Thorne Lay is Distinguished Professor of Earth and Planetary Sciences at the University of California Santa Cruz, where he has been located since 1990. Previously he was a faculty member at the University of Michigan, after receiving his Ph.D. in Geophysics from the California Institute of Technology in 1983. His research area is seismology, and includes studies of large earthquake ruptures, internal structure of the Earth, and seismic monitoring of nuclear testing treaties. He is author or co-author of 500 research publications including 2 textbooks, 1 monograph, 1 edited book, 4 major reports, 361 publications in refereed books and professional journals, and 131 technical reports, book reviews, news items, memorials and conference proceedings. In addition, he is author or co-author on 423+ published abstracts for meeting presentations. He is elected Fellow of the National Academy of Sciences, the American Academy of Arts and Sciences, and the American Association for the Advancement of Sciences. He received the Gold Medal of the Royal Astronomical Society, the Harry Fielding Reid Medal of the Seismological Society of America and the Inge Lehmann Medal and Macelwane Medal of the American Geophysical Union. He is an Honorary Professor of Xi’an Jiaotong University and recipient of a Star on the Mountain Award from the City of El Paso. Supervisors: Donald V. Helmberger and Hiroo Kanamori
Tsunami observations play an important role in resolving offshore earthquake slip distributions. ... more Tsunami observations play an important role in resolving offshore earthquake slip distributions. Nondispersive shallow‐water models are often used with a static initial sea surface pulse derived from seafloor deformation in computation of tsunami Green's functions. We compare this conventional approach with more advanced techniques based on a dispersive model with a static initial sea surface pulse and with the surface waves generated from kinematic seafloor deformation. These three sets of tsunami Green's functions are implemented in finite‐fault inversions with and without seismic and geodetic data for the 2010 Mentawai Mw 7.8 tsunami earthquake. Seafloor excitation and wave dispersion produce more spread‐out waveforms in the Green's functions leading to larger slip with more compact distribution through the inversions. The fit to the recorded tsunami and the deduced seismic moment, which reflects the displaced water volume, are relatively insensitive to the approach u...
Subduction zones play critical roles in the recycling of oceanic lithosphere and the generation o... more Subduction zones play critical roles in the recycling of oceanic lithosphere and the generation ofcontinental crust. Seismic imaging can reveal structures associated with key dynamic processes occurring in the upper-mantle wedge above the sinking oceanic slab. Three-dimensional images of reflecting interfaces throughout the upper-mantle wedge above the subducting Tonga slab were obtained by migration of teleseismic recordings of underside P- and S-wave reflections. Laterally continuous weak reflectors with tens of kilometers of topography were detected at depths near 90, 125, 200, 250, 300, 330, 390, 410, and 450 kilometers. P- and S-wave impedances decreased at the
330-kilometer and 450-kilometer reflectors, and S-wave impedance decreased near 200 kilometers in the vicinity of the slab and near 390 kilometers, just above the global 410-kilometer increase. The pervasive seismic reflectivity results from phase transitions and compositional zonation associated with extensive metasomatism involving slab-derived fluids rising through the wedge.
