Deep Earthquakes

Depth distribution and classification Earthquakes generally classified as Intermediate (70-400 km) Deep (400-690 km) Not a single earthquake deeper than 690 km It is quite possible that intermediate and deep earthquakes have different causes Cessation of deep earthquakes at 690 km appears to occur at the “660 km discontinuity” Earthquakes deeper than 660 occur where the discontinuity is depressed

Deep Earthquake Distribution with Depth Frohlich [1989]

Deep earthquakes are not frictional sliding Frictional resistance = μ FN FN is proportional to ρgh ρ is fairly constant for different rock types So frictional resistance exceeds strength of rock for earthquakes deeper than 20-30 km

Proposed intermediate & deep earthquake faulting mechanisms Dehydration embrittlement -- pore fluids from dehydration reactions allow brittle failure at high normal stress Transformational faulting -- metastable phase transformation forms small lenses of low viscosity stable phase, which coalesce and lead to shear failure Plastic shear instabilities/thermal runaway -- formation of a shear zone due to strain and temperature softening, with runaway shear heating leading to failure -- extreme case would be fault zone melting

How to allow frictional slip at high pressure? High fluid pressure reduces effective normal stress Fault motion occurs through ductile shear, not brittle friction

Transformational Faulting “Anticracks” are observed in metastable phase transitions Associated with reductions in volume Filled with very fine grained low-strength transformed material Shear failure occurs along the low strength crack.

Shear zones & thermal runaway

transformational faulting model Diagram of the transformational Faulting model – deep earthquakes occur in a wedge of metastable olivine Evidence against the transformational faulting model The width of the seismic region is greater than possible metastable wedge widths Seismological evidence for a metastable wedge is unclear

Intermediate depth earthquakes often show a double seismic zone

Intermediate depth eqs result from dehydration? Hacker et al., 2003

Double Seismic Zone depth thermally controlled  Hot Cold  Wei et al., 2017

Imaging dehydration in the subducting crust? Abers et al, 2013

Dehydration of uppermost mantle serpentine ?

Similarities between Shallow and Deep Earthquakes Shear failure on planar surfaces (double couple) Generally similar rupture parameters (velocity, stress drop) Similar aftershock decay and magnitude statistics Aftershocks align along mainshock fault planes Repeating earthquakes and earthquake triggering observed

Differences Between Deep and Shallow Earthquakes Similar magnitude-frequency relations worldwide Large earthquakes occur only where small events occur Many aftershocks Deep earthquakes Seismic zones show vastly different magnitude- frequency relations Largest earthquakes are isolated Fewer aftershocks by 1 to 2 orders of magnitude

Deep earthquakes result from shear failure along a plane (double-couple) Not CLVD Mechanisms not isotropic Kawakatsu, 1996 Kuge & Kawakatsu, 1993; data from Harvard CMT

Statistical properties of deep earthquake aftershock sequences Data from 1994 Tonga (Mw 7.6) Aftershock sequence (Wiens & McGuire, 2000)

Deep earthquake aftershocks align along the main shock fault plane Visualization of the Aftershock sequence of the 1994 Tonga deep earthquake (Wiens et al., 1994) Ellipsoids represent 95% confidence regions of relative locations; white is mainshock; green are aftershocks.

Magnitude-frequency as a function of slab thermal parameter Wiens & Gilbert, 1996

The largest deep earthquakes occur in aseismic regions August 19, 2002 Tonga second mainshock (white), foreshock (yellow), aftershocks (green), and background seismicity (blue). From Tibi, Wiens, Inoue (2003)

Important clues to the mechanics of deep earthquakes Deep earthquake rupture is sensitive to the slab temperature Deep earthquakes show both static and dynamic triggering Events in previously aseismic regions are particularly sensitive to dynamic triggering Repeating earthquakes recur on the same fault and show small recurrence times

Deep Earthquake rupture is temperature sensitive Magnitude-Frequency relations Aftershock productivity Wiens & Gilbert, 1996 Thermal parameter – the product of the subduction convergence rate and the age of the lithosphere – inversely proportional to slab temperature for a cooling halfspace thermal model and cooling by conduction

Deep earthquakes can be dynamically triggered - two examples - 2002 Tonga 1986 Tonga Both cases involve dynamic triggering in aseismic regions with time scales of minutes Suggests that slabs are surrounded by critically stressed regions where seismic rupture initiates with difficulty Triggered events do not correspond to passage of seismic waves and do not define a consistent strain velocity – probably involve non-linear initiation mechanisms Tibi, Wiens, & Inoue,2003

Repeating Deep Earthquakes Deep earthquakes show repeated earthquakes showing nearly identical waveforms, suggesting recurrent faulting at the same location (Wiens & Snider, 2001) This pair of co-located events show that even details of the rupture process may be repeated.

Repeated rupture along the same fault Repeating deep earthquakes can be located accurately (+ 1 km) using cross correlation methods These events define fault-like features containing events with identical focal mechanisms The rupture area of each event can be estimated from the duration and scaling relations These events define repeated ruptures along the same fault segment Repeated rupture on the same fault may be inconsistent with transformational faulting, which is a non-repeatable process Wiens and Snider, 2001

Complicated deformation in deep slabs

Does Transformational Faulting Explain deep earthquakes? Do Metastable olivine wedges exist?

Slab seismic zone is too wide for transformational faulting, dehydration embrittlement Slab Seismicity and aftershocks 1994 Tonga Event Fault Slip during mainshock 1994 Tonga Event McGuire et al, 1997 Blue – background seismicity White - mainshock epicenter Green - aftershocks Yellow – foreshock Red -rupture termination from directivity Dots – aftershocks X - rupture initiation F – foreshock T – rupture termination

Travel time anomalies from metastable and equilibrium slabs Slab modeling is carried out for the Tonga slab (Koper at al, 1998), With data from the Labatts Experiment Results are inconclusive – effects of a metastable wedge are subtile due to diffraction around the low velocity wedge

Waveform effects of a metastable olivine wedge A metastable wedge produces large later phases following P and S waves due to guided waves within the low velocity wedge. A systematic search suggests these later phases are not observed Koper & Wiens, 2000

Metastable wedge imaged with receiver functions ?

Mechanism of intermediate and deep earthquakes still uncertain Intermediate depth earthquakes are most likely associated with dehydration reactions (but what about the deep zone of DSZs?) The most popular models for deep earthquakes are ductile shear zones and transformational faulting The lack of field observations is very limiting – progress will come from comparison of seismological and experimental results

Do slabs contain a metastable olivine wedge? Forward modeling: takes advantage of knowledge about slab temperature, mineralogy, morphology to perform resolution tests. Slab models include: Temperature structure from finite difference thermal model spatial configuration of the slab, as inferred from seismicity Phase transformations and associated velocity changes Slab models can incorporate a metastable olivine wedge, defined by an isotherm Temperature structure is converted to a velocity anomaly by assuming derivative that is found using a grid-search