ADJOINT SEISMIC TOMOGRAPHY USING FULL-WAVEFORM AND ENVELOPE INVERSION

Abstract

Convergence and stability of seismic full-waveform adjoint schemes are greatly improved when data and synthetics are progressively presented to the inversion algorithms in a constructive multiscale approximation using a wavelet transform. Surface waves, when properly accounted for, can and do preserve a wealth of near-surface features that would be lost in conventional body-wave tomography. To incorporate surface waves in full-waveform inversion, I use seismic envelopes in the context of wavelet-based multiscale analysis, which lessens the nonlinearity and prevents cycle-skipping problems. Seismic tomography is at risk of mapping source and other systematic uncertainties into estimates of wavespeed variations. To alleviate the corresponding nonuniqueness in the inverse problem, I developed a double-difference method for elastic adjoint tomography that pairs up measurements from the same source recorded at distinct stations. In double-difference tomography even one earthquake is able to resolve significant details of the wavespeed structure, while the computational cost of the necessary adjoint simulations is largely unaffected.