Constraints on the origin of slab and mantle wedge anomalies in Tonga from the ratio of S to P velocities
Abstract
We examine two prominent upper mantle velocity anomalies in the southwest Pacific, the Tonga slab anomaly and the corresponding overlying mantle wedge anomaly, using data collected during a combined land-sea deployment of temporary seismometers. The linear geometry and small interstation spacing of the instruments yield high-resolution data along a cross section of the Tonga subduction zone, including the actively spreading Lau back arc basin. We estimate the relative variation of P and S velocity, often described as v = 8\uV s /8lnV p , for the slab and mantle wedge anomalies using two distinct methods: a linear regression of the P and S travel time residuals, and detailed modeling of the velocity structure using a three-dimensional finite difference travel time algorithm. The two methods yield similar results, with v of the slab being 1.1-1.5 and v of the mantle wedge being 1.2-1.3. These values are consistent with experimental data concerning the effect of temperature on P and S wave velocities in the upper mantle and are lower than what is expected for velocity anomalies generated by the presence of partial melt. These observations imply that either the theoretical estimates of v for partial melt are too large or very little partial melt is present beneath the Lau basin. In the latter case, melt must be quickly removed from the rock matrix, such that the velocity anomalies are due to increased temperature, and not melt. The bulk of the velocity anomaly in the mantle wedge can be explained by temperature anomalies of 400-600° C because of the amplification of temperature derivatives of seismic velocity by anelastic effects. Such large thermal anomalies, generated by decreased lithospheric thickness and mantle upwelling beneath the fast spreading Lau back arc basin, can still leave the mantle near the solidus, even after accounting for the effect of increased volatile content in the mantle wedge. The lower-amplitude velocity reductions in the deeper wedge are likely related to an increased concentration of volatiles from the subducting slab.
... Beside mantle models from the global tomography imaging, the intense seismic activity in the Tonga subduction region offers the opportunity for regional tomography studies with regional-scale details utilizing intermediate and deep-focus earthquakes in the slab. Between 1993 and 1995, elaborate land experiment SPASE and passive seismic ocean bottom experiment LABBATS were designed across the northern Tonga and Fiji (Wiens et al., 1995) providing information on the structure in the mantle wedge and the back-arc Lau Basin (Zhao et al., 1997;Xu and Wiens, 1997;Koper et al, 1999;Conder and Wiens, 2006). These experiments improved our knowledge of the region and emphasized the importance of regional studies; however, they retained the following limitations. ...
... However, Zhao et al. (1997) did not include the S-waves in their interpretation and, therefore, no Vp/Vs constraints on a possible distribution of melt and volatiles were provided. • Koper et al. (1999) used the P-and S-wave traveltimes from selected earthquakes to investigate a fractional change in the S-wave velocities relative to the P-wave velocities (ratio dlnVs/dlnVp). They found that this ratio was 1.2-1.3 for the whole mantle wedge region, consistently with the experimentally predicted values mapping the effect of temperature on the P and S velocities. ...
... The Tonga wedge and the associated back-arc Lau Basin are imaged by low-velocity anomalies (Zhao et al., 1997;Koper et al., 1999;Conder and Wiens, 2006) in agreement with the results from attenuation studies detecting an extensive zone of high attenuation in the uppermost mantle (Roth et al., 1999;Wei and Wiens, 2018). This low-velocity zone extends westwards afar off the active Lau spreading centre and imply high temperatures in the upper mantle related to the magma production generating the fast-spreading back-arc Lau Basin (Zhao et al., 1997;Martinez and Taylor, 2002). ...
