The thermal structure of cratonic lithosphere from global Rayleigh wave attenuation

The resolution of and level of agreement between different attenuation models have historically been limited by complexities associated with extracting attenuation from seismic-wave amplitudes, which are also affected by the source, the receiver, and propagation through velocity heterogeneities. For intermediate- and long-period Rayleigh waves, removing the amplitude signal due to focusing and defocusing effects is the greatest challenge. In this paper, three independent data sets of fundamental-mode Rayleigh wave amplitude are analyzed to investigate how three factors contribute to discrepancies between the attenuation models: uncertainties in the amplitude measurements themselves, variable path coverage, and the treatment of focusing effects. Regionalized pure-path and fully two-dimensional attenuation models are derived and compared. The approach for determining attenuation models from real data is guided by an analysis of amplitudes measured from synthetic spectral-element waveforms, for which the input Earth model is perfectly known. The results show that differences in the amplitude measurements introduce only very minor differences between the attenuation models; path coverage and the removal of focusing effects are more important. The pure-path attenuation values exhibit a clear dependence on tectonic region at shorter periods that disappears at long periods, in agreement with pure-path phase-velocity results obtained by inverting Rayleigh wave phase delays. The 2-D attenuation maps are highly correlated with each other to spherical-harmonic degree 16 and can resolve smaller features than the previous generation of global attenuation models. Anomalously low attenuation is nearly perfectly associated with continental cratons. Variations in lithospheric thickness are determined by forward modeling the global attenuation variations as a thermal boundary layer of variable thickness. Temperature profiles that satisfy the attenuation values systematically overpredict and underpredict Rayleigh wave phase velocity in cratons at short and long periods, respectively. Introducing a low-velocity layer at depths 60-80 km and a high-velocity layer that begins at 200 km can resolve the discrepancy. The former is consistent with receiver-function detections of a mid-lithospheric discontinuity, and the latter may correspond to the Lehmann discontinuity.