TOPOGRAPHY OF THE 400-KM DISCONTINUITY FROM OBSERVATIONS OF LONG-PERIOD P400P PHASES

New evidence is presented for lateral variation of the depth of upper mantle discontinuities. This evidence is derived from an analysis of amplitude of P400P phases (underside reflections from the 400 km discontinuity) for great circle paths with bounce points in the upper mantle near the Hawaii-Emperor hotspot chain. The data are long-period, from WWSSN, SRO and RSTN stations. The epicentral distance ranges from 95-degrees to 115-degrees, which ensures that no other major phases arrive in the time window of interest. Observed P400P amplitudes are compared with synthetic amplitudes for both a laterally homogeneous model and models including 3-D topography on the discontinuities. The observations permit models of topography that can be characterized by undulations of the 400 km discontinuity with wavelengths in the range 600-2000 km, and peak-to-peak amplitude of the order of 30-40 km. Such long-wavelength structure causes quite strong (de)focusing of P(d)P phases; synthetic P(d)P waveforms reveal that topography not only affects its amplitude, but also smears out the P(d)P wave, causing overlap between different P(d)P phases. This mechanism is proposed as the reason for the lack of well-developed P(d)P phases reported in the literature. The fact that the waveforms can be seriously distorted prohibits the use of P(d)P phases in a waveform inversion. The curvature of the discontinuity can be explained as the signature of convection currents crossing the discontinuity. Although the data cover the Hawaii hotspot, no signature of a plume structure can be detected in the data, probably because the resolving power of the data is too low.