The Tonga subduction zone in the south-west Pacific is the fastest convergent plate boundary in the world with the most active mantle seismicity. This zone shows unique tectonic features including Samoan volcanic lineament of plume-driven origin near the northern rim of the Tonga subducting slab. The proximity of the Samoa hotspot to the slab is enigmatic and invokes debates on interactions between the Samoa plume and the Tonga subduction. Based on long-term observations of intermediate and deep-focus Tonga earthquakes reported in the Global Centroid Moment Tensor (CMT) catalog, we provide novel detailed imaging of this region. Accurate traveltime residua of the P- and S-waves recorded at two nearby seismic stations of the Global Seismographic Network are inverted for the P- and S-wave velocities and their ratio and reveal their pronounced lateral variations. In particular, they differ for the southern and northern parts of the Tonga subduction region. While no distinct anomalies are detected in the southern Tonga segment, striking low-velocity anomalies associated with a high Vp/Vs ratio are observed in the northern Tonga segment close to the Samoa plume. These anomalies spread through the whole upper mantle down to depths of ~ 600 km. Together with the fast extension of the northern back-arc Lau Basin, slab deformation and geochemical enrichment in the northern Tonga region, they trace deep-seated magmatic processes and evidence an interaction of the Tonga subduction with the Samoa plume.
... Geophysical and mineral-physics studies suggest that mantle temperature variations produce R S,P of magnitudes in the range 1.1-2.2 (Cammarano et al., 2003;Goes et al., 2000;Koper et al., 1999). The effect of composition on V P and V S is small compared to that of temperature (for most plausible compositions) because of the very strong temperature sensitivity and therefore distinguishing between the two remains complicated (Goes et al., 2000). ...
... Previous studies using the ratio of P wave to S wave relative traveltime residuals found that regions with anomalies caused by thermal variations only have a S,P ratios that vary between 1.8 and 2.2 (R S,P = 1.1-1.3, Chung, 1971;Koper et al., 1999). Several other studies found that an a S,P of around 2.9 may be on the high side to invoke only temperature perturbations (e.g., Gao et al., 2004), and values >2.9 likely require the presence of compositional heterogeneities (Rocha et al., 2011). ...
Independent models of P wave and S wave velocity anomalies in the mantle derived from seismic tomography help to distinguish thermal signatures from those of partial melt, volatiles, and compositional variations. Here we use seismic data from SW Europe and NW Africa, spanning the region between the Pyrenees and the Canaries, in order to obtain a new S‐SKS relative arrival‐time tomographic model of the upper mantle below Iberia, Western Morocco, and the Canaries. Similar to previous P wave tomographic results, the S wave model provides evidence for (1) subvertical upper‐mantle low‐velocity structures below the Canaries, Atlas Ranges, and Gibraltar Arc, which are interpreted as mantle upwellings fed by a common lower‐mantle source below the Canaries; and (2) two low‐velocity anomalies below the eastern Rif and Betics that we interpret as the result of the interaction between quasi‐toroidal mantle flow induced by the Gibraltar slab and the mantle upwelling behind it. The analysis of teleseismic P wave and S wave arrival‐time residuals and the conversion of the low‐velocity anomalies to temperature variations suggest that the upwellings in the upper mantle below the Canaries, Atlas Ranges, and Gibraltar Arc system may be solely thermal in nature, with temperature excesses in the range ~100–350 °C. Our results also indicate that local partial melting can be present at lithospheric depths, especially below the Atlas Ranges. The locations of thermal mantle upwellings are in good agreement with those of thinned lithosphere, moderate to high heat‐flow measurements, and recent magmatic activity at the surface.
... We should note that there are large variations in δ ln V S /δ ln V P and ν ≡ δ ln ρ/δ ln V S in the earth. Koper et al. (1999) estimated δ ln V S /δ ln V P to be in the range 1.1-1.5 in the Tonga Slab. Romanowicz (2001) estimated δ ln V S /δ ln V P to be larger than 2.5 in the lower mantle at larger length scales. ...
Recent seismological observations focusing on the collapse of an impulsive
wavelet revealed the existence of small-scale random heterogeneities in the earth medium.
