Michael S. Thorne

A Post-Perovskite Lens and D'' Heat Flux Beneath the Central Pacific

Temperature gradients in a low-shear-velocity province in the lowermost mantle (D″ region) beneath the central Pacific Ocean were inferred from the observation of a rapid S-wave velocity increase overlying a rapid decrease. These paired seismic discontinuities are attributed to a phase change from perovskite to post-perovskite and then back to perovskite as the temperature increases with depth. Iron enrichment could explain the occurrence of post-perovskite several hundred kilometers above the core-mantle boundary in this warm, chemically distinct province. The double phase-boundary crossing directly constrains the lowermost mantle temperature gradients. Assuming a standard but unconstrained choice of thermal conductivity, the regional core-mantle boundary heat flux (∼85 ± 25 milliwatts per square meter), comparable to the average at Earths surface, was estimated, along with a lower bound on global core-mantle boundary heat flow in the range of 13 ± 4 terawatts. Mapped velocity-contrast variations indicate that the lens of post-perovskite minerals thins and vanishes over 1000 kilometers laterally toward the margin of the chemical distinct region as a result of a ∼500-kelvin temperature increase.

Geometric Spreading of Pn and Sn in a Spherical Earth Model

Geometric spreading of Pn and Sn waves in a spherical Earth model is different than that of classical headwaves and is frequency dependent. The behavior cannot be fully represented by a frequency-independent power-law model, as is com- monly assumed. The lack of an accurate representation of Pn andSn geometric spread- ing in a spherical Earth model impedes our ability to characterize Earth properties including anelasticity. We conduct numerical simulations to quantify Pn and Sn geometric spreading in a spherical Earth model with constant mantle-lid velocities. Based on our simulation results, we present new empirical Pn and Sn geometric- spreading models in the form Gr;f ��� 10 n3� f� =r0�� r0=rn1� flogr0=r�� n2� fand nif �� ni1� logf=f0�� 2 � ni2 logf=f0 �� ni3, where i � 1 ,2 , or 3;r is epicentral distance; f is frequency; r0 � 1 km; and f0 � 1 Hz. We derive values of coefficients nij by fitting the model to computed Pn and Sn amplitudes for a spherical Earth model having a 40-km-thick crust, generic values of P and S velocities, and a constant-ve- locity uppermost mantle. We apply the new spreading model to observed data in Eur- asia to estimate average Pn attenuation, obtaining more reasonable results compared to using a standard power-law model. Our new Pn and Sn geometric-spreading models provide generally applicable reference behavior for spherical Earth models with con- stant uppermost-mantle velocities.

Fine-Scale Ultra-Low Velocity Zone Layering at the Core-Mantle Boundary and Superplumes

Ultra-low velocity layering at the Earth’s core-mantle boundary (CMB) has now been detected using a variety of seismic probes. P- and S-wave velocity reductions of up to 10’s of percent have been mapped in a thin (5–50 km) layer, which commonly underlies reduced seismic shear wave speeds in the overlying few 100 km of the mantle. Ultra-low velocity zones (ULVZ) contain properties consistent with partial melt of rock at the very base of the mantle. Strong evidence now exists for a significant density increase in the layer (∼5–10% greater than reference models), which must be included in dynamical scenarios relating ULVZ partial melt to deep mantle plume genesis. 3-D geodynamical calculations involving an initially uniform dense layer in the lowermost few 100 km of the mantle result in thermo-chemical piles that are geographically well-correlated with seismic tomography low velocities, when past plate motions are imposed as a surface boundary condition. The hottest lower mantle regions underlay edges of the dense thermo-chemical piles. A scenario is put forth where these piles geographically correlate with ultra-low velocity zones, and subsequent mantle plume genesis.

A compositional origin to ultralow-velocity zones

We analyzed vertical component short-period ScP waveforms for 26 earthquakes occurring in the Tonga-Fiji trench recorded at the Alice Springs Array in central Australia. These waveforms show strong precursory and postcursory seismic arrivals consistent with ultralow-velocity zone (ULVZ) layering beneath the Coral Sea. We used the Viterbi sparse spike detection method to measure differential travel times and amplitudes of the postcursor arrival ScSP and the precursor arrival SPcP relative to ScP. We compare our measurements to a database of 340,000 synthetic seismograms finding that these data are best fit by a ULVZ model with an S wave velocity reduction of 24%, a P wave velocity reduction of 23%, a thickness of 8.5 km, and a density increase of 6%. This 1:1 VS:VP velocity decrease is commensurate with a ULVZ compositional origin and is most consistent with highly iron enriched ferropericlase.

Seismic array constraints on the D″ discontinuity beneath Central America

We analyzed 16,150 transverse component seismic recordings from 54 deep-focus earthquakes in the South American and Caribbean regions recorded at broadband stations in North America between 2005 and 2012. We treated subgroups of seismic stations within 3° radius geographical bins as seismic arrays and performed vespagram analysis. We focused on the S, ScS, and Scd arrivals and collected data in the epicentral distance range from 55° to 90°. In particular, we searched for D′′ discontinuity presence in the vespagrams in a 25° by 35° (or 1520 by 2130 km) area beneath Central America. Analysis of these data showed 125 clear Scd observations, 180 Scd observations of lesser quality, and 343 nonobservations. We produced a new map of the discontinuity height beneath Central America. Our map shows an average discontinuity height of 286 ± 6 km (σ =76 km). The region is punctuated by a large topographic high centered at approximately 10°N and 90°W with a maximum height of 380 km. Two smaller topographic highs are located at approximately 4°N and 81°W (discontinuity height of 320 km) and at 4°N and 70°W (height of 315 km). The observation of multiple Scd arrivals collocated with the strongest gradients in inferred topography provides evidence for topographic variation on the discontinuity rather than multiple discontinuities. The regions where the discontinuity has the greatest height can be explained by localized enrichment of mid-ocean ridge basalt from the subducted Farallon slab impinging on the core-mantle boundary.

