GEOPHYSICAL RESEARCH LETTERS, VOL. 22, NO. 7, PAGES 747-750, APRIL 1, 1995 The style of the Tyrrhenian subduction Anna Maria Marotta Dipartimento di Geologia e Geofisica, Universit& di Bari Roberto Sabadini Dipartimento di Scienze della Terra, Universit/x di Milano Abstract. Momentum and energy equations are solved simultaneously for an incompressibleviscous fluid in or- der to model the changesin the shape of a subducted slab when active convergence between the subducting and overriding plates coxaxes to the end and slab pull becomes the dominant tectonic mechanism. This model can be applied to the Tyrrhenian domain, where it has been suggested that active convergence terminated about 7-9 Myr ago. During the active phase the an- gle of immersion of the slab at intermediate depths be- tween 100 and 270 km is small, about 45ø-50 ø, and large, about 80ø-90 ø, at depths greater than 300 kin. The phase of passivegravitational sinking is character- ized by a substantial modification in the shape of the slab, with a large angle of immersion of 70 o now at intermediate depths, decreasingto 500 in proximity of the tip of the slab. When the shape of the modelled slab is compared with the seismogenic portion of the subducted Ionian lithosphere in the Tyrrhenian, our re- sults are consistent with subduction driven by slab pull and with cessationof active convergence between 7-9 Myr before present. Introduction A quantitative approach to the modelling of subduc- tion in the Tyrrhenian basin requires the development of numerical models that account for the peculiarities in the tectonic style of this area. Subduction underneath the Calabrian arc plays a key role, as suggested by a wide literature, not quoted here for brevity. Somegeo- logical investigationsseem to indicate that the style of subductionchanged about 8 Myr ago, from a configura- tion of a neutral arc, with active convergence between the subducting and overriding plates, into a new tec- tonic style where the gravitational sinking of the sub- ducted slab becomes dominant [Pataccaet al., 1993]. This tectonic processis tested by means of a finite el- ement scheme in which momentum and temperature equationsfor a viscous fluid are solvedfor active conver- gence followedby gravitational sinkingof the subducted Copyright 1995 by the American Geophysical Union. Paper number 95GL00635 0094-8534/95/95GL-00635503.00 plate. Predictions on the shape of the subducted slab, based on the changes with depth of the dip, are coxat- pared with the hypocentral distribution of deep earth- quakes. The flow, driven by a velocity boundary condition or by density contrasts, is confined in a two-dimensional Cartesian geometry and the equation of motion, conti- nuity and energy are solved simultaneouslyfor an infi- nite Prandtl number and incompressibleviscous fluid; the equation of motion is given by 0- X72ff- X7p + RaTE (1) where u is the velocity, p the pressure, T the tempera- ture, Ra the Rayleigh number for convectionin a fluid heated from below and Els the vertical unit vector. The energy equation reads OT = V2T- (2) ot These equations have been non-dimensionalized such that the Rayleigh number Ra = pgc•ATd3/•v charac- terizesthe problem, with p the density,g the gravity, c• the coefficient of thermal expansion, AT the tempera- ture difference between the surface and the bottom of the upper mantle, d the thickness of the upper man- ale, • the the thermal diffusivity and v the viscosity, specified in the caption; the lateral dintension of the cell is 2d. Velocity components are set to zero at the bottom and at the vertical walls of the cell; the non- dimensional temperature used in the calculation is de- finedby T'= (T-T,)/(Tb- T,) where Tb= 1600øKis the bottom temperature and T, = 300øK is the surface temperature, both held constant. The penalty func- tion method has been used within a two-dimensional finite-element scheme [Sewell, 1985].The quality of the solution for temperature dependent viscosities has been testedwith previous convection results [Quareniet al., 1985]. Results Fig. I deals with the temperature fields of the two vis- cous models considered in our calculations; the bottom of the lithosphere is defined by the 1400øK isotherm, and contour lines are separated by 200øK. Panels(a) deals with the phase of active convergence, as depicted by the black arrow, while (b)and (c) correspond to 747