Finite difference modelling of P-wave scattering in the upper mantle

Scattering within the upper mantle is prominently documented in the so-called high-frequency teleseismic P-n phase, generated by velocity fluctuations in the upper mantle. This phase is seen in many data sets from active and passive seismology. Whenever the source-generated frequencies are high enough (5-10 Hz) to allow their observation at large distances (> 2000 km), this phase appears in the recordings of the Russian PNE programme, in other long-range refraction experiments, and in numerous recordings of earthquakes. Using a 2-D finite difference scheme we calculate complete synthetic seismograms in order to relate the scales of the fluctuation of the elastic parameters to wavefield properties. We discuss the influence of critical parameters on the scattering properties of the upper mantle and on the propagation mechanism of the teleseismic P-n: variations in vertical and horizontal correlation lengths, rms velocity fluctuations, thickness of the heterogeneous layer, and cross-correlation of P- and S-wave velocities. A teleseismic P-n phase arises only if the velocity fluctuations are strong enough to cause multiple scattering and if the horizontal correlation length exceeds the vertical by an order of magnitude. Our favourite model for the Russian PNE profiles includes a 100-150-km-thick zone of scatterers below the Moho, containing heterogeneities with a horizontal correlation length of 20 km, a vertical correlation length of 0.5 km and an rms velocity perturbation of 2 per cent, while keeping the P- and S-wave velocities correlated. As the teleseismic P-n is often observed, this scale of velocity fluctuations seems to be characteristic of the uppermost mantle. Anisotropy is likely to contribute to the inferred velocity fluctuations.