Figure 1. Numerical simulation of thermal structure for eastern Aleutian subduction zone 50 Ma after initiation (i.e., steady state) based on model of Huang et al. (2000).

Abstract

The thermal structures of subduction zones vary in response to several parameters including age of subducted lithosphere, rate of convergence, and magnitude of frictional heating. Cooler subduction zones (e.g, Marianas, Tonga) are characterized by older plates and faster subduction (and vice versa); associated volcanic arcs have relatively steeply dipping slabs, greater slab length (as measured down-dip from trench to deepest earthquakes), narrow arc-trench gap, and magmas exhibiting stronger enrichments of fluid-mobile trace elements (FMEs; e.g., B, As, Sb). Warmer subduction zones (e.g., Cascades) have smaller slab dip, shorter slab length, wider arc-trench gap, and magmas exhibiting small to non-existent enrichments of FMEs. Slab thermal structure appears to have a fundamental influence on the chemistry of arc magmas in that warmer slabs will dehydrate at shallower depths and thereby carry smaller inventories of water and FMEs to subarc depths. To evaluate this hypothesis, it is necessary to predict relative thermal gradients for specific volcanic arcs.

We have developed a computer model to simulate the thermal structure of subducted slabs (Figure 1) using site-specific subduction parameters and slab geometries (defined by Benioff zone earthquake distributions), all of which vary substantially from one volcanic arc to another. Our results indicate that slab geometry exerts a significant influence on the calculated thermal structures; this factor has largely been ignored in previous thermal models. Most notably, slab-surface P-T curves for cool and warm subduction zones differ considerably (Figure 2). However, even the warmest model slabs rarely attain temperatures high enough to induce slab melting unless frictional heating contributions are unrealistically high. These results are in accord with observed variations in FME-enrichments, but pose problems for petrologic models invoking slab partial melting to explain inferred incorporation of sediment or oceanic crust contributions to arc magmas.

An important implication is that transfer of subducted components to arc magma sources most likely is achieved via transport in slab-derived fluids (as opposed to silicate melts). Also, high temperatures of arc basalts must be inherited from melting of relatively warm regions within the overlying mantle wedge, which may be convecting due to frictional coupling to the downgoing slabs.

A significant outstanding problem is that to date there are few independent constraints on absolute slab temperatures derived from thermal modelling. We are attempting to develop limits on maximum slab temperatures at subarc depths based on the mass balance of input and output inventories of boron in selected arcs. In essence, the fraction of initial boron retained in the slab is determined by the maximum temperature attained. Thus, the output inventory represented in a given arc segment determines a minimum slab inventory at depth, which in turn places an upper limit on the slab temperature at that subduction zone.

Figure 2. Comparison of model slab-surface P-T paths for the eastern Aleutian subduction zone. (PP = Pongo & Peacock, 1995, models A, B, and C reflect increasing shear stress; AK99 = Huang et al., 1999, progressively warmer curves reflect increasing shear stress; other curves from Peacock and Hyndman, 1999 and Oleskovich et al., 1999).