The radiative transfer theory (RTT) is often used for the study of the propagation and
scattering of wavelet intensities, the mean square amplitude envelopes through random
media. For the statistical characterization of the power spectral density function (PSDF)
of the random fractional fluctuation of velocity inhomogeneities in a 3-D space, we use
an isotropic von Kármán-type function characterized by three parameters: the root
mean square (RMS) fractional velocity fluctuation, the characteristic length, and the
order of the modified Bessel function of the second kind, which leads to the power-law
decay of the PSDF at wavenumbers higher than the corner. We compile reported statistical
parameters of the lithosphere and the mantle based on various types of measurements for a
wide range of wavenumbers: photo-scan data of rock samples; acoustic well-log data; and
envelope analyses of cross-hole experiment seismograms, regional seismograms, and
teleseismic waves based on the RTT. Reported exponents of wavenumber are distributed
between −3 and −4, where many of them are close to −3. Reported RMS fractional
fluctuations are on the order of 0.01–0.1 in the crust and the upper mantle. Reported
characteristic lengths distribute very widely; however, each one seems to be restricted
by the dimension of the measurement system or the sample length. In order to grasp the
spectral characteristics, eliminating strong heterogeneity data and the lower mantle
data, we have plotted all the reported PSDFs of the crust and the upper mantle against
wavenumber for a wide range (10−3–108 km−1). We find that the spectral
envelope of those PSDFs is well approximated by the inverse cube of wavenumber. It suggests that the earth-medium randomness has a broad
spectrum. In theory, we need to re-examine the applicable range of the Born approximation
in the RTT when the wavenumber of a wavelet is much higher than the corner. In
observation, we will have to carefully measure the PSDF on both sides of the corner. We
may consider the obtained power-law decay spectral envelope as a reference for studying
the regional differences. It is interesting to study what kinds of geophysical processes
created the observed power-law spectral envelope at different scales and in different
geological environments in the solid earth medium.
... Assuming equality of P-and S-velocity fluctuations (d ln V S =d ln V P 1) simplifies the complexity of the g functions. Koper et al. (1999) obtained 1.1-1.5 for d ln V S =d ln V P from measurements in the crust and upper mantle between 10 and 100 km depth. ...
Using envelopes of regional high frequency seismograms synthesized by an algorithm from 3-D radiative transport theory (RTT), we investigate tradeoffs between the velocity fluctuation parameter epsilon , the corner parameter "a" of the von Kármán heterogeneity spectrum, and intrinsic attenuation 1/Qint. In the frequency band 1 to 4 Hz and range 100 to 1500 km, where S waves comprise the crustally guided Lg wave, the tradeoff between epsilon and "a" often lies between two regimes of scattering, one in which epsilon trades off with "a" such that either epsilon**2*a or epsilon**2/a is approximately constant. We suggest methods for identifying the boundary between these regimes: (1) varying and , and intrinsic attenuation using RTT to synthesize waveform envelopes for a small number of frequencies, and/or (2) varying epsilon and "a" , and 1/Qint in a larger series of simulations appropriate for multiple lapse-time window analysis (MLTWA). A stable region for calibrating MLTWA plots, where the seismic albedo is close to 0.5, is helpful in reducing the uncertainties of intrinsic and scattering attenuation when one of the attenuations is much larger than the other.
Bull. Seism. Soc. Am., vol. 109, pp. 312-321, 2019.
doi: 10.1785/0120180231
... Experimental verifications of this assumption are few. From a local array above the Tonga subduction zone, Koper et al. (1999) estimated d ln V S /d ln V P to be in the range 1.1 to 1.5 at scale lengths on the order of 10 to 100 km. In the lower mantle at larger scale lengths (~1000 km), d ln V S /d ln V P has been estimated to be > 2.5 (Romanowicz, 2001). ...