Anthropogenic sources stimulate resonance of a natural rock bridge

The natural modes of vibration of bedrock landforms, as well as the sources and effects of stimulated resonance remain poorly understood. Here we show that seismic energy created by an induced earthquake and an artificial reservoir has spectral content coincident with the natural modes of vibration of a prominent rock bridge. We measured the resonant frequencies of Rainbow Bridge, Utah using data from two broadband seismometers placed on the span, and identified eight distinct vibrational modes between 1 and 6 Hz. A distant, induced earthquake produced local ground motion rich in 1 Hz energy, stimulating a 20 dB increase in measured power at the bridges fundamental mode. Moreover, we establish that wave action on Lake Powell, an artificial reservoir, generates microseismic energy with peak power ~1 Hz, also exciting resonance of Rainbow Bridge. These anthropogenic sources represent relatively new energy input for the bridge with unknown consequences for structural fatigue.

D″ discontinuity structure beneath the North Atlantic from Scd observations

We analyzed transverse and radial component recordings from the 2010 M6.3 southern Spain earthquake (depth = 620 km) recorded on 370 broadband stations in North America. We grouped these seismograms into subarrays and applied fourth root vespa processing (vespagram analysis) in order to enhance low-amplitude arrivals. These vespagrams show clear Scd arrivals which indicate the existence of the D″ discontinuity beneath the North Atlantic Ocean (45–60°N, 45–55°W). These observations are best fit with a +2–4% velocity increase at the top of the D″ discontinuity at a height above the core-mantle boundary of 304 ± 14 km. We do not observe Scd arrivals at the eastern end of our study region which is consistent with the presence of the easternmost edge of the ancient Farallon plate.

Ambient resonance of Mesa Arch, Canyonlands National Park, Utah

We analyzed the resonance characteristics of a prominent natural arch in Canyonlands National Park, Mesa Arch, as measured from ambient seismic data. Evaluating spectral and polarization attributes, we distinguished the first four resonant frequencies of the arch, 2.9, 6.0, 6.9, and 8.5 Hz, as well as basic properties of the associated mode shapes. We then affirmed experimental data using 3-D numerical modal analysis, providing estimates of material properties and clarifying vibrational mode shapes. Monitoring resonant frequencies over time, we searched for shifts associated with changing environmental conditions and long-term progressive damage. We measured ~3% direct daily variation in resonant frequency associated with changing rock temperature, thermal stress, and stiffening of the rock matrix. Independent tilt data showed similar diurnal cycles associated with thermoelastic stresses and deformation of the arch. We observed no permanent resonant frequency shifts related to irreversible damage of Mesa Arch during our study period.

Use of Seismic Resonance Measurements to Determine the Elastic Modulus of Freestanding Rock Masses

Measuring the elastic modulus of in situ rock masses over scales of tens of meters remains an important challenge in experimental rock mechanics. Here we present a new approach using ambient resonance measurements of freestanding rock landforms to identify vibrational modes, which are then matched with 3D numerical models implementing bulk, globally representative material properties. The result is an experimentally determined, albeit numerically calibrated, estimate of rock mass elastic modulus. We demonstrate the approach at five natural rock arches in southern Utah, each formed in Navajo Sandstone, where we have acquired resonance data and matched experimental resonant modes using 3D numerical modal analysis. Two material properties can be varied to match experimental data: density and modulus. We hold density constant, applying measured or reference values, and solve for elastic modulus using a forward approach. The resolved modulus is representative of the global small-strain dynamic behavior, integrating rock mass heterogeneity over the scale of the feature. The technique works well for freestanding geological landforms that exhibit clear vibrational modes. Errors arise with uncertain mechanical boundary conditions or strong material anisotropy. The resolved modulus values add relevant information describing the variation of elastic properties over scales from lab samples to in situ rock masses.

Estimate of the Rigidity of Eclogite in the Lower Mantle From Waveform Modeling of Broadband S-to-P Wave Conversions

Broadband USArray recordings of the 21 July 2007 western Brazil earthquake (Mw=6.0; depth = 633 km) include high-amplitude signals about 40 s, 75 s, and 100 s after the P wave arrival. They are consistent with S wave to P wave conversions in the mantle beneath northwestern South America. The signal at 100 s, denoted as S1750P, has the highest amplitude and is formed at 1,750 km depth based on slant-stacking and semblance analysis. Waveform modeling using axisymmetric, finite difference synthetics indicates that S1750P is generated by a 10 km thick heterogeneity, presumably a fragment of subducted mid-ocean ridge basalt in the lower mantle. The negative polarity of S1750P is a robust observation and constrains the shear velocity anomaly δVS of the heterogeneity to be negative. The amplitude of S1750P indicates that δVS is in the range from −1.6% to −12.4%. The large uncertainty in δVS is due to the large variability in the recorded S1750P amplitude and simplifications in the modeling of S1750P waveforms. The lower end of our estimate for δVS is consistent with ab initio calculations by Tsuchiya (2011), who estimated that δVS of eclogite at lower mantle pressure is between 0 and −2% due to shear softening from the poststishovite phase transition.