Computing Rates and Distributions of Rock Recovery in Subduction Zones

Buchanan Kerswell1, Matthew Kohn1, Taras Gerya2,

1Department of Geology & Environmental Earth Science, Miami University, Ohio

2Department of Earth Sciences, ETH-Zürich

October 26 2022

Acknowledgments

High performance computing

  • Borah (Boise State University)
  • Euler (ETH Zürich).

Special thanks


Glossary

Term Description
markers numerical objects representing rock bodies in geodynamic simulations
rocks subduction-related rock samples w/ PT estimates and other metadata
recovery detachment of rock bodies from the subducting plate (max PT\(^*\))
recovery mode region in PT space with high sample density [high frequency]
PT pressure temperature
HP high pressure
OP oceanic plate
UP upper plate

Summary

Research questions

Where are rocks recovered along subduction interface shear zones?

How do recovery rates and distributions vary among subduction zones?

How do numerical and empirical PT distributions compare?




Main findings

Markers are overwhelmingly detached from ≤ 1 GPa

PT grads correlate w/ OP age & UP thickness, while depths correlate w/ velocity

Few markers detach from the highest-density regions of natural samples

Background  Where are rocks recovered from? How many?

Few rocks are recovered, but all come from the interface

Calvert et al. (2020)

The interface

Deformation styles

  • Stick-slip [discrete brittle]
  • Shear zone [mixed]
  • Coupled [distributed viscous]
  • Depends on dehydration rxns


Seismic properties

  • Low velocity zones (LVZs)
    • Few kms thick
    • High \(V_p/V_s\) ratio
    • High Poisson ratio

Background  Where are rocks recovered from? How many?

Geophysical images hint at shear zone structure…

Tewksbury-Christle & Behr (2021)

…and SZ rocks preserve exquisite detail

Behr & Bürgmann (2021)

Background  Where are rocks recovered from? How many?

Recovery is expected at key rheologic transitions…

Behr & Bürgmann (2021)

…and not expected beyond viscous coupling

Agard (2021)

Background  Where are rocks recovered from? How many?

Samples are distributed pretty smoothly…

Data from Penniston-Dorland et al. (2015)

…yet some clusters (modes) are apparent

Data from Agard et al. (2018)

Methods  Trace & classify markers in numerical experiments

Interface rheology controlled by hydrologic model

Kerswell et al. (2021)

Model setup

Fixed parameters

  • Rheologic model
  • Hydrologic model
  • Material properties
  • Boundary conditions


Varied parameters

  • Velocity (40-100 km/Ma)
  • OP age (32-110 Ma)
  • UP thickness (46-94 km)


Hydrologic model

  • Cont. slab dehydration
  • Atg forms weak interface

Methods  Trace & classify markers in numerical experiments

Recognizing recovery is an unsupervised classification problem

Kerswell et al. (in prep.)

Classification

Algorithm

  1. Trace markers
  2. Get max PTs*
  3. Apply clustering
  4. Clusters located @:
    ≥ 3 ˚C/km &
    ≤ 1300 ˚C &
    ≤ 3.4 GPa
    are recovered


Uncertainty

  • Count: \(\pm\) 43 (2 \(\sigma\))
  • Ratio: \(\pm\) 0.23% (2 \(\sigma\))


* along prograde path

Methods  Trace & classify markers in numerical experiments

Kerswell et al. (in prep.)

Determining maximum PTs of recovered markers

  • Max P&T determined independently
  • Heating after peak burial overestimates PT grads
  • Most markers have small time gaps bw max P&T

Results  How does recovery vary among subduction zones?

Stronger fill colors = stronger monotonic correlation

Kerswell et al. (in prep.)

Recovery modes vs. initial conditions

Modes

  • 1 = most markers
  • 2 = deepest markers


Correlations

  • grads → age & UPT
  • depths → velocity
  • temps → UPT
  • rates → UPT

Results  The marker recovery gap

Slow slabs = deeper recovery

All markers

Young slow slab

Old slow slab

Kerswell et al. (in prep.)

Results  The marker recovery gap

Fast slabs = shallower recovery

All markers

Young fast slab

Old fast slab

Kerswell et al. (in prep.)

Results  The marker recovery gap

Thin upper plates = warmer recovery

All markers

Young slab, thin upper plate

Old slab, thin upper plate

Kerswell et al. (in prep.)

Results  The marker recovery gap

Thick upper plates = cooler recovery

All markers

Young slab, thick upper plate

Old slab, thick upper plate

Kerswell et al. (in prep.)

Conclusions & Implications  

Recovery depths

Marker recovery modes correspond with mechanical transitions inferred from seismic imaging


  • Underplating/mélange at 1 GPa (e.g. Bostock, 2013)
  • Minor recovery near viscous coupling depth at ~2.3 GPa

SZ settings

Marker show appreciable deviations from the rock record except for young, slow OPs with thin UPs


  • Increasing average T does not fill in the marker recovery gap
  • Recovery rates are not correlated with OP age or velocity
  • Recovery rates are poor for thin UP lithospheres

Recovery gap

Across 64 numerical experiments, less than 1% of markers are recovered from 1.8-2.2 GPa and 500-625 ËšC


  • Poor implementation of detachment mechanisms (modeling bias)
  • Rock PTs are systematically misinterpreted (petrologic bias)
  • Rocks are (re)sampled from the same conditions (scientific bias)
  • Rocks are recovered early and/or during short-lived events (tectonic bias)

References

Agard, P. (2021). Subduction of oceanic lithosphere in the alps: Selective and archetypal from (slow-spreading) oceans. Earth-Science Reviews, 214, 103517.
Agard, P., Plunder, A., Angiboust, S., Bonnet, G., & Ruh, J. (2018). The subduction plate interface: Rock record and mechanical coupling (from long to short time scales). Lithos, 320-321, 537–566.
Behr, W. M., & Bürgmann, R. (2021). What’s down there? The structures, materials and environment of deep-seated slow slip and tremor. Philosophical Transactions of the Royal Society A, 379(2193), 20200218.
Bostock, M. (2013). The moho in subduction zones. Tectonophysics, 609, 547–557.
Calvert, A., Bostock, M., Savard, G., & Unsworth, M. (2020). Cascadia low frequency earthquakes at the base of an overpressured subduction shear zone. Nature Communications, 11(1), 1–10.
Kerswell, B., Kohn, M., & Gerya, T. (2021). Backarc lithospheric thickness and serpentine stability control slab-mantle coupling depths in subduction zones. Geochemistry, Geophysics, Geosystems, 22(6), e2020GC009304.
Penniston-Dorland, S., Kohn, M., & Manning, C. (2015). The global range of subduction zone thermal structures from exhumed blueschists and eclogites: Rocks are hotter than models. Earth and Planetary Science Letters, 428, 243–254.
Tewksbury-Christle, C., & Behr, W. (2021). Constraints from exhumed rocks on the seismic signature of the deep subduction interface. Geophysical Research Letters, 48(18).