ABSTRACT Applications of teleseismic P-wave back-projection to image gross characteristics of lar... more ABSTRACT Applications of teleseismic P-wave back-projection to image gross characteristics of large earthquake finite-source ruptures have been enabled by ready availability of large digital data sets. Imaging with short-period data from dense arrays or broadband data from global networks can place constraints on rupture attributes that otherwise have to be treated parametrically in conventional modeling and inversion procedures. Back-projection imaging may constrain choice of fault plane and rupture direction, velocity, duration and length for large (M>~8.0) earthquakes, and can robustly locate early aftershocks embedded in mainshock surface waves. Back-projection methods seek locations of coherent energy release from the source region, ideally associated with down-going P wave energy. For shallow events, depth phase arrivals can produce artifacts in back-projection images that appear as secondary or even prominent features with incorrect apparent source locations and times, and such effects need to be recognized. We apply broadband P-wave back-projection imaging to the 29 September 2009 Samoa (Mw8.2) and 30 September 2009 Sumatra (Mw7.6) earthquakes using data from globally distributed broadband stations and compare results to back-projections of synthetic seismograms from finite-source models for these events to evaluate the artifacts from depth phases. Back-projection images for the great normal-faulting Samoa event feature two prominent bright spots, which could be interpreted to correspond to two distinct slip patches, one near the epicenter in the outer trench slope and the other approximately 80 km to the west near the plate boundary megathrust where many aftershocks occurred. This interpretation is at odds with finite-fault modeling results, which indicate a predominantly bilateral rupture in the NW-SE direction on a steeply dipping trench slope fault, with rupture extending about 60 km in each direction. Back-projections of data and synthetic seismograms from the finite-fault modeling with common source-receiver geometries are nearly identical, with both having the two prominent bright regions as well as many secondary features. The prominent feature to the west is an artifact resulting from constructive interference of azimuthally varying depth phases; this coincidentally projects in the same region as many aftershocks. Similar analysis for the Sumatra event yields back-projections of data and synthetics for a finite-fault model that closely match and indicate a very compact source with virtually all of the coherent radiation located within 20km of the epicenter. Weak features in the back-projections with arrival times consistent with pP and sP are visible offset from the epicenter. Back-projection methods can be useful for constraining aspects of large earthquake rupture processes, particularly if the variation of waveforms across the imaging network is small, but it is always important to assess what features of the back-projections are artifacts from the path geometry or depth phases to avoid misinterpreting the images, particularly when using globally distributed stations rather than large-aperture arrays.
ABSTRACT We use thousands of seismograms from South and Central American earthquakes recorded by ... more ABSTRACT We use thousands of seismograms from South and Central American earthquakes recorded by western North American seismic networks to image the lowermost mantle beneath Central America using a 3D Kirchhoff migration scheme. P wave studies of the deep mantle often rely on some form of stacking of many records in order to enhance the signal-to-noise ratio of weak phases generated by deep structure, such as reflections off of the D" discontinuity. These methods, however, often assume one-dimensional structure, which is at odds with the evidence for significant heterogeneity. Kirchhoff migration is a three-dimensional stacking method that allows interactions with structure off of the source-receiver plane, thus imaging a much larger volume and avoiding false projections of scattered arrivals onto specular reflectors. The D" discontinuity beneath Central America has been readily observed in S wave studies and may be the result of the shear wave velocity increase associated with the recently discovered perovskite to post-perovskite phase transition. This phase transition is expected to have weaker effects on P wave velocities than on S wave velocities and the sharpness of this transition is unknown. We observe structures consistent with a discontinuity about 200 km above the core-mantle boundary (CMB). The fact that this is seen at all in short period data suggests that its boundary must be less than 10 to 20 km thick, while observation with broadband data exclude the possibility of it being a thin layer or lamella. Whether the discontinuity is co-located for both P and S waves is difficult to resolve given uncertainties in the long-scale velocity heterogeneity. In addition, both broadband and short period P wave data sets reveal a sharp out-of-plane scatterer, which may be located close to the CMB. The short period data also indicate reflectivity about 400 km above the CMB, well above the aforementioned D" discontinuity, and similar reflectivity is observed under the Central Pacific. This feature appears to be more consistent with a discontinuity than a scatterer, is hinted at in the broadband data set and is not observed with S waves.
For nearly 25 years, seismic velocity heterogeneity in Earth's mantle has been mapped by tom... more For nearly 25 years, seismic velocity heterogeneity in Earth's mantle has been mapped by tomographic methods, and used to infer global mantle dynamics [e.g., Dziewonski, 1984]. Two nearly antipodal large low shear velocity provinces (LLSVPs) in the lowermost mantle beneath the Pacific Ocean and Africa are surrounded by higher than average velocities. The observation that LLSVPs tend to underlie hotspots and the high velocities tend to underlie subduction has been long suggested to support whole mantle convection. In more recent years, regional high-resolution waveform studies of structure throughout the mantle depend on the global models as a long wavelength background framework. This presentation will highlight recent regional work in the lowermost mantle, as well as in the mantle transition zone, which argue for a mantle with diverse thermal and chemical complexities. In the deepest mantle, ultra-low velocity zones are found in isolated patches, and may geographically concentrate near LLSVP margins; especially given the seismic evidence for a chemically distinct origin to LLSVPs and geodynamic evidence for LLSVP margins being the hottest mantle rock on Earth. Shear velocity discontinuities in the lowermost 200-300 km of the mantle (the so-called D" discontinuity) are well-established features, with recent work documenting topography and suggesting the discontinuity originates from the post-perovskite phase transition. In the mantle transition zone, a richness in information is emerging about the topography on the 410 and 660 km phase boundaries, as well as fine scale additional layering. These findings are discussed in reference to the global models, which define the background thermodynamical setting for the high resolution findings. The high resolution studies compliment inferences born from tomographically derived long wavelength volumetric heterogeneity, in that they essentially depict 'roughness' which unveils and discriminates between regional chemical/mineralogical versus thermal heterogeneity. That is, the two together infer geology and hence and provide a strong conduit to research in other geodisciplines.
