No abstract

S U M M A R YWe examine seismic velocity and anisotropy at the interface between the Australian and the subducting Pacific Plate in southern North Island, New Zealand using receiver functions from teleseismic earthquakes recorded on two L-shaped seismic arrays. IRIS station SNZO makes up the corner of the southern array. Velocity models are tested with a ray theoretical method that allows both dipping boundaries and anisotropy, assuming a 2-D structure beneath each array. The strongest arrival on the northern array consistently appears at 2 s. The arrival is modelled by a jump in velocity at a flat boundary at 12 km, which may be caused by a change in metamorphic grade. This prominent arrival correlates with a weaker arrival at 1.5 s (7 km) at station SNZO.Isotropic layers dipping with the slab can explain many features in the first 20 s of the records, but anisotropy is needed to explain both the transverse receiver functions and a change in the polarity of a prominent Ps arrival at 3 s delay, relative to P, on the radial component. This arrival is consistent with a change between an upper layer that is either isotropic or anisotropic with a horizontal symmetry axis, and a supraslab metasediment layer having anisotropy with a vertical slow axis of symmetry. There is tradeoff between boundary dip, the plunge of the axis of symmetry, layer thickness and the amount of anisotropy. A dip of 20 • (shallowing to 15 • under the Tararua array), thickness of 4 km, and a vertical symmetry axis fits the backazimuth variation of the amplitudes and polarities of the arrivals with anisotropy of 4-6 per cent. Observed layer multiples are smaller than synthetic predictions. To explain this discrepancy, either layer boundaries are rough, attenuation is strong, or the layer boundaries are more gradational at the periods (1 s) used in this study compared to previous studies (∼5 s).