Crustal anisotropy beneath P acific O cean‐ I slands from harmonic decomposition of receiver functions

Karato

et al. 2016

Self Cite

Receiver function observations in the oceanic upper mantle can test causal mechanisms for the depth, sharpness, and age dependence of the seismic wave speed decrease thought to mark the lithosphere‐asthenosphere boundary (LAB). We use a combination of frequency‐dependent harmonic decomposition of receiver functions and synthetic forward modeling to provide new seismological constraints on this “seismic LAB” from 17 ocean‐bottom stations and 2 borehole stations in the Philippine Sea and northwest Pacific Ocean. Underneath young oceanic crust, the seismic LAB depth follows the ∼1300 K isotherm but a lower isotherm (∼1000 K) is suggested in the Daito ridge, the Izu‐Bonin‐Mariana trench, and the northern Shikoku basin. Underneath old oceanic crust, the seismic LAB lies at a constant depth ∼70 km. The age dependence of the seismic LAB depth is consistent with either a transition to partial‐melt conditions or a subsolidus rheological change as the causative factor. The age dependence of interface sharpness provides critical information to distinguish these two models. Underneath young oceanic crust, the velocity gradient is gradational, while for old oceanic crust, a sharper velocity gradient is suggested by the receiver functions. This behavior is consistent with the prediction of the subsolidus model invoking anelastic relaxation mediated by temperature and water content, but is not readily explained by a partial‐melt model. The Ps conversions display negligible two‐lobed or four‐lobed back azimuth dependence in harmonic stacks, suggesting that a sharp change in azimuthal anisotropy with depth is not responsible for them. We conclude that these ocean‐bottom observations indicate a subsolidus elastically accommodated grain‐boundary sliding (EAGBS) model for the seismic LAB. Because EAGBS does not facilitate long‐term ductile deformation, the seismic LAB may not coincide with the conventional transition from lithosphere to asthenosphere corresponding to a change in the long‐term rheological properties.

Section: Evaluation Of the Melt And Anisotropy Modelsupporting

confidence: 69%

Section: Methods and Resultsmentioning

confidence: 99%

Nature of the seismic lithosphere‐asthenosphere boundary within normal oceanic mantle from high‐resolution receiver functions

Olugboji

Karato

et al. 2016

Self Cite

“…Rychert et al () estimate Sp receiver functions, which typically involve lower frequencies than Ps receiver functions, to infer an underplated layer beneath the Galapagos Islands that extends to 37 ± 7‐km depth. Olugboji and Park () reported underplated layers beneath 11 ocean‐island stations within the Pacific Ocean basin, using a harmonic regression of Ps receiver functions over back azimuth to investigate anisotropic layering with frequency cutoff 1 Hz. With this low‐passed restriction, Olugboji and Park () typically inferred thick layers beneath their stations, roughly 30 km for the combination of crust and underplated layer.…”

Section: Seismic Evidence For Anisotropic Underplated Layersmentioning

confidence: 99%

“…In the case of Hawaii, data from ocean‐bottom seismometers have detected a sub‐Moho layer, attributed to magmatic underplating, that extends hundreds of kilometers from the loci of active hot spot volcanism (Leahy et al, ). Olugboji and Park () reported that the underplated layer at 11 hot spot islands contained anisotropy that weakly correlated with inferred stress directions at the time of island formation.…”

Section: Introductionmentioning

confidence: 99%

“…Leahy and Park () estimated radial receiver functions with frequency cutoffs up to f c = 4.0 Hz to detect thin layered structures. Olugboji and Park () estimated both radial and transverse RFs with f c = 1.0 Hz with harmonic regression (Bianchi et al, ; Park & Levin, ; Schulte‐Pelkum & Mahan, ) to resolve anisotropy within the sub‐Moho underplated layer. Here we apply the harmonic RF regression of the latter study with the higher‐frequency cutoffs of Leahy and Park () to provide a high‐resolution sample of seismic structures that must be explained by any hot spot‐underplating model, and to reconcile previous analyses.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Why Is Crustal Underplating Beneath Many Hot Spot Islands Anisotropic?

