tables 1–5). The uncertainties for the clear
Table 1. Final shear (S) and compressional wave (P) velocity uncertainties
for VARNET lines A and B.
Profile
Line A (S)
Line B (S)
Total number of picks
Picks used
T rms
χ2
1156
1152
0.089
2.64
1584
1569
0.09
3.21
Profile
Total number of picks
Picks used
T rms
χ2
line A (P)
1230
1226
0.071
1.557
line B (P)
1436
1424
0.068
1.276
The traveltime fit was assessed by monitoring the total number of picks
used, the root-mean-square traveltime residual (T rms ), and the normalized
misfit parameter (χ 2 ). The overall picking uncertainties vary between
±0.05 and 0.15 s, depending on the signal-to-noise ratio.
P m P and S m S phases, based on a similar analysis, are better than
±0.1 km s−1 . Corresponding depth uncertainties for mid to lower
crustal interfaces are typically in the range of 0.5–1 km and are
consistent with the values for line B obtained by Masson et al.
(1998).
The evaluation of uncertainties is more difficult for intracrustal
phases because the signal-to-noise ratio is lower, except where clear
‘bursts’ in energy occur at post-critical offsets, as in Figs 2 and 5.
Where these phases are strongest the uncertainty in intracrustal
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
Table 2. Shear wave model sensitivities for line A.
Layer no.
Phase
V s (km s−1 )
T rms (s)
χ2
NPTS
1
3
3
1–3
1
1–3
1–7
6–7
6–7
1–7
Sg
Sg
Sg
Sg
SmS
SmS
SmS
SmS
SmS
SmS
0.00
0.05
0.10
0.05
0.00
0.05
0.05
0.05
0.10
0.02
0.060
0.210
0.416
0.340
0.060
0.184
0.482
0.182
0.330
0.211
1.446
6.870
27.080
17.760
1.980
5.430
23.560
5.290
17.430
6.990
107
107
107
104
265
258
254
251
251
254
The model was assessed by monitoring the number of picks used (NPTS),
the root-mean-square traveltime residual (T rms ), and the normalized misfit
parameter (χ 2 ). A perturbation of V s is added to ‘Layer no’. and the
effect is measured using ‘Phase’. During the analysis picking errors were
assigned to the various phases based on the final model (see Table 1).
Table 3. Compressional wave model sensitivities for line A. Model perturbation and analysis as in Table 2.
Layer number
Phase
V p (km s−1 )
T rms (s)
χ2
NPTS
1
3
3
1–3
1
1–3
1–7
6–7
6–7
1–7
Pg
Pg
Pg
Pg
PmP
PmP
PmP
PmP
PmP
PmP
0.00
0.05
0.10
0.05
0.00
0.05
0.05
0.05
0.10
0.02
0.043
0.079
0.150
0.136
0.082
0.087
0.163
0.082
0.122
0.093
1.386
4.127
12.005
10.127
1.434
4.842
16.708
4.102
7.859
5.622
156
149
142
143
192
192
192
192
192
192
velocity determinations is of the same order as uncertainties for the
reflected P m P and S m S phases.
5 VELOCITY MODELS
The resulting P- and S-wave velocity structures for both VARNET
lines A and B are shown in Figs 9 and 10. Seven distinct crustal
layers (L1–7) are defined along the derived models for both lines.
When changes in crustal composition, deduced from this velocity
structure and Poisson’s ratio variations, are presented later, it becomes apparent that the layered velocity structure reflects (to some
extent) the model parameterization (Zelt & Smith 1992). The P- and
S-wave velocity structure within these layers is very similar for the
two lines; however, the geometry of the individual layers changes
substantially. The models are first described and compared in terms
of P- and S-wave velocity structure, then variations in V p /V s ratios
are used to define Poisson’s ratio, which is an important diagnostic
for bulk crustal petrology and mineralogy.
11
Table 4. Shear wave model sensitivities for line B. Analysis as in Tables 2
and 3.
Layer number
Phase
V s (km s−1 )
T rms (s)
χ2
NPTS
1
3
3
1–3
1
1–3
1–7
6–7
6–7
1–7
Sg
Sg
Sg
Sg
SmS
SmS
SmS
SmS
SmS
SmS
0.00
0.05
0.10
0.05
0.00
0.05
0.05
0.05
0.10
0.02
0.105
0.106
0.220
0.160
0.075
0.110
0.340
0.100
0.200
0.125
1.112
1.130
4.750
2.690
2.190
4.860
47.250
3.660
15.890
6.310
554
548
550
552
258
258
256
256
254
258
Table 5. Compressional wave model sensitivities for line B. Analysis as in
Tables 2 and 3.
Layer number
Phase
V p (km s−1 )
T rms (s)
χ2
NPTS
1
3
3
1–3
1
1–3
1–7
6–7
6–7
1–7
Pg
Pg
Pg
Pg
PmP
PmP
PmP
PmP
PmP
PmP
0.00
0.05
0.10
0.05
0.00
0.05
0.05
0.05
0.10
0.02
0.040
0.070
0.125
0.099
0.082
0.085
0.139
0.080
0.105
0.083
1.286
4.022
12.835
8.102
1.159
1.160
2.948
1.159
1.913
1.223
237
235
237
233
248
245
245
247
248
248
the distribution of Old Red Sandstone (ORS) quartz rich, clastic
sediments in the Munster Basin.
