Home

Research

Teaching

Publications

Vita

Seismic tomography

 

Model S40RTS is our most recently developed shear velocity model of the mantle. The model represents isotropic shear velocity perturbations to the (anisotropic) PREM model that provide the best fit to 20 million Rayleigh wave dispersion [Van Heijst & Woodhouse, 1999], 500,000 shear-wave Traveltime [Ritsema & Van Heijst, 2001], and 1100 normal-mode Splitting function.measurements. The construction of S40RTS is discussed in Ritsema, J., A. Deuss, H.J. van Heijst & J.H. Woodhouse, Geophys. J. Int., 2011.

A gzipped TAR file that includes the model, codes to read the model coefficients, and GMT scripts for plotting cross-sections and maps can be downloaded. Since we compute the exact inverse, it is straightforward to construct models based on different dampings parameters (and hence different number of resolved unknowns), Contact me (jritsema AT umich DOT edu) to discuss how to best transfer thew files (they are too large to post here). Tomographic models from other research groups can be accessed via the REM website

(right) Maps of shear velocity perturbations (in %) at 100 km, 600 km (the transiztion zone), 1000 km and 2850 km (the core-mantle boundary region) depth in the mantle, according to model S40RTS. The shear velkocity is low (high) compared to PREM in regions shaded red (blue). .Shear velocity variations of more than 15% in the uppermost 100 km of the mantle are due to the ocean/continent variations and plate tectonics. High velocity anomalies in the transition zone indicate the position of slabs of subducted oceanic lithosphere. Broad low shear velocity structures beneath Africa and the central Pacific in the lower mantle are likely hot, but relatively dense (and stable) thermo-chemical piles..

 

Waveform analysis

While seismic tomography provides a large-scale atlas of wavespeed heterogeneity in the mantle, analyses of regional network data often provide complementary (and higher resolution) constraints of well-sampled regions in the mantle. Regional networks such as PASSCAL arrays, intended primarily for explorations of the lithosphere, have provided spectacular recordings of delayed and distorted body wave signals due to heterogeneous structure in the deep mantle.

(left) Tanzania Network recordings of S waves generated by earthquakes in Indonesia. S waves (the largest amplitude signals) from the Banda Sea earthquake, which have propagated through the D" region, are "split". S waves from a closer event in Java (at the same azimuth), which turn in the mid-mantle, are not split. This suggest the presence of shear velocity anisotropy at the base of the mantle beneath the Indian Ocean (adapted from Ritsema [2000]).

3D simulations of wave amplitudes

 
Tomographic models are the products of linearized inversions, relying on simplified theories for wave propagation. While resolution tests are insightful in understanding the effects of incomplete data coverage and inversion regularization, we need to resort to accurate 3D synthetics to determine how complex waveform effects (e.g., wave diffraction), omitted in the forward theory, affect the tomographic model. We are working with the Spectral-element method to determine 3D (i.e., S20RTS) waveform fits of surface waves, whether finite-frequency effects can be observable in body wave traveltimes, and whether body wave amplitude ratios may be useful data types for global tomography.

(right) Maps of the amplitude ratio SS/S (referenced to the PREM predicted ratio), plotted at the SS surface reflection point and expanded into degree-6 spherical harmonics. The map at the top is the observed SS/S ratio measured using low-pass filtered (T>15 s) transverse component data. At the bottom, the map of SS/S is determined using SEM synthetics for model S20RTS. The correlation between the two maps is significant (53%), suggesting that the variation of body wave amplitude is, to some degree, predictable by elastic wave speed variation. However, the amplitude of the SS/S is under-predicted by S20RTS. This may be due to the presence of (a) complex anelastic structure in the mantle (S20RTS assumes PREM anelasticity), or (b) strong wavespeed gradients in the mantle that are suppressed (by choice!) in the tomographic inversion.