Heat flow and elevation of the marginal basins of the western Pacific

High heat flow and shallow crustal depths have been observed in many marginal basins landward of the trench-island arc system in the western Pacific. All the basins in the north-western Pacific where adequate heat-flow data exist, except the deep west Philippine basin, show a high mean heat flow. This is in marked contrast to the southwestern Pacific, where only the Fiji plateau and Lau basin have high heat flow. All active midocean ridges show a uniform relationship between heat flow and elevation: high and shallow where the ridge is young, and uniform and deep where the oceanic crust is old. In the absence of magnetic data in the marginal basins, this relationship has been used to determine whether or not these basins have the same thermal and dynamic structure as the midocean ridges. The heat flow and elevation data from the southwestern Pacific basins are compatible with the ridge data, whereas heat flow and elevation data from the Japan abyssal plain, Parece Vela basin, and the Shikoku basin in the northwestern Pacific have too high a heat flow for their mean depth. Convincing evidence has been presented that the marginal basins of the northwestern Pacific owe their origin and development to large-scale crustal extension landward of the island arc system. However, the intrusion of oceanic crust associated with this extension cannot alone adequately account for the heat flow and elevation data. Three other factors are discussed that, in addition to the intrusion process, may account for the observed distribution of heat flow and crustal depth: (1) continuous incipient intrusion, which raises the ‘effective’ thermal conductivity of the crust and hence the heat flow in these basins; (2) development of a thinner lithospheric plate landward of the island arc system of the northwestern Pacific; and (3) dynamic forces acting on the upper mantle landward of the downgoing slab, decreasing the observed elevation of the marginal basins. There are problems with all three of these factors, but a model using only the second is preferred, because it is simple to develop and is compatible with rigid plate tectonics.