Towards the northern end of the profiles (Figs 9a and 10a) higher
velocities (5.9–6.1 km s−1 ), for example between 140 and 200 km
on line A, correlate with Namurian to Westphalian sediments within
the Clare Basin (Fig. 1) and its offshore extension, as defined by a
broad Bouguer anomaly high (Readman et al. 2003). A third underlying layer in the upper crust, extending to ca. 10 km depth on both
profiles has velocities in the range of 5.8–6.2 km s−1 , at the ends
of both profiles, decreasing to lower velocities (5.7–5.8 km s−1 ) towards the centre of both profiles. This region of lower velocities in
the upper crust corresponds to a region of relatively lower Bouguer
gravity that may be related to late Caledonian granitic rocks at depth,
representing the structural continuation of large granite bodies exposed in eastern Ireland (Readman et al. 1997).
The mid crust, between ca. 10 and 23 km depth, consists of two
layers. Velocities in the first layer lie between 6.0 and 6.3 km s−1 and
overlie a layer in which they vary from 6.4 to 6.6 km s−1 . Within the
lower crust, which is also subdivided into two layers, the velocity
structure is remarkably uniform. An upper layer, 3–5 km thick, has
velocities of 6.8–6.9 km s−1 and overlies a layer of similar thickness, with a velocity between 7.0 and 7.2 km s−1 , on both profiles
(Figs 9a and 10a). Upper mantle velocities of 8.0–8.1 km s−1 are
defined along line A by post-critical P n phases, observed at offsets
beyond 130 km.
5.1 P-wave velocity models
The velocity within the shallow parts of the crust (i.e. < 6 km depth)
shows large lateral variations form north to south, which are broadly
similar on both profiles (Figs 9a and 10a). In the south (between 0
and 100 km on line B and 0–60 km on line A) P-wave velocities are
lowest and vary from 5.0 to 5.2 km s−1 near the surface to 5.8 km s−1
at depth. The extent of these lower velocities correlates well with
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation 5.2 S-wave velocity models
Variation in the S-wave velocity structure (Figs 9b and 10b) along
both lines (A and B) is much smoother than the variation in P-wave
velocities. The main features defined in the P-wave model for both
lines are also present in the S-wave model, and the variations in
velocity are not unusual.
12
F. Hauser et al.
(a)
Line A
S
Killarney-Mallow
Fault
Munster Basin
Shannon
Estuary
4
Clare Basin
8
5.6
5.9
5.5
6.05
5.8
6.0 - 6.1
6.3
6.4
6.5
6.5
6.6
6.8
6.9
7.0
7.2
8.0
8.0
(b)
Killarney-Mallow
Fault
Munster Basin
4
Shannon
Estuary
3.3 (1.66)
3.1 (1.74)
L2
3.55 (1.69)
3.2 (1.75)
3.4 (1.70)
3.7 (1.76)
3.9 (1.74)
4.0 (1.75)
3.2 - 3.3 (1.78)
3.4
3.4 (1.75)
3.5 (1.72)
3.45 (1.70)
L3
3.55 (1.74)
3.5 (1.70)
L4
3.6 (1.69)
Clare Basin
8
2.9
3.0 (1.76)
3.5
N
6.2
6.0
6.0
3.0
11
6.0
5.8
5.9
S
N
5.9 - 6.0
5.4
5.7
5.3
11
3.6 (1.74)
L5
3.75 (1.76)
L6
L7
3.95 (1.75)
4.05
4.6 (1.74)
4.1 (1.76)
Mantle
Figure 9. (a) P-wave velocity model for line A (for location, see Fig. 1). Numbers indicate the P-wave velocities in km s−1 . Open triangles mark the shot
points, while filled triangles show location of seismic sections in Figs 2–4. (b) S-wave velocity model for line A. Numbers are S-wave velocities in km s−1 .
Numbers in brackets are the V p /V s ratios. Dashed lines indicate the area resolved by reflected and/or refracted rays.
A notable exception is the laterally uniform S-wave velocity at
80 km along line B (at ca. 5 km depth) in the upper crust, where
a low velocity zone is present in the P-wave model. The S-wave
velocities are laterally very constant, particularly in the middle and
lower crust (varying by less than 0.05 km s−1 ). Upper mantle shear
wave velocities, defined by weak S n arrivals at large offset on the
longer line A, are about 4.6 km s−1 (Fig. 9b).
6 VA R I AT I O N S I N P O I S S O N ’ S R AT I O
The P- and S-wave crustal velocities are used to compute variations
in Poisson’s ratio to provide constraints on crustal composition.
Temperature and pressure variations within the crust have little
effect on Poisson’s ratio above pressures of 100–200 MPa. Numerous studies have shown that above these pressures fluid filled open
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
(a)
13
Line B
S
Killarney-Mallow
Fault
Munster Basin
33
N
19
5.0
16
5.5 -
5.4 - 5.7
5.5
5.7
5.8
5.6 - 5.9
5.9
6.0
5.9
6.0
5.0
5.9
6.05
6.1
6.0 - 6.15
6.25
6.4
6.25
6.5 - 6.6
6.6
6.6
6.8
6.9
6.9
7.1
7.15
8.0
(b)
S
Killarney-Mallow
Fault
Munster Basin
33
3.0 (1.67)
3.3 (1.64)
N
19
16
3.0
3.3 (1.69)
3.4 (1.73)
L2
3.4 (1.67)
3.5
L3
3.5 - 3.6 (1.66)
3.6 (1.69)
L4
3.65 (1.71)
L5
3.75 (1.76)
3.9 (1.77)
4.05 (1.76)
4.6 (1.74)
L6
L7
Mantle
Figure 10. (a) P-wave velocity model for line B (for location, see Fig. 1). Numbers indicate the P-wave velocities in km s−1 . Triangles mark the shot points,
with filled triangles showing locations of seismic sections in Figs 5–7. (b) S-wave velocity model for line B. Numbers are S-wave velocities in km s−1 . Numbers
in brackets are the V p /V s ratios. Dashed lines indicate the area resolved by reflected and/or refracted rays (see Fig. 8).
fractures and pore spaces in rocks generally become closed and variations in Poisson’s ratio become solely dependant on rock mineralogy (Christensen 1996). This means that laboratory determinations
of Poisson’s ratio should be directly applicable over a wide range of
crustal depths below about 3–4 km. Poisson’s ratios for both profiles
A and B were computed from the derived P- and S-wave models.
