A Comparison of Synthetic Seismograms for 2D Structures: Semianalytical versus Numerical

Abstract

Teleseismic wave fields occasionally exhibit rapid changes in travel times and waveforms over distances less than several great-circle degrees when observed at broadband arrays. These rapid changes in wave field suggest the existence of significant structural transitions occurring over scales of several hundred kilometers or less in the mid- and deep mantle. Although approximate analytical methods based on raytracing can be readily adapted to structures having arbitrarily small scale lengths, it is important to validate their accuracy against the predictions of numerical methods. Here we compare synthetics from an approximate ray-based method WKBJ modified (WKM) against the pseudospectral method for a 2D model of the S-velocity anomaly associated with the South African plume. This model consists of a uniform 3% decrease in S velocity over a broad (>10°) region of the mid- and deep mantle beneath South Africa, contiguous at its bottom with a thin (100- to 200-km-thick) zone of low velocity extending 30° westward toward South America along the core-mantle boundary. Transitions between anomalous and radially symmetric structures of the test model are sharp, occurring over l0 km or less. SV and SH wave fields synthesized by the WKM and pseudospectral methods in this model generally agree with each other well. Slight mismatches in the two methods can be understood as the result of either differences in model parameterization or the effects of asymptotic approximations in the ray-based WKM method.