Abstract

Abstract Three factors have hampered attempts to find the causes of compression and extension within overriding plates of subduction zones: 1. (1) adjacent subduction zones are often similar in many subduction parameters, 2. (2) proposed causal variables are themselves correlated, and 3. (3) physical models are limited by uncertainties in physical quantities and complexity of the subduction process. Nevertheless, modern subduction zones provide a rich source of observations that help to distinguish among models. We consider ten variables or combinations of variables which may control strain regime behind trenches: convergence rate (Vc), slab age (As), shallow slab dip (Ds), absolute motion (Voa), asthenospheric flow, slab pull force, As + Vc, As + Voa, Vc + Ds + Voa, and Vc + Ds + As. Ten observations are used to test these models. A key prerequisite to these tests is the semiquantitative classification of the strain regime of each modern subduction zone into one of seven strain classes, from active back-arc spreading (class 1) through active folding and foreland thrusting (class 7). Statistical analysis indicates that the modern variance in strain classes is not adequately accounted for by any single or dual-variable model, but the combination of Vc, Ds, and either AsorVoa can account for 77% of this variance (R = 0.88). Rapid changes in strain class both with time and along a subduction zone are similarly more consistent with the 3-variable models. The role of Vc is also supported by maximum compression orientations parallel to Vc; the role of Ds is also supported by the tendency for oceanic overriding plates to be more extensional than continental overriding plates. Two factors suggest that the Vc + Ds + As model is more likely than Vc + Ds+ Voa: 1. (1) the dominance of one or more of the variables Vc, As, andDs in other subduction relations (slab length, earthquake moment, trench relative depth, deep dip, and arc-trench gap), and 2. (2) the correlation between strain class and maximum earthquake moment (which is affected by Vc and As). However, no confident discrimination between the two 3-variable models is yet possible.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call