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation The value of Poisson’s ratio (σ ) for seismically isotropic materials
(see Christensen (1996) for detailed discussion) as a function of V p
and V s velocities is defined as
σ =
1
1
1−
2
(V p /Vs )2 − 1
(1)
14
F. Hauser et al.
Table 6. Average values of seismic properties in different types of crust
from Christensen & Mooney (1995) and Christensen (1996) compared with
the Irish crustal measurements (this study).
V p (km s−1 ) V s (km s−1 ) V p /V s
Crust type
Oceanic crust
Arc crust
Average continental
Irish crust (this study)
6.50
6.69
6.45
6.46
3.49
3.78
3.65
3.73
1.89
1.77
1.77
1.73
σ
t (km)
0.305
0.266
0.265
0.247
7 .0
39.0
41.0
32.0
6.1 Petrological implications
A comprehensive analysis of the experimental data that defines
the relationship between the bulk mineralogy of crustal rocks and
Poisson’s ratio (Christensen 1996) can be used to infer the chemical
composition of the crust. Poisson’s ratio, measured at 600 MPa,
shows a very high degree of correlation with silica content (%SiO 2 )
for a variety of metasedimentary and igneous rocks including felsic
granulites, metagreywackes, granites and diorites. For crustal rocks
with more that 55 per cent SiO 2 , σ varies systematically with SiO 2
content according to the simple quadratic rule:
Sensitivity factor, f(η)
%SiO2 = 100.9 − 496σ 2
Vp / Vs
Figure 11. The sensitivity factor f (η) plotted against the V p /V s ratio showing how relative errors in Poisson’s ratio relate to velocity uncertainties (see
text for more information).
and the values are shown on the S-wave model profiles in Figs 9b
and 10b. The mean value of σ for the entire crust is given in Table
6, together with average values for crustal thickness and V p and
V s wave velocities. These are compared with the global averages
reported by Christensen & Mooney (1995) for accreted arc, average
continental crust and oceanic crust.
Errors in Poisson’s ratio, with respect to errors in V p and V s
depend very strongly on the V p /V s ratio, and this is important when
considering the implications for crustal petrology and chemistry.
The error in σ associated with the respective errors in velocities
(V p and V s ) is greater for lower values of V p /V s (Christensen
1996) and is given by
σ = f (η) V p − Vs V V σ p
s
(2)
2
2η
where η = V p /V s and the sensitivity factor, f (η) = (η2 −1)(η
2 −2)
Fig. 11 shows how the sensitivity factor f(η) for the errors varies
with V p /V s over the range in σ values found in this study. The error
estimates in the P- and S-wave velocity models, given in Tables 2,
3, 4 and 5 (0.05 km s−1 ) for VARNET lines A and B, imply that the
corresponding errors in σ are typically better than 0.02 for the whole
range in σ values determined from the analysis of the seismic wideangle data. For example, perturbing the S-wave velocity values by
more than ±1 per cent very noticeably reduces the visual fit of the
calculated traveltime curves to the onset time of very clear S-wave
arrivals for both S g and S m S phases. It also significantly degrades
the value of both the RMS residual and the chi-squared measure
(see Tables 2, 3, 4 and 5). This means that variations in σ of within
∼0.01 across the crustal models are significant when it comes to
inferring changes in crustal petrology and composition.
(3)
with a correlation coefficient r ∼ 0.99 over a broad range in σ
from 0.1 (quartzite) to 0.30 for slates (Christensen 1996). This is
because in rocks with bulk SiO 2 content in excess of 55 per cent,
compressional wave velocities decrease and shear wave velocities
increase with SiO 2 content.
The validity of this relationship appears to be borne out by numerous studies, using different seismological techniques to resolve
crustal properties, in terranes of different geological age and geographical locality (Christensen 1996; Egorkin 1998; Chevrot &
van der Hilst 2000; Morozov et al. 2003; Mechie et al. 2005). In
metabasic rocks with less than 55 per cent SiO 2 content, which are
thought to be common in the lower crust (Holbrook et al. 1988)
and upper mantle, there is no correlation between silica content and
Poisson’s ratio.
Laboratory measurements indicate that for seismic wave speeds
that encompass the range in mineral assemblages typical of lower to
middle crustal rocks, eq. (3) can be applied to deduce silica content
when Poisson’s ratio (σ ) is less than 0.27, and if the compressional
wave speed (V p ) is below 7.4 km s−1 (see Fig. 13 and Table 3 in
Christensen (1996). Since the seismic properties of the crust determined along lines A and B (Fig. 1) satisfy these criteria, they can
be used to infer the bulk silica content of the entire crust and to
constrain its chemical composition at various depths.
Results from this calculation are shown in Figs 12(a) and (b),
where the variation of silica content in the crust along both profiles
is presented as grey-scale images. Variations in σ in the upper crust
are consistent with the mineralogy of rock exposed on the surface
along the profiles, particularly with the disposition of carbonate and
quartz lithologies within the Clare and Munster basins (Fig. 1).
