Peridotite is widely considered to be the dominant component of the upper most mantle. Seismic velocities of a dry peridotite determined in the laboratory at high pressure and hypersolidus temperature show a rapid decrease with increasing temperature. The strong temperature dependence of velocities can be used to estimate temperature and melt fraction in the low-velocity zone of the Earth. The laboratory results show that the pressure dependence of both velocity and melt fraction appears to be well accounted for by the pressure dependence of the solidus temperature of the peridotite, i.e., homologous temperature dependence. This observation allows us to extrapolate laboratory results to higher pressures (greater depths) with some confidence, requiring only a knowledge of the solidus as a function of pressure. Using the ratios of lithospheric to asthenospheric velocities, temperature and melt fraction in the asthenosphere (the low-velocity zone) can be determined from laboratory velocity data. Examples of thermal structures of the upper mantle beneath oceanic plates are presented.

We chose three locations of geophysical interest: Iceland Plateau, Pacific Ocean, and Philippine Sea, where reliable seismic velocities have been determined in the upper mantle from surface waves studies. A large amount of partial melt (≥5 vol %) and a higher temperature than the dry peridotite solidus are inferred in 0--5 m.y. asthenosphere under the Iceland Plateau and the active marginal basin of the east Philippine Sea. The partial melt zone appears to extend deeper (>100 km) and to greater ages (>20 m.y.) in the Pacific Ocean region than under the slowly spreading Iceland Plateau. Partial melting is not expected in the asthenosphere older than 5 m.y. under the Iceland Plateau. Even though the volume fraction of partial melt is not so large (≤3 vol %) beneath the Pacific plate, the melting zone may extend to a distance of above 2000 km from the ridge. Many seamounts observed in the Pacific Ocean may result from this vast region of melting under the plate. We suggest a linear relation between the width of melting zone and the plate velocity. The laboratory results are also applied to other low-velocity zones.

It is suggested that a velocity drop up to about 6% in the asthenosphere does not necessarily require the existence of partial melt but can be explained by high subsolidus temperature. This implies that partial melt may not exist generally but only in limited areas in the low-velocity zone. Comparison of seismic results with theoretical models may overestimate the temperature and melt fraction, if the velocity drop at subsolidus temperature is neglected. Temperature and melt fraction obtained in this study are discussed together with results from heat flow and electrical studies. Almost the same temperatures as inferred from seismic velocity and a dry peridotite solidus are calculated from heat flow data, indicating that the upper mantle under mid-ocean ridges may be fairly dry. Even a water- or CO2-undersaturated solids (0.1 wt %) does not give a temperature consistent with the lithospheric geotherm and laboratory velocity results. It is noted that electrical conductivity studies generally yield lower temperatures and larger melt fractions than this study. Unknown pressure, frequency, and impurity effects on laboratory conductivities of partially molten mantle rocks are suggested as possible reasons for this discrepancy. ¿ American Geophysical Union 1989