ABSTRACT An abrupt 1-2.5% S-wave velocity increase has been observed in the lowermost mantle bene... more ABSTRACT An abrupt 1-2.5% S-wave velocity increase has been observed in the lowermost mantle beneath the Cocos plate in several studies. This is commonly attributed to the perovskite to post-perovskite phase transition. This phase transition is expected to have much weaker effects on P-wave velocities than on S-wave velocities and the depth range of the transition is expected to depend on Al and Fe content of the perovskite. We image lowermost mantle P-wave reflectivity beneath the Cocos plate using 1D stacking and 3D Kirchhoff migration. Our carefully processed data set comprises 8000 seismograms from deep South American earthquakes recorded by broadband and short-period seismic networks in western North America. Stacked P-wave source wavelets are deconvolved from the data for each event, allowing band-pass filtered signals to be combined for many events. Events are discarded if individual event-stacks do not show an impulsive PcP arrival with significant signal-to- noise ratio. Depth shifts are applied to each event to align PcP arrivals in the combined stacks. These shifts increase systematically from south to north. We observe a widespread weak P-wave reflector about 320 km above the CMB modeled well by a P-wave velocity (Vp) change of -0.2 to -0.4%, depending on the thickness of the velocity change. The depth of this relatively flat reflector is tightly constrained and is a few tens of km shallower than the local S-wave reflector, which may have regional topography of up to 100 km. We model a clear feature in the P-wave reflectivity with a change in dVp/dZ about 180 km above the CMB, accompanied by a sharp 0.2% increase in Vp. Different narrow band filters up to 2 Hz and forward modeling of double-array stacks show that this small velocity increase must occur over less than 10 km in depth. This also does not directly correspond to any significant feature in the S-wave velocity structure. The high signal-to-noise ratio of our locally binned data stacks allows us to preclude the existence of any other abrupt Vp change stronger than ±0.1 to ±0.4%. These upper bounds are influenced by the source-receiver geometry of the data and the thickness of the velocity change. A strong, positive velocity contrast, out-of-plane scatterer is observed in the migrations, and appears to originate shallower than 200 km above the CMB. Finally, we show that the PREM values for P-wave attenuation in the lowermost mantle are too low, and need to be at least three times greater locally to match our broadband observations.
ABSTRACT A large, high quality P-wave data set comprising short-period and broadband signals samp... more ABSTRACT A large, high quality P-wave data set comprising short-period and broadband signals sampling four regions in the lowermost mantle beneath the Cocos plate, Mexico, the north Pacific, and the central Pacific is analyzed using regional one-dimensional double-array stacking and synthetic modeling. A data-screening criterion retains only events with stable PcP energy in the final data stacks used for modeling and interpretation. This significantly improves the signal in the stacks compared to stacks that uses all observations and allows tight bounds to be placed on P-wave velocity structure above the core-mantle boundary (CMB). The PcP reflections under the Cocos plate are well-modeled without any ultra-low velocity zone from 5 - 20°N. The data stacks for paths with PcP reflection points below the eastern equatorial Pacific, Mexico and the Gulf of Mexico have PcP images that are well-matched with the simple IASP91 structure. Beneath the north Pacific near the Aleutian Islands, PcP arrivals are compatible with LVZ (dlnVp of about 0 to -3%) only a few km thick. Data sampling a 200 X 200 km patch of the lowermost mantle beneath the central Pacific confirm the presence of a low velocity zone just above the CMB about 15 km thick, with very little topography, dlnVp of around -3 to -4% and dlnVs that varies from -4 to -8%. The data are insensitive to any density contrasts. The velocity drops observed in this region are small compared to some ULVZ observations beneath other parts of the Pacific, which have dlnVp and dlnVs of -10 and -30%, respectively. P- and S-wave data in this region also indicate laterally varying discontinuities shallower in D". The data sample the margin of the central Pacific's large low shear velocity province and our results may reflect a gradual change in Fe or Al content. Our results indicate localized occurrence of detectable LVZ structures rather than ubiquitous ULVZ structure and show structural differences between circum-Pacific regions and the large low shear velocity province under the central Pacific.