Throughout the investigated crustal section, the value of Poisson’s
ratio is consistently less than that of the average petrological crustal
model of Christensen & Mooney (1995) and Christensen (1996)
(see Table 6) and agrees with independent estimates from receiver
function studies near line A (Landes et al. 2006). This demonstrates
that it is more felsic (∼70 per cent SiO 2 ) in bulk composition than
‘average’ continental crust (∼60 per cent SiO 2 ) reported in many
publications since the 1920s (see Table 9 in Christensen & Mooney
(1995)). The variations in the upper to mid crust are consistent with a
granite-granodiorite-diorite or meta-greywacke bulk composition.
More significantly, within the lower 10 km of crust the value of
σ is 0.257–0.265 (corresponding to a bulk silica content of ∼64 per
cent), which is markedly less than the global average value of about
0.281 predicted in the average crustal petrology model (Christensen
1996). The calculated σ value for the Irish lower crust is consistent
with a more felsic granulite composition, which is supported by
petrophysical, petrological and geochemical measurements on exceptionally well-preserved lower crustal xenoliths in Carboniferous
volcanics from central Ireland (Van den Berg et al. 2005).
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
(a)
Line A
S
Killarney-Mallow
Fault
Shannon
Estuary
Munster Basin
0.2
0.257
0.287
0.248
0.242
L3
0.221
0.242
L4
0.226
0.257
L5
0.256
0.260
L6
0.258
N
Clare Basin
0.244
L2
0.205
15
0.256
L7
0.260
0.249
Mantle
Line B
(b)
S
Killarney-Mallow
Fault
Munster Basin
0.201
0.237
0.241
0.251
0.202
L2
0.248
0.21
N
0.25
0.231
L3
0.214
L4
0.235
L5
0.232
0.262
(~64%) 0.265
(~64%) 0.264
L6
L7
Mantle
Figure 12. (a) Silica content (%SiO 2 ) determined from eq. (3) for Poisson’s ratio values less than 0.27 across line A. The upper part of the crust to a depth of
about 17 km is characterized by higher silica content (up to about 75 per cent) suggesting quartzofeldspathtic mineralogies with rock types such as granodiorites,
metagreywackes and quartzofeldspathic sediments. The variation in the upper crust correlates broadly with the surface distribution of carbonate and silica rich
Palaeozoic rocks. The lower crust below 17 km is more depleted in silica and consistent with a dioritic mid crust and a lower crust (ca. 10 km thick) composed
of restitic silica rich (∼ 64 per cent) paragneisses (see text). (b) Silica content (%SiO 2 ) for Poisson’s ratio values less than 0.27 across line B. As on line A,
the upper part of the crust to a depth of about 17 km is characterized by a higher silica content up to about 75 per cent, consistent with quartzofeldspathic
rock mineralogies. The variation in the upper crust, between distances of 0 and 65 km, correlates broadly with the silica rich Palaeozoic clastic rocks of the
tectonically inverted Munster basin (see Fig. 1). The zone of low SiO 2 content at the northern end of the line, between a distance of 100 and 140 km, correlates
with Carboniferous rock within the offshore extension of the Clare Basin. SiO 2 percentages within the crust below 17 km are consistent with a dioritic mid
crust and a lower crust (ca. 10 km thick) composed of restitic metapelites and psammites. The higher SiO 2 content at 80 km distance and 20 km depth suggests
a more felsic granodioritic composition. This region appears to continue upwards towards the surface, where it lies on the flanks of a Bouguer gravity low
perhaps due to buried granite plutons at depth (see text).
6.2 Correlations with lower crustal xenoliths
from central Ireland
Lower crustal xenoliths from central Ireland, close to the trace of
the Iapetus Suture Zone, were brought to the surface during lower
Carboniferous volcanism and consist predominantly of granulite
facies (garnet-sillimanite-K-feldspar) metapelites and psammites
(see Fig. 1 and Van den Berg et al. (2005) for detailed xenolith
locations). This relatively felsic composition of the xenolith suite
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation is unusual, compared with the predominantly mafic composition of
lower crustal xenoliths, found worldwide (Griffin & O’Reilly 1987;
Burke & Fountain 1990; Rudnick & Fountain 1995).
Many of the metapelitic xenoliths are peraluminous khondalites (garnet-sillimanite-quartz-K-feldspar-rutile ± plagioclase ±
ilmenite ± biotite assemblages) and probably represent the residue
of partial melting of lower crustal metasedimentary rocks. This is
consistent with their high modal abundance of aluminosilicate minerals such as sillimanite and garnet (Van den Berg et al. 2005),
F. Hauser et al.
Table 7. Major element compositions of lower crustal xenoliths from central
Ireland comprising khondalite (mean of 16 samples), kinzigite, psammite
and granitic orthogneiss (individual representative samples).
81.38
0.57
7.83
6.11
0.40
1.18
0.79
0.08
1.43
0.07
−0.10
99.74
72.65
0.37
13.63
3.27
0.05
0.72
3.18
4.16
0.58
0.06
1.26
99.97
65.81
0.87
15.79
6.79
0.29
2.25
2.89
1.94
1.77
0.16
1.28
99.86
kh = khondalite, ki = kinzigite, ps = psammite, orth = granitic
orthogneiss. Following determination of the loss on ignition (LOI),
whole-rock analyses were obtained by standard XRF methods on fused
glass discs at the University of Leicester.
high Al 2 O 3 contents (Table 7) and the presence of quartzofeldspathic leucosomes (Van den Berg et al. 2005). Some samples rich
in sillimanite aligned along the foliation are highly anisotropic with
respect to both P- and S-wave velocities (up to 14 per cent).
