High Resolution Studies of Deep Earth Structure

Citation

Xiaoming, Ding (1998) High Resolution Studies of Deep Earth Structure. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/z21r-bk54. https://resolver.caltech.edu/CaltechTHESIS:08032023-144000226

Abstract

Recent advances in seismic tomography has imaged major deep structure in the lower mantle. The ring of fast velocities originally derived from global long-period inver­sions has been resolved into interspersed sheet-like structure which appears to be old slabs. Beneath some of the structure, there are high velocity zones (HVZ) with variable thickness approaching the core mantle boundary (CMB). Detailed broadband modeling of waveforms produced by seismic paths sampling one of these zones yields a picture of D" zone, with one thermal negative above the layer and one positive approaching the CMB, which would seem to be quite compatible with pancaked slab debris. In contrast, modeling waveforms produced by sampling the slowest velocity regions along the CMB reveals a thin ultra low velocity zone (ULVZ). The dimensions of these zones range from a few hundred km beneath Iceland to a few thousand km beneath Africa.

Seismic data recorded on TERRAscope and Berkeley Digital Seismic Network are used to study the HVZ beneath Central America. Modeling these waveforms (P, SV and SH) constitutes a major portion of this thesis. Two modeling strategies were employed in the thesis: (1) Assume a "Lay type" D" with a sharp velocity discontinuity; (2) Assume an upper transition zone approaching D", and a lower transition zone approaching the CMB (old slabs). Our preferred model following strategy (1) (Chapter 2) has an S discontinuity 200 km above the CMB with 3% jump and a negative gradient in the D" layer. It fits S, SKS, Scd on radial component and S and Scd phase on tangential component in the distance range of 78° to 92°. A 2D model with decreasing thickness of D" layer toward North is also investigated. It fits the data better at the distances beyond 90° and produces reasonable good fit of the S and Scd phases. This suggests that the typical negative gradient in the "Lay type" D" may not be necessary as in SDH. P waveform modeling, on the other hand, shows no indication of a corresponding Ped phase. The PREM model fits the P travel time and waveform well. Our preferred model following strategy (2) (Chapter 3) has a positive gradient initiated 350 km above the CMB with a sharper increase near 200 km and a strong negative radient begins at about 100 km. This model can explain both P and S waveforms.

In Chapter 4 the ULVZ beneath Iceland and Africa are addressed. The major phases used to study the ULVZ are SKS and S_[P_(diff)]KS which travels along the CMB as P at both the core entry (S_[P_(diff)]KS) and exit (SK_[P_(diff)]S) locations. A major structure beneath Iceland (SK_[P_(diff)]S) as identified from data recorded on stations in Northern Europe appears to be shaped like a dome, 80 km high, 200 km wide with a 10% drop in P and S velocities. The data for Africa is less complete but highly anomalous. Shear wave record sections across Africa and Europe containing the cross-over from S to SKS and extended core-phases (75° to 120°) are presented from deep South American events. These are compared against corresponding synthetics for various tomography models computed with a new 2D synthesizing technique (Appendix A). Some of the most recent models, Grand [1994], explain the observation for African data better than lD models. However, considerable fine tuning is required in D" to explain abrupt changes in S and ScS waveforms and the extreme cases in SKS-S travel times. Essentially, Grand's anomaly needs to be increased to -4% with evidence for a strong plume (1500 km of vertical structure with -4% velocity drop) to explain the SKS travel times and waveform data. The plume is located along the eastern edges of the basal low velocity region.

By studying the various branches of the core phases PKP, it has become quite clear that North-South paths in the inner-core appear faster than East-West paths. Moreover, the broadband seismograms associated with these paths are distinct. The reason for this difference is not known but suggests a lower (anisotropic) inner-core with an upper (isotropic) inner-core which may have variable thickness. Modeling of long period and broadband data for such structures is, also, addressed in Chapter 5 of this thesis.

Thesis Files