ABSTRACT Using data from South American earthquakes recorded by seismic stations in Europe and Af... more ABSTRACT Using data from South American earthquakes recorded by seismic stations in Europe and Africa, and using Tonga-Fiji events recorded in South America with reverse paths from South America events to Tonga stations, we investigate properties of D" beneath the Atlantic and the southern Pacific. By measuring splits between the SV and SH components of ScS and S or Sdiff phases, we infer D" anisotropy magnitude. Broadband data, along with some digitized analog WWSSN data, were analyzed for deep focus earthquakes (> 100 km). All data are corrected for predictions of upper mantle anisotropy, and we assume that D" has vertical transverse isotropy (VTI) in our initial analysis. Beneath the Atlantic we find SV-SH splits averaging 0.8 s (minimum = -0.9 s; maximum = 3.6 s) in the northern regions where path coverage is densest and averaging 0.3 s (minimum = -1.6 s; maximum =1.7 s) in the south Atlantic. Approximate strength of inferred anisotropy averages around 0.4% in the northern regions and 0.2% in the south. The geometry of anisotropy is difficult to constrain, so the data are compared to synthetics to narrow the possible models responsible for the splitting. Differential travel times between S(Sdiff)(SH)-SKS(SV) and ScS(SH)-S(SH) are used to deduce isotropic heterogeneity in D", with the times corrected for aspherical mantle structure above D". We calculate differential travel time residuals with respect to PREM averaging -0.04 s for S-SKS and 0.25 s for ScS-S. Heterogeneity inferred from S-SKS differential times was averaged to be 0.62% and for and ScS-S was 1.2% which agrees well with high-resolution tomographic model predictions. All data are inspected for Scd arrivals, the seismic phase attributed to a reflection off a high velocity layer in D", to assess the possible presence of a D" discontinuity. In contrast to high velocity regions of the circum-Pacific and the low velocity central Pacific, our results show relatively reduced magnitudes of anisotropy in a region characterized as a transition from high-to-low isotropic heterogeneity (west-to-east). We relate these patterns to deep mantle dynamics and discuss implications of our observations for boundary layer processes.
The discovery of the post-perovskite phase transition presents a new context for interpreting sei... more The discovery of the post-perovskite phase transition presents a new context for interpreting seismological observations of deep mantle structure. As is the case for seismic velocity discontinuities detected in the upper mantle transition zone, interpretation as an expression of a phase change provides an opportunity to infer thermal and dynamical effects that are otherwise not directly sensed by seismology. Extensive
Global Love and Rayleigh wave phase velocity variations from 85 to 250 s are determined using abo... more Global Love and Rayleigh wave phase velocity variations from 85 to 250 s are determined using about 30,000 seismograms from earthquakes with M ≥ 6.0. The phases used are G1, R1, G2, and R2 arrivals. The data were obtained from the Chinese Digital Seismic Network, GEOSCOPE, and Global Seismic Network archives for the time period January 1980 to June 1992. All seismograms underwent careful quality control in the time and frequency domains. We consider the inversion damping and model parameterization issues, examining the data statistics and assuming that errors in data and model spaces are Gaussian random distributions with zero mean. It is not stable to invert for a model with resolution higher than about spherical harmonic degree, l = 22, given the path sampling and scatter of our data set, and the low frequency data only warrant expansion up to about l = 12. We find that a block model parameterization, with its local basis functions, has fewer artifacts than a high order spherical harmonic inversion which must be heavily damped. However, spherical harmonic functions provide intrinsically superior recovery of the long‐wavelength structure. Our final models involve a hybrid parameterization, in which an initial iteration retrieves the low order spherical harmonic components that are used as an aspherical reference model for performing a final block model inversion. The optimal block size decreases with frequency. This approach exploits the separate attributes of global and local basis functions for resolving different components of the heterogeneity, allows for variable resolution for different frequencies, and provides a natural framework for embedding localized high resolution regional inversions into the global model. We perform inversions both with and without corrections for shallow crustal structure. At the shorter periods the phase velocity maps strongly reflect the surface tectonics, correlating with continental shields, mid‐ocean ridges, and tectonically active areas. The tectonic correlation diminishes gradually with period.