Thermobarometry indicates that the metapelitic xenoliths were
entrained from ca. 20–28 km depth within the lowermost crust
and rapidly transported to the surface. Rarer mafic granulites also
occur and are sourced from depths of 30–35 km, i.e. from below
the present-day seismic refraction Moho. However, the origin and
petrological significance of the mafic granulites is not understood
because of their rarity and high degree of alteration.
Excluding highly anisotropic samples, the velocity measurements
of the granulite facies metapelite xenolith suite from the Irish Midlands (red circles in Fig. 13) correlate very well with the wide-angle
seismic velocities and V p /V s ratios. This suggests that the samples
are close to the bulk composition and mineralogy of the lower crust
beneath Ireland. Moreover, the composition inferred from the σ values (∼64 per cent SiO 2 , Fig. 12b—layers L6 and L7) is remarkably
close to the unweighted arithmetic mean of whole-rock chemical
analyses of the central Irish xenoliths suite (Table 7) comprising
khondalite, kinzigite, psammite and granitic orthogneiss.
7 DISCUSSION
In a global context the seismic velocity structure of the crust beneath central and southwest Ireland is anomalous with a crustal
thickness of 30–34 km, well below the global average of 41 km
reported by Christensen & Mooney (1995) and is more typical of
extended crust in their classification. However, there this similarity
with ‘extended crust’ breaks down as the average crustal velocity of the Irish crust is identical to the global average value of
∼6.45 km s−1 (Table 6). The mean value of Poisson’s ratio (σ ) for
the entire crust (0.24–0.25) is below the global average (0.265) and
its value for the lower 10 km of crust (0.258) is particularly anomalous, indicating an unusually silica rich composition compared to
that indicated elsewhere from seismic observations and studies of
lower crustal xenoliths (Holbrook et al. 1988; Rudnick & Fountain
1995). These results are robustly supported and ‘ground-truthed’ by
the geochemical and petrophysical measurements on lower crustal
xenoliths from the Irish Midlands.
Layer 6
Layer 5
a
5
58.18
1.13
15.90
5.47
0.07
3.62
6.90
2.58
1.96
0.45
3.62
99.86
0.1
51.04
1.42
25.78
12.30
0.64
3.50
0.70
0.94
3.13
0.06
0.36
99.87
Layer 4
a
a
0.1
0
Mean (unweighted)
5
Orth
0.2
ps
0.3
0
ki
5
SiO 2
TiO 2
Al 2 O 3
Fe 2 O 3
MnO
MgO
CaO
Na 2 O
K2O
P2O5
LOI
Total
kh
0.3
Lithology
Layer 7
0.2
0
16
a
Figure 13. Plot of compressional (V p ) versus shear wave (V s ) velocities
with lines of constant Poisson’s ratio (e.g. 0.25) indicated. Measured seismic wide-angle velocities (solid black symbols), corrected back to room
temperature (using the same thermal properties as in Van den Berg et al.
(2005)) are indicated for the various mid to lower crustal model layers in
Figs 9 and 10 (e.g. Layer 4–7). Green circles are experimental data from the
ACCRETE experiment (Morozov et al. 2003) measured at different pressures (400–600 MPa) and room temperature. Red circles are measurements
from the Central Irish Xenolith Suite (CIXS), measured at 600 MPa and
room temperature (from Van den Berg et al. 2005). Seismic velocity properties are consistent with diorite/granodiorite composition for Layer 4, based
both on the ACCRETE measurements and the compilation of petrophysical measurements from Christensen (1996). Metapelitic granulites from the
Central Ireland Xenolith Suite have velocity properties identical to Layers
6 and 7. Samples labelled (a) show a high degree (>10 per cent) of seismic
anisotropy, due to the presence of aligned mineral phases, predominantly
sillimanite. Excluding the highly anisotropic samples, the experimental data
on xenolith and rock samples are remarkably close to the model for mid
to lower crustal composition (this study). Lower crustal restites from ACCRETE samples (green dots near V p = 7 km s−1 ) are likely more mafic in
composition than the CIXS examples (i.e. their Poisson’s ratio is appreciably
higher).
The combined evidence from this seismic study and the results
from the Irish Midlands xenolith suite (Van den Berg et al. 2005)
provides support for accumulation of new crust by tectonic accretion of largely juvenile volcaniclastic sediments, mainly of Lower
Palaeozoic age, close to the Iapetus suture zone, followed by partial
melting and emplacement of granitic plutons during and following the Caledonian Orogeny (Klemperer et al. 1991; Daly & Van
den Berg 2004). The geochemical and geochronological measurements from lower crustal xenoliths date this melting event to the
late Silurian/early-mid Devonian, towards the end of Caledonian
deformation.
This age is what would be expected from the conventionally
accepted view of the Caledonian tectonics of Ireland. U-Pb zircon ages from some lower crustal xenoliths, however, indicate that
partial melting of the felsic lower crust persisted until the late Devonian to early Carboniferous, implying that it was anomalously hot
and therefore rheologically very mobile well after the end of Caledonian deformation. Whether these younger events are connected
to the extensional tectonism that led to the formation of ensialic
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
Carboniferous basins across southern and central Ireland is uncertain and requires further investigation (Daly & Van den Berg 2004).
As the value of Poisson’s ratio is important in defining the bulk
composition of the crust, there have been many worldwide studies
using receiver functions based on differential P m S-P and Pp m S-P
traveltimes, to determine crustal thickness and the value of the ratio.
