No abstract

When continents rift to form new ocean basins, the rifting is sometimes accompanied by massive igneous activity. We show that the production of magmatically active rifted margins and the effusion of flood basalts onto the adjacent continents can be explained by a simple model of rifting above a thermal anomaly in the underlying mantle. The igneous rocks are generated by decompression melting of hot asthenospheric mantle as it rises passively beneath the stretched and thinned lithosphere. Mantle plumes generate regions beneath the lithosphere typically 2000 km in diameter with temperatures raised 100–200°C above normal. These relatively small mantle temperature increases are sufficient to cause the generation of huge quantities of melt by decompression: an increase of 100°C above normal doubles the amount of melt whilst a 200°C increase can quadruple it. In the first part of this paper we develop our model to predict the effects of melt generation for varying amounts of stretching with a range of mantle temperatures. The melt generated by decompression migrates rapidly upward, until it is either extruded as basalt flows or intruded into or beneath the crust. Addition of large quantities of new igneous rock to the crust considerably modifies the subsidence in rifted regions. Stretching by a factor of 5 above normal temperature mantle produces immediate subsidence of more than 2 km in order to maintain isostatic equilibrium. If the mantle is 150°C or more hotter than normal, the same amount of stretching results in uplift above sea level. Melt generated from abnormally hot mantle is more magnesian rich than that produced from normal temperature mantle. This causes an increase in seismic velocity of the igneous rocks emplaced in the crust, from typically 6.8 km/s for normal mantle temperatures to 7.2 km/s or higher. There is a concomitant density increase. In the second part of the paper we review volcanic continental margins and flood basalt provinces globally and show that they are always related to the thermal anomaly created by a nearby mantle plume. Our model of melt generation in passively upwelling mantle beneath rifting continental lithosphere can explain all the major rift‐related igneous provinces. These include the Tertiary igneous provinces of Britain and Greenland and the associated volcanic continental margins caused by opening of the North Atlantic in the presence of the Iceland plume; the Paraná and parts of the Karoo flood basalts together with volcanic continental margins generated when the South Atlantic opened; the Deccan flood basalts of India and the Seychelles‐Saya da Malha volcanic province created when the Seychelles split off India above the Réunion hot spot; the Ethiopian and Yemen Traps created by rifting of the Red Sea and Gulf of Aden region above the Afar hot spot; and the oldest and probably originally the largest flood basalt province of the Karoo produced when Gondwana split apart. New continental splits do not always occur above thermal anomalies in the mantle caused by plumes, but ...

The equations governing the movement of the melt and the matrix of a partially molten material are obtained from the conservation of mass, momentum, and energy using expressions from the theory of mixtures. The equations define a length scale d c called the compaction length, which depends only on the material properties of the melt and matrix. A number of simple solutions to the equations show that, if the porosity is initially constant, matrix compaction only occurs within a distance ~<5 C of an impermeable boundary. Elsewhere the gravitational forces are supported by the viscous stresses resulting from the movement of melt, and no compaction occurs. The velocity necessary to prevent compaction is known as the minimum fluidization velocity. In all cases the compaction rate is controlled by the.properties of the matrix. These results can only be applied to geological problems if the values of the permeability, bulk and shear viscosity of the matrix can be estimated. All three depend on the microscopic geometry of the melt, which is in turn controlled by the dihedral angle. The likely equilibrium network provides some guidance in estimating the order of magnitude of these constants, but is no substitute for good measurements, which are yet to be carried out. Partial melting by release of pressure at constant entropy is then examined as a means of produced melt within the earth. The principal results of geological interest are that a mean mantle temperature of 1350 °C is capable of producing the oceanic crustal thickness by partial melting. Local hot jets with temperatures of 1550 °C can produce aseismic ridges with crustal thicknesses of about 20 km on ridge axes, and can generate enough melt to produce the Hawaiian Ridge. Higher mantle temperatures in the Archaean can produce komatiites if these are the result of modest amounts of melting at depths of greater than 100 km, and not shallow melting of most of the rock. The compaction rate of the partially molten rock is likely to be rapid, and melt-saturated porosities in excess of perhaps 3 per cent are unlikely to persist anywhere over geological times. The movement of melt through a matrix does not transport major and trace elements with the mean velocity of the melt, but with a slower velocity whose magnitude depends on the distribution coefficient. This effect is particularly important when the melt fraction is small, and may both explain some geochemical observations and provide a means of investigating the compaction process within the earth.

