Multiple phase analysis of the shear velocity structure in the D″ region beneath Alaska

Abstract

A large data set (more than 990 total waveforms) comprised of North American station recordings of a variety of long‐period S phases (S, ScS, Sdiff, sS, sScS, SKS, SKKS) from 36 earthquakes in western Pacific subduction zones is analyzed to determine characteristics of the shear velocity structure in the D″ region beneath Alaska. Three hypotheses that have been proposed to explain observed ScS precursors and S waveform distortions for this region are appraised with the multiple phase data set, and the radial and lateral variations of the shear velocity structure in the 1500‐km broad portion of D″ sampled are also assessed. The observed travel time and waveform variations may result from diffraction by near‐source slab structure, scattering by low‐velocity heterogeneity in the lower mantle, or interaction with complex velocity gradients in D″ associated with a deep mantle velocity discontinuity. We utilize waveform modeling and differential time analysis to resolve greater detail of the structure than can presently be obtained by standard travel time tomography techniques, and find that the discontinuity model is most consistent with the data. Subdivision of the data by source region and source to receiver azimuth establishes that the observations exhibit fairly stable behavior, which is consistent with extensive local stratification of D″. The data show sufficient coherence over the 1500‐km broad region that we find it useful to present a local average radial model. The preferred model, SYLO, has a 2.75% velocity increase 243 km above the core mantle boundary and a negative velocity gradient within the D″ layer. Analysis of SKS‐S, SKS‐ScS, and SKS‐SKKS differential times also indicates that the outermost 100 km of the core below the fast D″ region underlying Alaska must be slower than the PREM structure. The observations exhibit significant scatter, most of which can be attributed to unmodeled heterogeneity in the midmantle and near‐source regions, as well as lateral variations in the D″ region. The complex structure of D″ that we prefer may be associated with large‐scale downwelling in the lower mantle, but it is not certain whether it is a purely thermal or a combined chemical and thermal effect.