S U M M A R YBroad-band P-and S-waves from earthquakes in South America recorded at Californian network stations are analysed to image lateral variations of the D -discontinuity beneath the Cocos plate. We apply two array processing methods to the data set: a simplified migration method to the P-wave data set and a double-array method to both the P-and S-wave data sets, allowing us to compare results from the two methods. The double-array method images a dipping reflector at a depth range from 2650 to 2700 km in the southern part of the study area. We observe a step-like topography of 100 km to a shallower reflector at about 2600 km depth to the north, as well as evidence for a second (deeper) reflector at a depth range from 2700 to 2750 km in the north. Results from the simplified migration agree well with those from the double-array method, similarly locating a large step in reflector depth in a similar location (about 2650 km depth in the south and about 2550 km in the north) as well as the additional deeper reflector at the depth of about 2750 km in the north. Waveform modelling of the reflected waves from both methods suggests a positive velocity contrast for S waves, but a negative velocity contrast for P waves for the upper reflector in agreement with predictions from mineral physical calculations for a post-perovskite phase transition. The data also show some evidence for the existence of another deeper reflector that could indicate a double intersection of the geotherm with the post-perovskite stability field, that is, the back-transformation of post-perovskite to perovskite close to the core-mantle boundary.

Abstract.We investigate the fine structure of the D" layer beneath the southwestern Pacific using shortperiod data from 7 deep events in the Fiji-Tonga region recorded at the J-Array in Japan. We use a double array migration method which integrates source array beamforming with receiver array beamforming. The phaseweighted stacking technique, which reduces incoherent noise by employing complex trace analysis, is also applied. This combination greatly enhances small phases common to the source and receiver beams. We present coherent phases originating from the lowermost mantle using these new array techniques. The results indicate that seismic energy is reflected at two negative discontinuities near 2550 and 2650 kin. The migration shows only weak evidence for scattering objects which may be an indication that scattering in the D" layer beneath the southwestern Pacific is weak.

The P and S wave velocity structure of the D″ layer beneath the southwestern Pacific was investigated by using short‐period data from 12 deep events in the Tonga‐Fiji region recorded by the J‐Array and the Hi‐net (two large‐aperture seismic arrays) in Japan. Reflected wave beam forming (RWB) and a migration method were used to extract weak signals originating from heterogeneities in the lowermost mantle. In order to acquire high resolution a double‐array method was applied to the data. The results of the RWB method indicate that seismic energy is reflected at discontinuities near the depths of 2520 and 2650 km, which have a negative P wave velocity contrast of 1% at the most. In addition, there is a positive seismic discontinuity at a depth of 2800 km. In the case of the S wave, reflected energy is produced almost at the same depth (2550 km depth). An apparent depth shift (50 km) of the discontinuity at the depth of 2850 km may indicate that the S wave velocity reduction in the lowermost mantle is ∼2–3 times stronger than that of P. A two‐dimensional cross section, constructed with the RWB method, suggests that the observed discontinuities can be characterized as intermittent lateral heterogeneities whose lateral extent is a few hundred kilometers. The migration shows weak evidence of scattering objects which belong to the seismic discontinuities detected by the RWB method. These anomalous structures may represent a part of hot plume generated beneath the southwestern Pacific in the lowermost mantle.

S U M M A R YBroad-band data from South American earthquakes recorded by Californian seismic networks are analysed using a newly developed seismic wave migration method-the slowness backazimuth weighted migration (SBWM). Using the SBWM, out-of-plane seismic P-wave reflections have been observed. The reflection locations extend throughout the Earth's lower mantle, down to the core-mantle boundary (CMB) and coincide with the edges of tomographically mapped high seismic velocities. Modelling using synthetic seismograms suggests that a narrow (10-15 km) low-or high-velocity lamella with about 2 per cent velocity contrast can reproduce the observed reflected waveforms, but other explanations may exist. Considering the reflection locations and synthetic modelling, the observed out-of-plane energy is well explained by underside reflections off a sharp reflector at the base of the subducted lithosphere. We also detect weaker reflections corresponding to the tomographically mapped top of the slab, which may arise from the boundary between the Nazca plate and the overlying former basaltic oceanic crust. The joint interpretation of the waveform modelling and geodynamic considerations indicate mass flux of the former oceanic lithosphere and basaltic crust across the 660 km discontinuity, linking processes and structure at the top and bottom of the Earth's mantle, supporting the idea of whole mantle convection.

S U M M A R YWe have developed a new array method combining conventional migration with a slownessbackazimuth deviation weighting scheme. All seismic traces are shifted based on the theoretical traveltime of the scattered wave from specific gridpoints in a 3-D volume. Observed slowness and backazimuth are calculated for each raypath and compared with theoretical values in order to estimate slowness and backazimuth deviations. Subsequently, stacked energy calculated by a conventional migration method is weighted by the slowness and backazimuth deviations to suppress any arrival energy whose slowness and backazimuth are inconsistent with the expected theoretical values. This new method was applied to two P-wave data sets which comprise (1) underside reflections at the 410 and 660 km mantle discontinuities and (2) D reflections as well as their corresponding synthetic data sets. The results show that the weighting scheme dramatically increases the resolution of the migrated images and enables us to obtain wellconstrained, focused images, making upper-mantle discontinuities and D reflections more distinct by reducing their surrounding energy.