Examination of more than 100 fault plane solutions for earthquakes within the Alpide belt between the Mid-Atlantic ridge and Eastern Iran shows that the deformation at present occurring is the result of small continental plates moving away from Eastern Turkey and Western Iran. This pattern of movement avoids thickening the continental crust over much of Turkey by consuming the Eastern Mediterranean sea floor instead. The rates of relative motion of two of the small plates involved, the Aegean and the Turkish plates, are estimated, but are only within perhaps 50 per cent of the true values. These estimates are then used to reconstruct the geometry of the Mediterranean 10 million years ago. The principal difference from the present geometry is the smooth curved coast which then formed the southern coast of Yugoslavia, Greece and Turkey. This coast has since been distorted by the motion of the two small plates. Similar complications have probably been common in older mountain belts, and therefore local geological features may not have been formed by the motion between major plates.A curious feature of several of the large shocks for which fault plane solutions could be obtained for the main shock and one major aftershock was that the two often had different mechanisms. 113 20 40 60 FIG. 3. Epicentres of foci at and below 50 km between 1961 and 1970. The depth of the foci given by the USCGS were used to obtain this figure, and may well be in error by 50 km. Since focal depths of greater than 50 km are believed to occur only where plates are being destroyed, the activity in this figure suggests that crustal shortening is now taking place over a wide region in the Eastern Mediterranean and South-west Asia. * Dielines of Figs 1 and 9 at a scale of 1:12 233 OOO, Figs 15-18 at 1: 1 million and Figs 24, 25, 29 and 3 1 at 1 :2.5 million can be provided at cost.

Seismic refraction results show that the igneous section of oceanic crust averages 7.1±0.8 km thick away from anomalous regions such as fracture zones and hot‐spots, with extremal bounds of 5.0–8.5 km. Rare earth element inversions of the melt distribution in the mantle source region suggest that sufficient melt is generated under normal oceanic spreading centers to produce an 8.3±1.5 km thick igneous crust. The difference between the thickness estimates from seismics and from rare earth element inversions is not significant given the uncertainties in the mantle source composition, though it is of the magnitude that would be expected if partial melt fractions of about 1% remain in the mantle and are not extracted to the overlying crust. The inferred igneous thickness increases to 10.3±1.7 km (seismic measurements) and 10.7±1.6 km (rare earth element inversions) where spreading centers intersect the regions of hotter than normal mantle surrounding mantle plumes. This is consistent with melt generation by decompression of the hotter mantle as it rises beneath spreading centers. Maximum inferred melt volumes are found on aseismic ridges directly above the central rising cores of mantle plumes, and average 20±1 and 18±1 km for seismic profiles and rare earth element inversions respectively. Both seismic measurements and rare earth element inversions show evidence for variable local crustal thinning beneath fracture zones, though some basalts recovered from fracture zones are indistinguishable geochemically from those generated on normal ridge segments away from fracture zones. This is consistent with a model where the melt generated beneath spreading ridges is redistributed to intrusive centers along the ridge axis, from where it may flow laterally along the axis at crustal or surface levels. The melt may sometimes flow into the bathymetric lows associated with fracture zones. Oceanic crust created at very slow‐spreading ridges, and in regions adjacent to some continental margins where rifting was initially very slow, exhibits anomalously thin crust from seismic measurements and unusually small amounts of melt generation from rare earth element inversions. We attribute the decreased mantle melting on very slow‐spreading ridges to the conductive heat loss that enables the mantle to cool as it rises beneath the rift.