Tsunami observations play an important role in resolving offshore earthquake slip distributions. ... more Tsunami observations play an important role in resolving offshore earthquake slip distributions. Nondispersive shallow‐water models are often used with a static initial sea surface pulse derived from seafloor deformation in computation of tsunami Green's functions. We compare this conventional approach with more advanced techniques based on a dispersive model with a static initial sea surface pulse and with the surface waves generated from kinematic seafloor deformation. These three sets of tsunami Green's functions are implemented in finite‐fault inversions with and without seismic and geodetic data for the 2010 Mentawai Mw 7.8 tsunami earthquake. Seafloor excitation and wave dispersion produce more spread‐out waveforms in the Green's functions leading to larger slip with more compact distribution through the inversions. The fit to the recorded tsunami and the deduced seismic moment, which reflects the displaced water volume, are relatively insensitive to the approach u...
Subduction zones play critical roles in the recycling of oceanic lithosphere and the generation o... more Subduction zones play critical roles in the recycling of oceanic lithosphere and the generation ofcontinental crust. Seismic imaging can reveal structures associated with key dynamic processes occurring in the upper-mantle wedge above the sinking oceanic slab. Three-dimensional images of reflecting interfaces throughout the upper-mantle wedge above the subducting Tonga slab were obtained by migration of teleseismic recordings of underside P- and S-wave reflections. Laterally continuous weak reflectors with tens of kilometers of topography were detected at depths near 90, 125, 200, 250, 300, 330, 390, 410, and 450 kilometers. P- and S-wave impedances decreased at the
330-kilometer and 450-kilometer reflectors, and S-wave impedance decreased near 200 kilometers in the vicinity of the slab and near 390 kilometers, just above the global 410-kilometer increase. The pervasive seismic reflectivity results from phase transitions and compositional zonation associated with extensive metasomatism involving slab-derived fluids rising through the wedge.
ABSTRACT Applications of teleseismic P-wave back-projection to image gross characteristics of lar... more ABSTRACT Applications of teleseismic P-wave back-projection to image gross characteristics of large earthquake finite-source ruptures have been enabled by ready availability of large digital data sets. Imaging with short-period data from dense arrays or broadband data from global networks can place constraints on rupture attributes that otherwise have to be treated parametrically in conventional modeling and inversion procedures. Back-projection imaging may constrain choice of fault plane and rupture direction, velocity, duration and length for large (M>~8.0) earthquakes, and can robustly locate early aftershocks embedded in mainshock surface waves. Back-projection methods seek locations of coherent energy release from the source region, ideally associated with down-going P wave energy. For shallow events, depth phase arrivals can produce artifacts in back-projection images that appear as secondary or even prominent features with incorrect apparent source locations and times, and such effects need to be recognized. We apply broadband P-wave back-projection imaging to the 29 September 2009 Samoa (Mw8.2) and 30 September 2009 Sumatra (Mw7.6) earthquakes using data from globally distributed broadband stations and compare results to back-projections of synthetic seismograms from finite-source models for these events to evaluate the artifacts from depth phases. Back-projection images for the great normal-faulting Samoa event feature two prominent bright spots, which could be interpreted to correspond to two distinct slip patches, one near the epicenter in the outer trench slope and the other approximately 80 km to the west near the plate boundary megathrust where many aftershocks occurred. This interpretation is at odds with finite-fault modeling results, which indicate a predominantly bilateral rupture in the NW-SE direction on a steeply dipping trench slope fault, with rupture extending about 60 km in each direction. Back-projections of data and synthetic seismograms from the finite-fault modeling with common source-receiver geometries are nearly identical, with both having the two prominent bright regions as well as many secondary features. The prominent feature to the west is an artifact resulting from constructive interference of azimuthally varying depth phases; this coincidentally projects in the same region as many aftershocks. Similar analysis for the Sumatra event yields back-projections of data and synthetics for a finite-fault model that closely match and indicate a very compact source with virtually all of the coherent radiation located within 20km of the epicenter. Weak features in the back-projections with arrival times consistent with pP and sP are visible offset from the epicenter. Back-projection methods can be useful for constraining aspects of large earthquake rupture processes, particularly if the variation of waveforms across the imaging network is small, but it is always important to assess what features of the back-projections are artifacts from the path geometry or depth phases to avoid misinterpreting the images, particularly when using globally distributed stations rather than large-aperture arrays.
