The rock beneath our feet is not still. Roughly 100 to 300 kilometres down, the mantle flows like extremely stiff toffee, creeping a few centimetres a year. You cannot see it, drill to it, or feel it, but you can hear it, because that slow flow leaves a fingerprint on every earthquake wave that passes through. This paper follows the chain from deep motion to measurable signal, and is careful about what the measurement constrains.
The mantle is a fluid on geological time
On human timescales the mantle is solid rock. On geological timescales it convects, and the distinction is one of viscosity rather than of state. Flow rates are centimetres a year, comparable to plate motion, because plate motion is largely what this flow drives.
That flow is not a curiosity. It sets the boundary conditions for everything shallower: how basins subside, where heat arrives, how stress is distributed in the crust above.
Flow leaves a fabric
The mantle's dominant mineral, olivine, is crystallographically anisotropic and it does not tumble randomly in a shear flow. It rotates toward alignment, so a region that has been deforming in a consistent direction ends up with its crystals statistically combed the same way. Geologists call this lattice-preferred orientation; it is a memory of deformation frozen into the rock.
The number that matters downstream is not the flow rate but how much alignment the rock has actually acquired. Weak or recently reorganised flow leaves a weak fabric, and everything that follows scales with the fabric rather than with the motion that made it.
Anisotropy is a directional speed limit
Aligned crystals make the aggregate anisotropic: a seismic wave travelling along the fabric goes faster than one travelling across it. Upper-mantle anisotropy of a few per cent is typical.
A few per cent sounds small and is not. Seismology measures arrival times precisely, and the effect accumulates over hundreds of kilometres of path, which is what makes it observable at the surface at all.
The signal is shear-wave splitting
A shear wave entering anisotropic rock splits into two components, one polarised along the fast direction and one across it. They travel at different speeds, so they arrive separated in time. That separation is the measurement.
Two numbers come out: the delay between components, which grows with path length and anisotropy strength, and the fast polarisation direction, which points along the fabric and therefore along the flow that made it.
Mantle flow rate
Typical upper-mantle anisotropy
Split delay, few hundred km
Quantities measured
What the measurement does not tell you
The delay accumulates along the entire path, which means a single station constrains a column, not a depth. A one-second delay is equally consistent with strong anisotropy in a thin layer and weak anisotropy through a thick one. That ambiguity is intrinsic to the measurement rather than a limitation of the instrument.
Resolving it requires additional constraint: dense station arrays for lateral variation, multiple phases with different paths, or forward modelling that couples the flow field to the predicted signal. Any interpretation that assigns a depth to a single-station splitting measurement without one of those is asserting more than the data supports.
From single station to a coupled model
The stronger approach runs the chain forwards. Model the flow, predict the fabric it produces, compute the anisotropy that fabric implies, synthesise the splitting a station would observe, and compare with what was recorded. Then adjust the flow model rather than the fabric, because the flow is the thing you actually want to know.
That loop is expensive: each iteration is a Stokes solve for the flow and a wave calculation for the signal. It is also where surrogate models earn their place, since the loop is a parameter sweep and sweeps are what amortisation is for.
Why an energy operator should care
Three reasons, in increasing directness.
Basin history. Mantle flow drives the vertical motions that create and destroy accommodation space. A basin's subsidence history is not fully explained by local processes, and the deep field is part of the explanation.
Stress. The crustal stress field an operator works within is anchored by deeper dynamics. Fracture orientation, fault reactivation potential and wellbore stability all read from that field.
Induced seismicity. Reactivation risk depends on how close a fault sits to failure, which depends on the ambient stress, which is set by processes that begin well below any reservoir.
Limitations
The chain from flow to signal is well established in outline and uncertain in detail. Fabric development depends on strain history, temperature, water content and deformation mechanism, and the mapping from fabric to elastic tensor carries its own assumptions. The instruments here are schematic illustrations of the relationships rather than measured or modelled results, and the figures quoted are representative scales rather than values from a specific study.
Key takeaways
- Mantle flow at centimetres a year aligns olivine crystals into a fabric, and it is the fabric strength rather than the flow rate that governs everything downstream.
- Aligned crystals make the rock anisotropic by a few per cent, which is observable because seismology measures arrival times precisely over long paths.
- A shear wave splits into fast and slow components, and the delay plus the fast polarisation direction are what a station measures.
- The delay accumulates along the whole path, so one station constrains a column and not a depth: assigning a depth without arrays, multiple phases or forward modelling asserts more than the data supports.
- For an operator the relevance runs through basin history, the ambient stress field, and therefore fault reactivation and induced-seismicity risk.
References
[1] Zhang, S. and Karato, S. Lattice preferred orientation of olivine aggregates deformed in simple shear. Nature, 1995.
[2] Silver, P. G. Seismic anisotropy beneath the continents: probing the depths of geology. Annual Review of Earth and Planetary Sciences, 1996.
