The northern Red Sea is a continental rift in the process of transition from continental to oceanic rifting. We present 191 new heat flow measurements from the northern Red Sea forming three traverses across the water covered portion of the rift. The heat flow across the rift systematically increases from values of about 125 mW/m2 seaward of the coasts to average values greater than 250 mW/m2 in the axial depression. The heat flow measurements are evaluated for environmental disturbances. These are found to be generally small. The largest estimated disturbance results from the relief of the seafloor and of the top of a subbottom evaporite layer. The relief on these surfaces can account for the 20% point to point scatter typically observed in the heat flow measurements. Limits are placed on systematic disturbances to the heat flow pattern across the rift. The estimated largest systematic disturbance results from sediment blanketing which may cause a reduction in the heat flow on the order of 10%.

The heat flow variation across the rift is then utilized to examine the lithospheric thermal structure and the geometry and mechanism of extension. We employ a two-dimensional time-dependent numerical technique to follow the advection and diffusion of heat for a simple shear model and various pure shear models of lithospheric extension. The model of simple shear lithospheric extension along a planar shallow dip detachment produces significantly lower than observed heat flow, results in relatively small amounts of thermal thinning of the lithosphere compared to pure shear extension, and does not result in conditions likely to generate melt. The pure shear models with widening or constant width zones of extension also do not match the observed heat flow nor generate partial melt. The modeling study indicates that the actively extending region must become narrower through time in order to match the heat flow high in the center of the rift, generate some degree of partial melt and produce crustal subsidence within the observed width of the rift. A narrowing of the zone of active extension is also consistent with the pattern of tectonic activity inferred from sedimentary structures observed on seismic reflection profiles which indicate an abatement of activity in the marginal areas and intense present-day activity in the axial depression. In all of the models, an additional component of lithospheric heating other than that produced by extension appears necessary to match heat flow values in the marginal and near coastal areas, as well as explain the observed magnitude of rift flank uplift. Several scenarios in which the zone of extension eventually becomes narrow are consistent with the heat flow data alone, including those resulting in seafloor spreading for the past several million years. However, additional information provided by gravity, magnetic, and seismic reflection and refraction data appear to be most consistent with a history in which a period of relatively uniform extension across the rift is followed by a concentration of extension toward the center of the rift.

Such a model retains a small thickness of continental crust in the center of the rift and does not necessitate active seafloor spreading in the northern Red Sea. The three heat flow traverses indicate that maximum lithospheric thinning is occurring within the axial depression in the northern Red Sea which is very close to initiating seafloor spreading. There is evidence that upper crustal extension is locally concentrated in specific areas along the strike of the axial depression forming ''deeps'' associated with large recent intrusions. These deeps may constitute a precursory stage to seafloor spreading and appear to be nucleating at discrete and regularly spaced sites along the axis of the rift. The deeps are located midway between regularly spaced cross-trending fault zones or accommodation zones which, in the marginal areas, separate sets of fault blocks along strike.

Crustal extension appears to be taken up more diffusely in the accommodation zones than in the intervening fault block domains which are characterized by fewer, larger faults. As a result, initial crustal rupturing and emplacement of large intrusive bodies occurs halfway between the accommodation zones where more focused crustal extension forms the deeps. Comparison with other rifts suggests that the pattern of fault block domains separated by cross-trending fault zones is a common feature, that it originates early in the development of a rift, and that it may continue to influence the development of the rift up to the establishment of a continuous seafloor spreading axis. ¿ American Geophysical Union 1989