Rye

2019

Self Cite

High‐frequency harmonic regression (2.0–4.0‐Hz cutoff) of receiver functions at two long‐running seismic observatories at mid‐Pacific hot spot islands confirms earlier detections of underplated material with seismic velocities intermediate to crust and mantle, and reveals it to be multilayered and anisotropic within ~30 km of the surface. Magmatic underplating beneath the oceanic Moho has been proposed to accompany basaltic melt that erupts at the seafloor and (eventually) atop a subaerial volcano. An alternate hypothesis is “metasomatic underplating” whereby crustal fractures developed during magma ascent allow seawater to infiltrate and to serpentinize the sub‐Moho mantle partially. Metasomatic underplating would lower seismic wave speeds, promote the buoyancy of the hot spot swell, and induce textural anisotropy as metamorphic expansion of olivine‐rich peridotite promotes a crack network along which serpentinization spreads. Differential expansion of mantle peridotite and crustal gabbro promotes cracks in the crust that offer new pathways for seawater to descend to the Moho, allowing metasomatic underplating to expand laterally and to contribute anisotropy to the underplated layer. Rare serpentinized mantle xenoliths confirm that crack textures can develop during serpentinization at depth. The discovery of iron‐oxidizing microbial mats on the seafloor flank of the Loihi volcano, and many locations of diffuse low‐temperature venting worldwide, is consistent with the circulation of metasomatic fluids with reducing chemistry, sourced from serpentinization at depth. Posteruptive uplift of Santa Maria Island (Azores), and asymmetry of the Hawaiian swell, suggests that underplating requires 2–4 Myr to complete, suggesting that fluid infiltration is slow, subject to cycles of blockage and fresh fracturing.

Reply to comment by Kawakatsu and Abe on “Nature of the seismic lithosphere‐asthenosphere boundary within normal oceanic mantle from high‐resolution receiver functions”

Olugboji

Karato

2016

Self Cite

Kawakatsu and Abe (2016) have highlighted the potential complicating effect of sediment reverberations on the analysis and interpretation of crust and mantle phases inferred from receiver functions analyzed from ocean-bottom seismograms. In their comment, they identify resonant peaks in the power spectrum at one of the stations, T06, and demonstrate with synthetic modeling how sedimentinduced resonances can cause instability in the recovered receiver-function (RF) traces. They also request a detailed explanation of how LQT rotation is conducted, and why its use leads to stable receiver functions in the analysis of . We welcome this query as an opportunity to highlight certain technical aspects of the data-analysis procedures used in . Our methods derive partly from methods recommended by previous studies of receiver functions estimated from seismic seafloor data, particularly the use of the modal wavefield decomposition (which we approximated by the LQT rotation) to suppress reverberation signals in the overlying water column. Approach Used in Calculating Receiver FunctionWe use multiple-taper correlation (MTC) to estimate receiver functions [Park and Levin, 2000], which estimates Ps converted-waves from the portions of the P, SV, and SH motions that are mutually coherent in narrow band-passes (J. Park and V. Levin, Statistics and frequency-domain moveout for multiple-taper receiver functions, submitted to Geophysical Journal International, 2016), discarding the incoherent portion of the wavefield. We apply moveout corrections to the MTC RFs before stacking in the frequency domain to enhance primary Ps conversions from deep interfaces, and suppress reverberations [Helffrich, 2006;Bianchi et al, 2010] (Park and Levin, submitted manuscript, 2016). Finally, synthetic modeling suggests that, even at our noisiest stations, defocusing of seismic energy within the sediment layer leads to reverberations of much smaller amplitude than predicted by simple 1-D models. Real seafloor seismic observations exhibit other evidence for resonant behavior that plausibly arise from ambient motions at the seafloor, because they can be detected in the seismic noise prior to P-wave arrival. Our identification of the deep LAB interface relies on the observed moveout of its negative-amplitude Ps with epicentral distance, when we analyze P-coda for receiver functions. This Ps conversion (with moveout) is absent when preevent data are analyzed. Adjustable Empirical LQT RotationWe use an empirical LQT rotation as an approximate method of wave-field decomposition [e.g., Vinnik, 1977;Bostock, 1998;Reading et al., 2003;Svenningsen, 2004]. An explicit modal wave-field decomposition is advocated by Bostock and Trehu [2012] to reduce contamination of water and sediment reverberations. The modal decomposition relies on either a collocated seafloor pressure sensor or prior knowledge of the sediment layer to effect its correction. The LQT rotation is crude by comparison, but can be scaled to less than a full rotation.We used an empirical r...