Differential Contraction of Subducted Lithosphere Layers Generates Deep Earthquakes

Abstract:

The origin of forces for generating deep earthquakes (>300 km depth) remains elusive. Published models for generating down-dip compressional stresses within the down-going slabs have potential problems to explain first order characteristics of deep earthquakes. These include: 1) stress magnitudes are too low compared to seismically implied pre-earthquake stress levels, 2) those based on slab buoyancy structures counter-act the subduction-facilitating slab pull force, inconsistent with the fact that most deep earthquakes occur within fast-subducting slabs, 3) fail to explain the maximum seismicity rate at ~600 km depth, and 4) inconsistent with depth-discontinuity of earthquake distribution within warm slabs (e.g., S. America), and the occurrence of outboard (e.g., Tonga) and isolated (e.g., Spain) deep earthquakes. A potential reason for these problems is because all the published models consider only the present-day snapshot of slabs, while in reality the past history of subduction may affect the present slab stress state significantly.

We propose a new mechanical model that considers the slab stress evolution during the entire history of subduction. The key mechanism of our model is shear strain accumulation between the crust and mantle lithosphere caused by differential volume changes during phase transformations as a slab sinks. This generates increasing down-dip compression inside the slab with depth, consistent with the global distribution of deep earthquake focal mechanisms. Using experimentally calibrated slab rheology, we show that the estimated distribution of slab internal stress agrees well with the peak seismicity rate at ∼600km depth, corresponding to an induced stress maximum. The model also predicts the depth-continuous seismicity within cold slabs and isolated seismicity swarm at ~600 km in warm slabs. This mechanism of residual stress within a cold slab also provides a solution to deep earthquakes occurring beneath former subduction zones such as western Tonga and Spain. We further suggest that this model may reconcile existing geodynamic paradoxes concerning the absolute strength of slabs, discrepancies on mantle rock strength between laboratory and geophysical inversions, as well as the competition between earthquake-generating stresses and slab-pull forces.