The global study of Zandt & Ammon (1995) focussed on variations
in Poisson’s ratio as a function of crustal age and concluded that the
ratio increased with the age of crustal formation, being higher in
Archaean than in Palaeozoic and Cenozoic regions. Measurements
of σ along deep-probing long-range seismic refraction profiles in
the former Soviet Union (Egorkin 1998) and in Australia (Chevrot
& van der Hilst 2000) suggest that σ tends to decrease with crustal
thickness in tectonic domains of Phanerozoic age (i.e. Palaeozoic
platforms and Cenozoic arcs). The results of the present study on
the Irish crust are consistent with the latter observation.
The xenolith suite from central Ireland is similar in petrology and
seismic properties to samples of mid to lower crustal rock gathered
across the Cenozoic Coast Mountains Batholith (Fig. 13) and the adjacent accreted terranes in southeastern Alaska and western British
Columbia, where similar controlled source seismic techniques have
been employed to investigate crustal composition (Morozov et al.
2003). This suggests that a fundamentally similar process of crustal
formation by melting and differentiation was at work in both regions during different geological epochs. Crustal growth in western
British Columbia is interpreted to have involved processes associated with the flow of hot crustal material during crustal shortening
and the vertical accretion of mantle derived mafic melts (Hollister
& Andronicos 2006). The crustal structure of the Jurassic to Eocene
Coast Mountains Orogen of southeast Alaska and British Columbia
shows large changes in V p , V s and Poisson’s ratio across a crustalscale suture zone within the lower 10–15 km of crust (Morozov
et al. 2003). Large amounts of exhumation of upper to middlecrustal rock is associated with gabbroic magmatism and crustal
underplating, the extraction of granites from the lower crust and
the consequent formation of lower crustal granulite facies restite
assemblages, rich in garnet and sillimanite (Hollister & Andronicos
2006).
By contrast, in the Irish crust, late Caledonian deformation during
the final closure of the Iapetus Ocean involved a large component
of sinistral strike slip on the major faults (Soper et al. 1992; Dewey
& Strachan 2003), rather than the great amounts of crustal shortening observed in the Canadian Cordillera. This explains the lack
of exhumation of deep (midcrustal) rock, the relatively low upper
crustal seismic velocities (Figs 9 and 10) and the absence of geological evidence for very large amounts of crustal thickening and
magmatic underplating (Van den Berg et al. 2005). Rocks of Ordovician/Silurian to early Devonian age are exposed below a late
Palaeozoic (late Devonian to Carboniferous) sedimentary sequence
across the midlands and south of Ireland, implying that uplift and
exhumation of rock was small. The crustal structure now observed
(this study) has remained essentially the same since the xenolith
suite was emplaced during the early Carboniferous.
The crustal model for the Irish crust, presented in this study,
shows a subdivision of the lower crust into two distinct layers with
an aggregate thickness of about 10 km and with a uniform Poisson’s
ratio (σ = 0.257–0.265) diagnostic of an unusual felsic composition (Figs 9, 10 and 12). There is no unambiguous seismological
evidence for a significant mafic (underplated) component within
the lower crust or near its base, as has been inferred on the basis of
receiver function studies, which are partially constrained by wideangle seismic data beneath Ireland (Shaw Champion et al. 2006).
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation 17
The rare enigmatic mafic granulite xenoliths, which the thermobarometric studies indicate are sourced from close to or below the
present-day seismic refraction Moho (Van den Berg et al. 2005),
have uncertain geochemical affinities. They could have come from
a thin crust/mantle transition boundary layer made from accreted
slivers of Ordovician ocean crust, or from mafic intrusions within
the mantle itself.
Thus the impact of post-Palaeozoic tectonic and magmatic events,
widely manifest throughout the North Atlantic region as large Mesozoic basin systems and early Cenozoic basaltic igneous provinces,
does not appear to have dramatically modified the structure of the
crust beneath southwest Ireland. It seems that the process of crustal
formation within central and southern Ireland was essentially completed by the early Devonian with some degree of intracrustal melting continuing into the latest Devonian (Daly & Van den Berg
2004).
8 C O N C LU S I O N S
1. A joint analysis of controlled source shear (V s ) and compressional (V p ) wide-angle seismic data from southwest Ireland has
been used to derive the distribution of Poisson’s ratio (σ ) within the
crust. Using published petrophysical experimental data on V p , V s
and σ from different igneous and metasedimentary rocks a model
for crustal petrology and chemistry is developed. The mean value
of σ (∼0.247) for the Irish crust is unusually low compared to the
global average value (∼0.265) and indicates that it is unusually
felsic in bulk composition.
2. The values of Poisson’s ratio are highly variable in the uppermost 5 km of crust (σ = 0.20–0.28) and correlate well with the
surface distribution of silica (low σ ) and carbonate rich (high σ )
Palaeozoic sedimentary rocks. Values of Poisson’s ratio in the upper part of the crust, to a depth of about 15 km, suggest high silica
contents (SiO 2 ∼75 per cent) and quartzofeldspathic mineralogies,
such as granites/granodiorites and metagreywackes.
3. Within the lower 10 km of crust the variation in σ is smoother.
The mean value of σ (∼0.258) is anomalously low compared to the
global average (σ ∼ 0.281). This indicates that it is more depleted
in silica than the upper to mid crust; however, it still has an unusually silica-rich composition compared with the mafic granulite
composition that is generally considered typical.