ABSTRACT We use thousands of seismograms from South and Central American earthquakes recorded by ... more ABSTRACT We use thousands of seismograms from South and Central American earthquakes recorded by western North American seismic networks to image the lowermost mantle beneath Central America using a 3D Kirchhoff migration scheme. P wave studies of the deep mantle often rely on some form of stacking of many records in order to enhance the signal-to-noise ratio of weak phases generated by deep structure, such as reflections off of the D" discontinuity. These methods, however, often assume one-dimensional structure, which is at odds with the evidence for significant heterogeneity. Kirchhoff migration is a three-dimensional stacking method that allows interactions with structure off of the source-receiver plane, thus imaging a much larger volume and avoiding false projections of scattered arrivals onto specular reflectors. The D" discontinuity beneath Central America has been readily observed in S wave studies and may be the result of the shear wave velocity increase associated with the recently discovered perovskite to post-perovskite phase transition. This phase transition is expected to have weaker effects on P wave velocities than on S wave velocities and the sharpness of this transition is unknown. We observe structures consistent with a discontinuity about 200 km above the core-mantle boundary (CMB). The fact that this is seen at all in short period data suggests that its boundary must be less than 10 to 20 km thick, while observation with broadband data exclude the possibility of it being a thin layer or lamella. Whether the discontinuity is co-located for both P and S waves is difficult to resolve given uncertainties in the long-scale velocity heterogeneity. In addition, both broadband and short period P wave data sets reveal a sharp out-of-plane scatterer, which may be located close to the CMB. The short period data also indicate reflectivity about 400 km above the CMB, well above the aforementioned D" discontinuity, and similar reflectivity is observed under the Central Pacific. This feature appears to be more consistent with a discontinuity than a scatterer, is hinted at in the broadband data set and is not observed with S waves.
For nearly 25 years, seismic velocity heterogeneity in Earth's mantle has been mapped by tom... more For nearly 25 years, seismic velocity heterogeneity in Earth's mantle has been mapped by tomographic methods, and used to infer global mantle dynamics [e.g., Dziewonski, 1984]. Two nearly antipodal large low shear velocity provinces (LLSVPs) in the lowermost mantle beneath the Pacific Ocean and Africa are surrounded by higher than average velocities. The observation that LLSVPs tend to underlie hotspots and the high velocities tend to underlie subduction has been long suggested to support whole mantle convection. In more recent years, regional high-resolution waveform studies of structure throughout the mantle depend on the global models as a long wavelength background framework. This presentation will highlight recent regional work in the lowermost mantle, as well as in the mantle transition zone, which argue for a mantle with diverse thermal and chemical complexities. In the deepest mantle, ultra-low velocity zones are found in isolated patches, and may geographically concentrate near LLSVP margins; especially given the seismic evidence for a chemically distinct origin to LLSVPs and geodynamic evidence for LLSVP margins being the hottest mantle rock on Earth. Shear velocity discontinuities in the lowermost 200-300 km of the mantle (the so-called D" discontinuity) are well-established features, with recent work documenting topography and suggesting the discontinuity originates from the post-perovskite phase transition. In the mantle transition zone, a richness in information is emerging about the topography on the 410 and 660 km phase boundaries, as well as fine scale additional layering. These findings are discussed in reference to the global models, which define the background thermodynamical setting for the high resolution findings. The high resolution studies compliment inferences born from tomographically derived long wavelength volumetric heterogeneity, in that they essentially depict 'roughness' which unveils and discriminates between regional chemical/mineralogical versus thermal heterogeneity. That is, the two together infer geology and hence and provide a strong conduit to research in other geodisciplines.
