Models for convergence, subduction and back-arc development

Abstract

Abstract Largely because of the wide variety of observational constraints which must be satisfied, the search for a viable driving mechanism is perhaps the most perplexing problem related to plate tectonics. The mechanism must be compatible with the rigid behavior of lithospheric plates, and with a wide range of plate sizes, shapes and motions. It must be consistent with complex configurations of plate boundaries and equally complex boundary interactions, such as the destruction of ridges at subduction zones. The mechanism must produce steady-state relative and absolute plate motions which persist for tens of millions of years, but must also account for sudden dramatic changes. Finally, the plate driving mechanism must be consistent with the non-Newtonian properties of olivine and with the fabrics of upper mantle peridotites. Mounting evidence suggests that plate motions result from forces associated with plate boundaries and that the principal resisting force is drag at the base of the lithosphere, particularly beneath continents Several investigators have suggested that gravitational forces acting on thermally-induced, lateral density variations in the upper mantle are the principal driving forces for plate tectonics. If so, plate motions are ultimately controlled by the temperature distribution in the upper mantle, and plate tectonics represents a state of dynamic equilibrium in which plate motions are both the cause and the consequence of temperature and density variations in the mantle. This concept requires that average absolute plate velocities be predictable from the characteristics of individual plates, and that plates tend to move down horizontal temperature gradients. A simple linear relation which includes contributions from ridge push (RP) , slab pull (SP) , trench suction (TS) and continental drag ( CD ): ( cm / y ) = (2.6 ± 0.4) + (4.8 ± 1.8) RP + (14.3 ± 1.7) SP +(3.5 ± 2.5) TS −(5.1 ±0.7) CD predicts plate velocities with an rms error of 0.44 cm/y, and a correlation coefficient of 0.98. That plate velocities can be accurately predicted from their own boundary configurations and proportions of continental lithosphere is strong evidence that plate motions result from negative buoyancy forces associated with plate boundaries.