4. A silica rich lower crust (SiO 2 ∼ 64 per cent) is supported
by petrophysical and geochemical data on exceptionally wellpreserved lower crustal xenoliths from nearby early Carboniferous
volcanic rocks. These xenoliths consist mostly of highly aluminous partially melted metapelites (khondalites) as well as normal
metapelites (kinzigites), psammites and granitic orthogneisses. The
khondalites have seismic properties identical to those derived from
the wide-angle seismic data for the lower crust. This suggests that
they faithfully represent its bulk chemical composition.
5. The combined xenolith data and seismic velocity models support tectonic theories for the origin of the central Irish crust by
tectonic accretion of predominantly sedimentary and volcaniclastic
material derived from oceanic, island-arc and continental margin
sources during the Caledonian orogenic cycle.
6. Partial melting of the crust at mid to lower crustal depths
began during the Caledonian producing granites and a chemically
differentiated crust. However, melting of the lower 10 km of crust
occurred as late as the latest Devonian; however, it is not known if
this is a discrete event or a continuation of the early melting history.
18
F. Hauser et al.
AC K N OW L E D G M E N T S
JSD and RvdB gratefully acknowledge the support of Enterprise
Ireland Basic Research Grant SC/1998/524 and a University College Dublin President’s Research Award. We also thank the late Tim
Brewer for access to XRF facilities at the University of Leicester.
The data gathered during the VARNET experiment were collected
using the seismic equipment of the GeoForschungsZentrum (GFZ)
Potsdam, the Dublin Institute for Advanced Studies and the University of Copenhagen. The support of the European Commission
by the EEC Human Capital and Mobility Program under contract
No. ERBCHRXCT940572 is acknowledged. Finally, we thank the
two anonymous reviewers and the editor Randy Keller for their detailed and constructive comments. This paper is contribution GP196
of the Geophysics Section of the Dublin Institute for Advanced
Studies.
REFERENCES
Bevington, P.R., 1969. Data Reduction and Error Analysis For The Physical
Sciences, McGraw-Hill, New York.
Burke, M.M. & Fountain, D.M., 1990. Seismic properties of rocks from an
exposure of extended continental crust—new laboratory measurements
from the Ivrea Zone, Tectonophysics, 182, 119–146.
Chevrot, S. & Van Der Hilst, R.D., 2000. The Poisson ratio of the Australian
crust: geological and geophysical implications, Earth Planet. Sci. Lett.,
183, 121–132.
Christensen, N.I., 1996. Poisson’s ratio and crustal seismology, J. geophys.
Res., 101, 3139–3156.
Christensen, N.I. & Mooney, W.D., 1995. Seismic velocity structure and
composition of the continental crust: a global view, J. geophys. Res., 100,
9761–9788.
Daly, J.S. & Van Den Berg, R. 2004. Ion microprobe U-Pb zircon evidence
from lower crustal xenoliths for Palaeozoic crust formation and crustal
development within the Caledonian suture zone in central Ireland. Geol.
Soc. Am. Abs. Prog., 36(5), 48.
Dewey, J.F. & Strachan, R.A., 2003. Changing Silurian-Devonian relative plate motion in the Caledonides: sinistral transpression to sinistral
transtension, J. Geol. Soc. Lond., 160, 219–229.
Egorkin, V.A., 1998. Velocity structure, composition and discrimination of
crustal provinces in the former Soviet Union, Tectonophysics, 298, 395–
404.
Griffin, W.L. & O’Reilly, S.Y., 1987. Is the continental Moho the crustmantle boundary?, Geology, 15, 241–244.
George, T.N., Johnson, G.A.L., Mitchell, M., Prentice, J.E., Ramsbottom,
W.H.C., Sevastopulo, G.D. & Wilson, R.B., 1976. A Correlation of Dinantian rocks in the British Isles. Geological Society Special Report
No. 7.
Hawman, R.B., Colburn, R.H., Walker, D.A. & Smithson, S.B., 1990. Processing and inversion of refracted and wide-angle reflection data from
1986 Nevada PASSCAL experiment, J. geophys. Res., 95, 4651–4691.
Higgs, K. & Russell, K.L., 1981. Upper Devonian microfloras from southest
Iveragh, County Kerry, Ireland, Bull. Geol. Surv. Ireland, 3, 17–50.
Holbrook, W.S., Gajewski, D., Krammer, A. & Prodehl, C., 1988. An interpretation of the wide-angle compressional and shear wave data in southwest Germany: Poisson’s ratio and petrological implications, J. geophys.
Res., 93, 12081–12106.
Hollister, L.S. & Andronicos, C.L., 2006. Formation of new continental
crust in Western British Columbia during transpression and transtension,
Earth planet. Sci. Lett., 249, 29–38.
Jacob, A.W.B., 1975. Dispersed shots at optimum depth—an efficient seismic source for Lithospheric studies, J. Geophys., 41, 63–70.
Jacob, A.W.B., Kaminski, W., Murphy, T., Philips, W.E.A. & Prodehl, C.,
1985. A crustal model for a northeast-southwest profile through Ireland,
Tectonophysics, 113, 75–103.
Klemperer, S.L., Ryan, P.D. & Snyder, D.B., 1991. A deep seismic reflection
transect across the Irish Caledonides. J. Geol. Soc. Lond., 148, 149–164.
Landes, M., Prodehl, C., Hauser, F., Jacob, A.W.B. & Vermuelen, N.J.,
2000. VARNET-96: influence of the Variscan and Caledonian orogenies
on crustal structure in SW Ireland, Geophys. J. Int., 140, 660–676.