This dissertation presents Radiative3D, a computer code for radiative transport simulations of seismic events, including earthquakes and explosions, in 3-D Earth models, and allowing output of seismogram envelopes from virtual seismometers or information representing bulk energy transport through the model. Radiative transport is an efficient algorithm for Monte-Carlo simulation of high-frequency elastic wavefield energy transport through models featuring both large-scale (larger than wavelength) structure, simulated by deterministic ray tracing, and small-scale structure (wavelength scale and smaller), simulated by a stochastic scattering process. The 3-D models are composed of a collection of adjoining model cells inside of which material properties have simple mathematical description. Within cells, the deterministic mechanism implements curved ray paths in linear gradient background media, and the stochastic mechanism implements pseudo-random preferential scattering based on a formulation that computes mean-free-path and scattering cross sections from a characterization of material heterogeneity in which fluctuation of elastic properties are assumed to follow a von Kármán spectrum of scale lengths. Between cells, material properties may be continuous or discontinuous, in which case reflections and refractions may occur. At the model’s surface, virtual seismometers may be emplaced to record signal channels representative of seismogram envelopes. After introducing the code and theory of operations, two chapters describing initial experiments with the code are presented. In Chapter 4, we present experiments in layered Earth models exploring the effects of variations in the heterogeneity spectrum and their effect on seismic coda generation. In Chapter 5 we present experiments in which crust structure is subject to localized thinning or thickening, simulating graben structures or mountain structures, and analyze their contribution to blockage of the Lg and Pg seismic phases at distant seismometers. We conclude with a roadmap describing the future continued development of Radiative3D.
... Experimental verifications of this assumption are few. From a local array above the Tonga subduction zone, Koper et al. (1999) estimated d ln V S /d ln V P to be in the range 1.1-1.5 at scale lengths on the order of 10-100 km. In the lower mantle at larger scale lengths (1000 km), d ln V S /d ln V P has been estimated to be >2.5 (Romanowicz 2001). ...
Radiative transport modeling can combine the effects of both large-scale (deterministic) and the small scale (statistical) structure on the coda envelopes of high frequency regional seismograms. We describe a computer code to implement radiative transport modeling that propagates packets of seismic body wave energy along ray paths through large-scale deterministic 3-D structure, including the effects of velocity gradients, intrinsic attenuation, source radiation pattern, and multiple scattering by layer boundaries and small scale heterogeneities specified by a heterogeneity spectrum. The spatial distribution of these energy packets can be displayed as time snapshots to aid in the understanding of regional phase propagation or displayed as a coda envelope by summing at receiver bins. These techniques are applied to earthquakes and explosions recorded in the Lop Nor, China region to model observed narrow band passed seismic codas in the 1 to 4 Hz band. We predict that source discriminants in this region based on P/Lg amplitude ratios will best separate earthquake and explosion populations at frequencies 2 Hz and higher.
... The presence of this layer (V P between 6.9 km/s and 8.2 km/s, V S < 4.35 km/s; density between 3.0 g/cm 3 and 3.3 g/cm 3 ) is not limited nor directly bound to subduction areas. The mantle wedge (metasomatized LID) is the portion of mantle material found between a subducting slab and the overriding lithosphere where plastic deformation and magmatism occur (Abers, 2005;Abers et al., 2006;Koper, Wiens, Dorman, Hildebrand, & Webb, 1999;Syracuse & Abers, 2006); outside of this geodynamic setting, mantle material with similar physical behavior has been simply described as "soft mantle". In this region, according to Van Keken (2003), hydrothermal uprising flow causes partial melting of subducting slab material due to dehydration. ...
... Only shallower than 50-75 km does the low-velocity zone steepen upward as hot material beneath a volcanic arc. In the low-velocity sector, Vs falls off faster than does Vp (Vp/Vs is high) and attenuation of S waves is markedly greater than of P waves: the sector is near solidus temperature, and is 400-600 K hotter than nearby slabs at the same depths [Koper et al., 1999]. Seismicity indicates the slab to be decoupled from, and to shear against, the accretionary wedge and the bottom of the overriding plate, but suggests no decoupling between slab and cool mantle beneath the hot medial zone. ...
... The presence of this layer (V P between 6.9 km/s and 8.2 km/s, V S < 4.35 km/s; density between 3.0 g/cm 3 and 3.3 g/cm 3 ) is not limited nor directly bound to subduction areas. The mantle wedge (metasomatized LID) is the portion of mantle material found between a subducting slab and the overriding lithosphere where plastic deformation and magmatism occur (Abers, 2005;Abers et al., 2006;Koper, Wiens, Dorman, Hildebrand, & Webb, 1999;Syracuse & Abers, 2006); outside of this geodynamic setting, mantle material with similar physical behavior has been simply described as "soft mantle". In this region, according to Van Keken (2003), hydrothermal uprising flow causes partial melting of subducting slab material due to dehydration. ...