ABSTRACT An abrupt 1-2.5% S-wave velocity increase has been observed in the lowermost mantle bene... more ABSTRACT An abrupt 1-2.5% S-wave velocity increase has been observed in the lowermost mantle beneath the Cocos plate in several studies. This is commonly attributed to the perovskite to post-perovskite phase transition. This phase transition is expected to have much weaker effects on P-wave velocities than on S-wave velocities and the depth range of the transition is expected to depend on Al and Fe content of the perovskite. We image lowermost mantle P-wave reflectivity beneath the Cocos plate using 1D stacking and 3D Kirchhoff migration. Our carefully processed data set comprises 8000 seismograms from deep South American earthquakes recorded by broadband and short-period seismic networks in western North America. Stacked P-wave source wavelets are deconvolved from the data for each event, allowing band-pass filtered signals to be combined for many events. Events are discarded if individual event-stacks do not show an impulsive PcP arrival with significant signal-to- noise ratio. Depth shifts are applied to each event to align PcP arrivals in the combined stacks. These shifts increase systematically from south to north. We observe a widespread weak P-wave reflector about 320 km above the CMB modeled well by a P-wave velocity (Vp) change of -0.2 to -0.4%, depending on the thickness of the velocity change. The depth of this relatively flat reflector is tightly constrained and is a few tens of km shallower than the local S-wave reflector, which may have regional topography of up to 100 km. We model a clear feature in the P-wave reflectivity with a change in dVp/dZ about 180 km above the CMB, accompanied by a sharp 0.2% increase in Vp. Different narrow band filters up to 2 Hz and forward modeling of double-array stacks show that this small velocity increase must occur over less than 10 km in depth. This also does not directly correspond to any significant feature in the S-wave velocity structure. The high signal-to-noise ratio of our locally binned data stacks allows us to preclude the existence of any other abrupt Vp change stronger than ±0.1 to ±0.4%. These upper bounds are influenced by the source-receiver geometry of the data and the thickness of the velocity change. A strong, positive velocity contrast, out-of-plane scatterer is observed in the migrations, and appears to originate shallower than 200 km above the CMB. Finally, we show that the PREM values for P-wave attenuation in the lowermost mantle are too low, and need to be at least three times greater locally to match our broadband observations.
ABSTRACT A large, high quality P-wave data set comprising short-period and broadband signals samp... more ABSTRACT A large, high quality P-wave data set comprising short-period and broadband signals sampling four regions in the lowermost mantle beneath the Cocos plate, Mexico, the north Pacific, and the central Pacific is analyzed using regional one-dimensional double-array stacking and synthetic modeling. A data-screening criterion retains only events with stable PcP energy in the final data stacks used for modeling and interpretation. This significantly improves the signal in the stacks compared to stacks that uses all observations and allows tight bounds to be placed on P-wave velocity structure above the core-mantle boundary (CMB). The PcP reflections under the Cocos plate are well-modeled without any ultra-low velocity zone from 5 - 20°N. The data stacks for paths with PcP reflection points below the eastern equatorial Pacific, Mexico and the Gulf of Mexico have PcP images that are well-matched with the simple IASP91 structure. Beneath the north Pacific near the Aleutian Islands, PcP arrivals are compatible with LVZ (dlnVp of about 0 to -3%) only a few km thick. Data sampling a 200 X 200 km patch of the lowermost mantle beneath the central Pacific confirm the presence of a low velocity zone just above the CMB about 15 km thick, with very little topography, dlnVp of around -3 to -4% and dlnVs that varies from -4 to -8%. The data are insensitive to any density contrasts. The velocity drops observed in this region are small compared to some ULVZ observations beneath other parts of the Pacific, which have dlnVp and dlnVs of -10 and -30%, respectively. P- and S-wave data in this region also indicate laterally varying discontinuities shallower in D". The data sample the margin of the central Pacific's large low shear velocity province and our results may reflect a gradual change in Fe or Al content. Our results indicate localized occurrence of detectable LVZ structures rather than ubiquitous ULVZ structure and show structural differences between circum-Pacific regions and the large low shear velocity province under the central Pacific.
