Central North Island, New Zealand, provides an unusually complete geological and geophysical record of the onset and evolution of subduction at a continental margin. Whereas most subduction zones are innately two‐dimensional, North Island of New Zealand displays a distinct three‐dimensional character in the back‐arc regions. Specifically, we observe “Mariana‐type” subduction in the back‐arc areas of central North Island in the sense of back‐arc extension, high heat flow, prolific volcanism, geothermal activity, and active doming and exhumation of the solid surface. Evidence for emplacement of a significant percent of new lithosphere beneath the central North Island comes from heat flux of 25 MW/km of strike (of volcanic zone) and thinned crust underlain by rocks with a seismic wave speed consistent with underplated new crust. Seismic attenuation (Qp−1) is high (∼240), and rhyolitic and andesitic volcanism are widespread. Almost complete removal of mantle lithosphere is inferred here in Pliocene times on the basis of the rock uplift history and upper mantle seismic velocities as low as 7.4 ± 0.1 km/s. In contrast, southwestern North Island exhibits “Chilean‐type” back‐arc activity in the sense of compressive tectonics, reverse faulting, low‐heat‐flow, thickened lithosphere, and strong coupling between the subducted and overriding plates. This rapid switch from Mariana‐type to Chilean‐type subduction occurs despite the age of the subducted plate being constant under North Island. Moreover, stratigraphic evidence shows that processes that define the extensional back‐arc area (the Central Volcanic Region) are advancing southward into the compressional system (Wanganui Basin) at about 10 mm/yr. We link the progression from one system to another to a gradual and viscous removal of thickened mantle lithosphere in the back‐arc regions. Thickening occurred during the Miocene within the Taranaki Fault Zone. The process of thickening and convective removal is time‐ and temperature‐dependent and has left an imprint in both the geological record and geophysical properties of central North Island, which we document and describe.

S U M M A R YNear vertical and wide-angle seismic data provide evidence for a gradational crust-mantle boundary in a depth range of 15-20 km beneath the Central Volcanic Region (CVR), New Zealand. This volcanic area includes the Taupo Volcanic Zone and is a direct extension of Tonga-Kermadec oceanic backarc spreading into continental lithosphere. Long-range seismic refraction data show velocities of 6 km s −1 and less within the top 15 km of the crust of the CVR. At a depth of 15 km compressional seismic velocities increase to 6.8 km s −1 , and then to 7.4 ± 0.2 km s −1 at ∼20 km depth. These 7.4 km s −1 seismic wave speeds are interpreted as anomalous upper mantle as beneath this level passive seismic studies show similar Pn wave speeds that increase slowly to ∼7.8 km s −1 at about 80 km deep. We interpret rocks between 15 and 20 km to be a layer of new crust formed by underplating. The strongest reflection observed, and what might also be interpreted as a reflection Moho, is from the top of the proposed underplated layer at 15 km depth. At 20 km depth no such distinct reflection is observed. Rather, wide-angle reflection data show a continuum of low-level reflectivity between 15 and at least 35 km depth, indicating some heterogeneity and/or structure within the lower crust and upper mantle. Thus the transition from lower crust to upper mantle is broad, and a conventional reflection Moho does not exist beneath the CVR. Buoyancy force calculations based on rock uplift for the central North Island indicate that the subjacent mantle, to a depth of 80-100 km is ∼70 kg m −3 or 2 per cent less dense than normal mantle. Best estimates attribute half of this density anomaly to the effects of increased temperature with additional contributions from partial melt (∼1.2 per cent) and melt residuum.

Eocene onset of subduction in the western Pacific was accompanied by a global reorganization of tectonic plates and a change in Pacific plate motion relative to hotspots during the period 52–43 Ma. We present seismic-reflection and rock sample data from the Tasman Sea that demonstrate that there was a period of widespread Eocene continental and oceanic compressional plate failure after 53–48 Ma that lasted until at least 37–34 Ma. We call this the Tectonic Event of the Cenozoic in the Tasman Area (TECTA). Its compressional nature is different from coeval tensile stresses and back-arc opening after 50 Ma in the Izu-Bonin-Mariana region. Our observations imply that spatial and temporal patterns of stress evolution during western Pacific Eocene subduction initiation were more varied than previously recognized. The evolving Eocene geometry of plates and boundaries played an important role in determining regional differences in stress state.

A program of explosion seismology in central North Island, New Zealand, discovered a strong reflector within the upper mantle. Reflections from this (PmP2) are spatially confined to come from an interface 35 km deep and directly beneath a 40 km‐wide, back‐arc extension zone with active volcanism, high heat flow, low Pn wave‐speeds and thinned crust. On the basis of relative reflection amplitudes, the mantle reflections are most readily explained by an interface with a negative seismic impedance contrast. A satisfactory fit is obtained for a layer with a 40–90% drop in S‐wave speed (Vs) compared to the surrounding mantle. We interpret this layer to be a 40 km‐wide reservoir of partial melt pooled at a thermal boundary layer within the upper mantle.

SUMMARY A recent seismic refraction study across southern Norway has revealed that the up to 2469 m high Southern Scandes Mountains are not isostatically compensated by a thick crust. Rather, the Moho depths are close to average for continental crust with elevations of ∼1 km. Evidence from new seismic data indicate that beneath the highest topography Moho depths are around 38–40 km. These measurements are ∼2 km deeper than early estimates interpolated from coarsely spaced refraction profiles, but up to 3 km shallower than Receiver Function estimates for the area. Moho depth variation beneath the mountains roughly correlates with changes in surface topography indicating that topography is, at least to the first order, controlled by crustal thickness. However, the highest mountains do not overlie the thickest crust and additional support for topography, for example from flexural strength in the lithosphere, low densities in the upper‐mantle or mantle dynamics, is likely. The relationship between topography and Moho depth breaks down for the Oslo Graben and the Fennoscandian Shield to the east and north. High density lower crustal rocks below Oslo Graben and increasing crust and lithospheric thicknesses below the Fennoscandian Shield may produce a negative correlation between topography and Moho depth.