The mechanisms driving plate motion and the Earth's geodynamics are still not entirely clarified. Lithospheric volumes recycled at subduction zones or emerging at rift zones testify mantle convection. The cooling of the planet and the related density gradients are invoked to explain mantle convection either driven from the hot interior or from the cooler outer boundary layer. In this paper we summarize a number of evidence supporting generalized asymmetries along the plate boundaries that point to a polarization of plate tectonics. W-directed slabs provide two to three times larger volumes to the mantle with respect to the opposite E- or NE-directed subduction zones. W-directed slabs are deeper and steeper, usually characterized by down-dip compression. Moreover, they show a shallow decollement and low elevated accretionary prism, a steep regional monocline with a deep trench or foredeep, a backarc basin with high heat flow and positive gravity anomaly. Conversely directed subduction zones show antithetic signatures and no similar backarc basin. Rift zones also show an asymmetry, e.g., faster Vs in the western lithosphere and a slightly deeper bathymetry with respect to the eastern flank. These evidences can be linked to the westward drift of the lithosphere relative to the underlying mantle and may explain the differences among subduction and rift zones as a function of their geographic polarity with respect to the "tectonic equator." Therefore also mantle convection and plate motion should be polarized. All this supports a general tuning of the Earth's geodynamics and mantle convection by astronomical forces.
A current issue on the settlement of the Americas refers to the lack of morphological affinities between early Holocene human remains (Palaeoamericans) and modern Amerindian groups, as well as the degree of contribution of the former to the gene pool of the latter. A different origin for Palaeoamericans and Amerindians is invoked to explain such a phenomenon. Under this hypothesis, the origin of Palaeoamericans must be traced back to a common ancestor for Palaeoamericans and Australians, which departed from somewhere in southern Asia and arrived in the Australian continent and the Americas around 40,000 and 12,000 years before present, respectively. Most modern Amerindians are believed to be part of a second, morphologically differentiated migration. Here we present evidence of a modern Amerindian group from the Baja California Peninsula in Mexico, showing clearer affinities with Palaeoamerican remains than with modern Amerindians. Climatic changes during the Middle Holocene probably generated the conditions for isolation from the continent, restricting the gene flow of the original group with northern populations, which resulted in the temporal continuity of the Palaeoamerican morphological pattern to the present.
The travel times of S waves from 20 earthquakes to stations in North America in the distance range 28° to 82° have been studied. The deviations from J-B times were analyzed into station, source and distance components using the least-squares time-term approach of Cleary and Hales. Station anomalies had a range of about eight seconds, as compared to three seconds for the P anomalies, and are believed to be caused largely by variations in the upper mantle velocity distribution. S residuals, like the P residuals, were generally positive in the western United States, and negative in the central and eastern United States. P and S residuals at the same station correlated with a coefficient of 0.75, the slope of the regression of S anomaly on P anomaly being 3.72. Corrections to J-B times for S were of the order of the standard errors of the determinations. Within the distance range of 28° to 82° large changes of the S travel times, such as were required by the lower mantle velocities proposed by MacDonald and Ness (1961), are not permitted by the present data. The analysis was checked by carrying out a univariate analysis of variance of the same data.
Relative S-wave station residuals have been determined over the Canadian long-period seismograph network and have been augmented by S residuals to the south into the United States taken from Poupinet (1977). The thick craton of the Canadian Shield is outlined by the S-residual contours with an inferred rapid thinning of the craton to the south except for a narrow central extension into Kentucky and Missouri. A comparison of these S residuals with P-wave residuals of Buchbinder and Poupinet (1977) reveals a low-velocity region in the west, a high-velocity zone centrally, and an area in the east where the S arrivals are late while the P arrivals are early.