ABSTRACT Using data from South American earthquakes recorded by seismic stations in Europe and Af... more ABSTRACT Using data from South American earthquakes recorded by seismic stations in Europe and Africa, and using Tonga-Fiji events recorded in South America with reverse paths from South America events to Tonga stations, we investigate properties of D" beneath the Atlantic and the southern Pacific. By measuring splits between the SV and SH components of ScS and S or Sdiff phases, we infer D" anisotropy magnitude. Broadband data, along with some digitized analog WWSSN data, were analyzed for deep focus earthquakes (> 100 km). All data are corrected for predictions of upper mantle anisotropy, and we assume that D" has vertical transverse isotropy (VTI) in our initial analysis. Beneath the Atlantic we find SV-SH splits averaging 0.8 s (minimum = -0.9 s; maximum = 3.6 s) in the northern regions where path coverage is densest and averaging 0.3 s (minimum = -1.6 s; maximum =1.7 s) in the south Atlantic. Approximate strength of inferred anisotropy averages around 0.4% in the northern regions and 0.2% in the south. The geometry of anisotropy is difficult to constrain, so the data are compared to synthetics to narrow the possible models responsible for the splitting. Differential travel times between S(Sdiff)(SH)-SKS(SV) and ScS(SH)-S(SH) are used to deduce isotropic heterogeneity in D", with the times corrected for aspherical mantle structure above D". We calculate differential travel time residuals with respect to PREM averaging -0.04 s for S-SKS and 0.25 s for ScS-S. Heterogeneity inferred from S-SKS differential times was averaged to be 0.62% and for and ScS-S was 1.2% which agrees well with high-resolution tomographic model predictions. All data are inspected for Scd arrivals, the seismic phase attributed to a reflection off a high velocity layer in D", to assess the possible presence of a D" discontinuity. In contrast to high velocity regions of the circum-Pacific and the low velocity central Pacific, our results show relatively reduced magnitudes of anisotropy in a region characterized as a transition from high-to-low isotropic heterogeneity (west-to-east). We relate these patterns to deep mantle dynamics and discuss implications of our observations for boundary layer processes.
The discovery of the post-perovskite phase transition presents a new context for interpreting sei... more The discovery of the post-perovskite phase transition presents a new context for interpreting seismological observations of deep mantle structure. As is the case for seismic velocity discontinuities detected in the upper mantle transition zone, interpretation as an expression of a phase change provides an opportunity to infer thermal and dynamical effects that are otherwise not directly sensed by seismology. Extensive
Global Love and Rayleigh wave phase velocity variations from 85 to 250 s are determined using abo... more Global Love and Rayleigh wave phase velocity variations from 85 to 250 s are determined using about 30,000 seismograms from earthquakes with M ≥ 6.0. The phases used are G1, R1, G2, and R2 arrivals. The data were obtained from the Chinese Digital Seismic Network, GEOSCOPE, and Global Seismic Network archives for the time period January 1980 to June 1992. All seismograms underwent careful quality control in the time and frequency domains. We consider the inversion damping and model parameterization issues, examining the data statistics and assuming that errors in data and model spaces are Gaussian random distributions with zero mean. It is not stable to invert for a model with resolution higher than about spherical harmonic degree, l = 22, given the path sampling and scatter of our data set, and the low frequency data only warrant expansion up to about l = 12. We find that a block model parameterization, with its local basis functions, has fewer artifacts than a high order spherical harmonic inversion which must be heavily damped. However, spherical harmonic functions provide intrinsically superior recovery of the long‐wavelength structure. Our final models involve a hybrid parameterization, in which an initial iteration retrieves the low order spherical harmonic components that are used as an aspherical reference model for performing a final block model inversion. The optimal block size decreases with frequency. This approach exploits the separate attributes of global and local basis functions for resolving different components of the heterogeneity, allows for variable resolution for different frequencies, and provides a natural framework for embedding localized high resolution regional inversions into the global model. We perform inversions both with and without corrections for shallow crustal structure. At the shorter periods the phase velocity maps strongly reflect the surface tectonics, correlating with continental shields, mid‐ocean ridges, and tectonically active areas. The tectonic correlation diminishes gradually with period.
Uploads
330-kilometer and 450-kilometer reflectors, and S-wave impedance decreased near 200 kilometers in the vicinity of the slab and near 390 kilometers, just above the global 410-kilometer increase. The pervasive seismic reflectivity results from phase transitions and compositional zonation associated with extensive metasomatism involving slab-derived fluids rising through the wedge.
330-kilometer and 450-kilometer reflectors, and S-wave impedance decreased near 200 kilometers in the vicinity of the slab and near 390 kilometers, just above the global 410-kilometer increase. The pervasive seismic reflectivity results from phase transitions and compositional zonation associated with extensive metasomatism involving slab-derived fluids rising through the wedge.