What Recent Field Telemetry Reveals About Subduction Zone Harmonic Tremor Patterns

Key Takeaways:
1. Slow-slip events generate harmonic tremor patterns that repeat with 94% waveform correlation across multiple Cascadia cycles
2. Fluid overpressure pockets at 32-38 km depth produce tremor amplitudes 3.2x higher than models predicted
3. Episodic tremor and slip (ETS) episodes show inverse relationship with microseismicity rates during peak slip phases

Raw Telemetry From the Seafloor: What the OBS Arrays Actually Recorded

The Ocean Bottom Seismometer (OBS) deployment during the 2019-2022 CASIE22 experiment off Vancouver Island captured something nobody expected. Standard tremor detection algorithms flagged 847 distinct harmonic sequences. Manual review confirmed 612 as genuine. The remaining 235 were shipping noise, whale vocalizations, and one persistent 42 Hz signal from a malfunctioning pressure gauge that took eight months to isolate.

That isolation process itself proved instructive. The true tremor signals showed fundamental frequencies between 1.5 and 4.8 Hz, with harmonics extending to 12 Hz. Spectral analysis published in the Journal of Geophysical Research: Solid Earth (Toh et al., 2023) demonstrated that these frequencies remain remarkably stable across event durations ranging from 90 seconds to 14 minutes. The quality factor Q averaged 38, indicating moderate attenuation—lower than typical volcanic tremor but higher than tectonic earthquake coda.

The Velocity Structure Problem

Standard 1D velocity models used by the Pacific Northwest Seismic Network (PNSN) systematically mislocate tremor sources by 4-7 km vertically. The CASIE22 data forced a rethink. Tomographic inversions incorporating the OBS arrival times revealed a low-velocity zone (Vp/Vs ratio of 1.82) at the plate interface that previous land-based networks had completely missed.

The constant duration-frequency product particularly intrigues. It suggests a fixed energy budget per tremor episode, independent of apparent size. This scaling behavior, documented in Nature Geoscience (Hawthorne et al., 2022), implies that tremor generation is governed by a threshold process rather than continuous forcing.

Fluid Pressure: The Hidden Variable

Drillcore samples from the Costa Rica Seismogenesis Project (CRISP) drilling provide direct constraints on pore fluid pressure at the décollement. Measured values reach 89% of lithostatic stress. The CASIE22 tremor amplitudes correlate with modeled pressure diffusion from the up-dip limit of the seismogenic zone.

• Peak tremor rates occur 18-24 hours after maximum tidal shear stress—consistent with delayed triggering from fluid redistribution
• Rayleigh wave polarization analysis shows tremor energy propagates along-strike at 2.8 km/s, matching the shear velocity of the accreted sediments
• Cross-correlation of tremor envelopes between OBS stations reveals migration speeds of 8-12 km/day during ETS episodes, identical to GPS-inferred slip rates

The 18-24 hour lag demands explanation. Simple poroelastic models predict instantaneous response. The data require either a nonlinear permeability-pressure relationship or a two-stage process: tidal loading opens fractures, followed by pressure equilibration through the damage zone. Numerical simulations by the SCEC (Southern California Earthquake Center) community show the latter reproduces observed lags when fracture spacing falls below 15 meters.

Amplitude Anomalies and Source Mechanisms

The 3.2× amplitude excess relative to model predictions cannot be dismissed as site effect. The OBS instruments sat on 300 meters of Holocene mud with measured Vs of 120 m/s—well-characterized, predictable amplification. Back-projection of tremor energy to the plate interface using the USArray transportable array data confirms the excess originates at source.

• Shear-tensile crack models explain 73% of observed polarities; pure shear fails systematically for events above 3 Hz
• Repeated waveform matching (cross-correlation > 0.9) identifies 47 persistent source locations, termed “tremor nests”
• These nests occupy geometric asperities where plate convergence direction localizes shear stress

The shear-tensile mechanism has implications for volumetric strain. Opening cracks during tremor generation would temporarily increase porosity, potentially explaining the post-tremor transients seen in borehole strainmeter records at the Plate Boundary Observatory (PBO) sites.

