Resolving the lithosphere-asthenosphere boundary with seismic Rayleigh waves

Abstract

SUMMARY Seismic surface wave measurements offer tight constraints on shear wave speed values within the lithosphere and asthenosphere. It is still a matter of debate, however, how accurately and under what conditions surface waves can resolve the depth and thickness of the lithosphere–asthenosphere boundary (LAB). We investigate the sensitivity of Rayleigh waves to LAB properties and find that if the LAB is associated with a 2–3 per cent shear speed reduction at a depth less than 100 km, then 10–20 km variations in its depth translate into phase-velocity changes of up to 1 per cent. Variations in the thickness of the LAB from zero to a few tens of kilometres cause much smaller phase-velocity changes (∼0.1 per cent), although the signal of the LAB-thickness variations is enhanced by a distinct frequency-dependent pattern of phase-velocity increases and decreases. For LAB depths increasing beyond 100 km, larger absolute variations in the LAB depth and thickness are needed to generate response of the same amplitude in phase velocities. We introduce a grid-search inversion of phase-velocity curves for the LAB depth and thickness, defining them as the middle and the width, respectively, of a depth interval with a linear shear-speed decrease in it. The inversion comprises dense sampling of the LAB depthLAB thickness parameter plane and non-linear, gradient-search inversions at every point; it accountsforanytrade-offsoftheLABparameterswithshearspeedvariationsaboveandbelow the LAB. Inversions of phase-velocity curves for the LAB depth and thickness show that the two parameters have uncorrelated uncertainties, with the LAB depth constrained better than the LAB thickness. Random errors in phase-velocity measurements have a limited effect on LAB-depth measurements, causing errors of only a few kilometres. Measurements of the LAB thickness are less robust and are not possible in the presence of large (∼1 per cent) random errors. Substantial systematic biases in the measurements (errors that persist over broad period ranges of tens of seconds) can affect both LAB-depth and LAB-thickness estimates; phasevelocity curves with such biases are not suitable for inversions for LAB properties. Resolving the LAB, however, in particular the LAB depth, should already be possible with some of the accurate, broad-band, phase-velocity measurements available today. Applying our method to phase-velocity data measured in Phanerozoic west-central Germany, we find best-fitting LAB depths in the 60–80 km range (the LAB depth always understood to be the middle of the shear-speed reduction depth range), with minimum misfits achieved near the 70-km LAB depth. Beneath the Archean Guyana Shield, we determine a 140–175 km LAB-depth range, with a 160 km best-fitting (mid-)LAB depth.