Landes, M., O’Reilly, B.M., Readman, P.W., Shannon, P.M. & Prodehl, C.,
2003. VARNET-96: three-dimensional upper crustal velocity structure of
SW Ireland, Geophys. J. Int., 153, 424–442.
Landes, M., Ritter, J.R.R., O’Reilly, B.M., Readman, P.W. & Do, V.C. 2006.
A N-S receiver function profile across the Variscides and Caledonides of
SW Ireland, Geophys. J. Int., 166, 814–824.
Lowe, C. & Jacob, A.W.B., 1989. A north—south seismic refraction profile
across the Caledonian Suture zone in Ireland, Tectonophysics, 186, 297–
318.
Masson, F., Jacob, A.W.B., Prodehl, C., Readman, P.W., Shannon, P.M.,
Schulze, A. & Enderle, U., 1998. A wide-angle seismic traverse through
the Variscan of Southwest Ireland, Geophys. J. Int., 134, 689–705.
Mechie, J., Abu-Ayyash, K., Ben-Avraham, Z., El-Kelani, R., Mohsen, A.,
Ruempker, G., Saul, J. & Weber, M., 2005. Crustal shear velocity structure across the Dead Sea Transform from two-dimensional modelling of
DESERT project explosion seismic data, Geophys. J. Int., 160, 910–924,
doi: 10.1111/j.1365-246X.2005.02526.x.
Morozov, I.B., Smithson, S.B., Chen, J. & Hollister, L.S., 2001. Generation
of new continental crust and terrane accretion in southeastern Alaska and
western British Columbia: constraints from P- and S-wave wide-angle
seismic data (ACCRETE), Tectonophysics, 341, 49–67.
Morozov, I.B., Christensen, N.I., Smithson, S.B. & Hollister, L.S., 2003.
Seismic and laboratory constraints on crustal formation in a former
continental arc (ACCRETE, southeastern Alaska and western British
Columbia), J. geophys. Res., 108, doi:10.1029/2001JB001740.
Nair, S.K., Gao, S.S., Liu, K.H. & Silver, P.G., 2006. Southern African
crustal evolution and composition: constraints from receiver function
studies, J. geophys. Res., 111, B02304, doi:10.1029/2005JB003802.
Naylor, D., Sevastopulo, G.D. & Sleeman, A.C., 1989. Subsidence history
of the South Munster Basin, Ireland, in The role of Tectonics in Devonian
and Carboniferous Sedimentation in the British Isles, Vol. 6, pp. 99–
109, eds Arthurton, R.S., Gutteridge, P. & Nolan, S.C. Occasional Publ.,
Yorkshire Geol. Soc.
Price, C.A. & Todd, S.P., 1988. A model for the development of the Irish
Variscides, J. Geol. Soc. Lond., 145, 935–939.
Readman, P.W., O’Reilly, B.M. & Murphy, T., 1997. Gravity gradients and
upper-crustal tectonic fabrics, Ireland, J. Geol. Soc. Lond., 154, 817–828.
Readman, P.W., O’Reilly, B.M., Murphy, T. & Makris, J. 2003. A gravity
anomaly map of the Irish western seaboard, Irish J. Earth Sci., 21, 133–
142.
Rudnick, R.L. & Fountain, D.M., 1995. Nature and composition of the
continental crust: a lower crustal perspective, Rev. Geophys., 33, 267–
309.
Shaw Champion, M.E., White, N.J., Jones, S.M. & Priestley, K.F., 2006.
Crustal velocity structure of the British Isles; a comparison of receiver
functions and wide-angle seismic data, Geophys. J. Int., 166, 795–813,
doi: 10.1111/j.1365-246X.2006.03050.x.
Soper, N.J., Strachan, R.A., Holdsworth, R.E., Gayer, R.A. & Greiling,
R.O., 1992. Sinistral transpression and the Silurian closure of the Iapetus,
J. Geol. Soc. Lond., 149, 871–880.
Stammler, K., 1994. SeismicHandler—programmable multichannel data
handler for interactive and automatic processing of seismological data,
Comp. Geosci., 19(2), 135–140.
Van Den Berg, R., Daly, J.S. & Salisbury, M.H., 2005. Seismic velocities
of granulite-facies xenoliths from Central Ireland: implications for lower
crustal composition and anisotropy, Tectonophysics, 407, 81–99.
Williams, E. et al., 1989. Tectonic controls and sedimentary response in
the Devonian-Carboniferous Munster and South Munster basins, SouthWest Ireland. in The Role of Tectonics in Devonian and Carboniferous
Sedimentation in the British Isles, Vol. 6, pp. 123–141, eds Arthurton,
R.S., Cutteridge, P. & Nolan, S.C. Occasional Publ., Yorkshire Geol. Soc.
Zandt, G. & Ammon, C.J., 1995. Continental crust composition constrained
by measurements of crustal Poisson’s ratio, Nature, 374, 152–154.
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation Constraints on crustal structure and composition
Zelt, C.A., 1999. Modelling strategies and model assessment for wide-angle
seismic traveltime data, Geophys. J. Int., 139, 183–204.
Zelt, C.A. & Forsyth, D.A., 1994. Modeling wide-angle seismic data for
crustal structure: southeastern Grenville province, J. geophys. Res., 99,
11687–11704.
C 2008 The Authors, GJI
C 2008 RAS
Journal compilation 19
Zelt, C.A. & Smith, R.B., 1992. Seismic traveltime inversion for 2D crustal
velocity structure, Geophys. J. Int., 108, 16–34.
Ziegler, P.A., 1986. Geodynamic model for the Palaeozoic crustal conditions of Western and Central Europe, Tectonophysics, 126, 303–
328.