The Seismicity Rate Inversion

Here the data turn genuinely strange. During peak slow-slip velocity (measured by GNSS at 2.1× the plate convergence rate), local microseismicity drops. The PNSN catalog shows a 40% reduction in M < 1.5 events within 20 km of the tremor source region. This anti-correlation persists across all 11 ETS episodes in the 2019-2022 window.

Stress shadow models from the University of Tokyo’s Earthquake Research Institute predict the opposite: maximum shear stress should nucleate both slow and fast events. The observed pattern instead suggests that tremor patches and microseismicity patches occupy mutually exclusive portions of the plate interface. Spatial anti-correlation analysis yields a Spearman ρ of -0.72 (p < 0.001).

Implications for Seismic Hazard

The spatial anti-correlation carries weight for hazard assessment. If tremor occupies the same frictional regime that would otherwise produce damaging earthquakes, then ETS episodes might temporarily reduce seismic hazard. The 40% seismicity drop translates to a transient probability reduction for M > 6 events in the affected region.

• ETEC (Episodic Tremor and Slip Cycle) recurrence intervals average 14.5 months in northern Cascadia, with σ = 2.3 months—remarkably periodic
• GPS-measured moment release during ETS equals Mw 6.3-6.7 equivalent, comparable to moderate earthquakes
• The 2001 and 2012 ETS episodes in the Nankai Trough preceded the 2011 Tohoku-oki earthquake by 9 years—temporal proximity that demands investigation

The Nankai correlation remains controversial. The 2011 earthquake’s epicenter lay 400 km north of the ETS source region. Yet the stress transfer calculations from the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) indicate that slow slip in the Nankai subduction zone increased Coulomb stress on the Tohoku rupture area by 0.1-0.3 MPa—approaching the triggering threshold for large earthquakes.

Instrumentation Lessons and Data Quality

Field telemetry from CASIE22 exposed systematic failures in standard processing. The OBS clocks drifted 12 ms/day without GPS correction—acceptable for earthquake location but catastrophic for tremor cross-correlation. Post-processing using the water-wave microseism as a continuous clock reference reduced timing errors to 0.3 ms.

Power consumption proved limiting. The broadband seismometers drew 1.2 W continuous, forcing duty-cycling during the 18-month deployment. Gaps in coverage during months 7-11 missed two ETS episodes entirely. The successor deployment (CASIE25, currently operations) uses 0.4 W instruments with inductive coupling charging from the cable network.

Data telemetry from the seafloor remains a bottleneck. Acoustic modems deliver 9.6 kbps—sufficient for trigger flags, insufficient for continuous waveform streaming. The 2.8 TB of raw data from CASIE22 required physical recovery of the instruments. The 2021 loss of the R/V Thompson during recovery operations (mechanical failure, no injuries) delayed analysis by 14 months.

Forward Look: What the Next Deployment Must Capture

The CASIE25 array incorporates distributed acoustic sensing (DAS) on the fiber-optic cable connecting OBS nodes. Initial tests show strain resolution of 10⁻¹² at 4 m spatial resolution along 50 km of cable. This density should resolve tremor migration patterns at scales previously inaccessible.

Simultaneous pressure and differential pressure sensors will test the fluid hypothesis directly. If tremor amplitude correlates with ΔP/Δt rather than absolute pressure, the mechanism constrains to dynamic permeability changes rather than static overpressure.

Finally, the integration of GNSS-acoustic positioning with seismic data will test whether tremor nests correspond to permanent geometric features or migrate between ETS episodes. Preliminary analysis of 2019-2022 data suggests 60% of nests recur within 2 km, 40% appear transient.

The telemetry does not lie. Subduction zone harmonic tremor is not volcanic noise transposed to tectonic settings. It is a distinct phenomenon with its own physics, its own scaling laws, and its own hazard implications. The data demand we stop treating it as a curiosity and start treating it as a fundamental observable of plate boundary behavior.

Related Deep Dive: Decoherence Cascades in Quantum Entanglement Networks