the fault line keeps its schedule secret

Problem statement

We cannot predict when or how large subduction zone earthquakes and volcanic eruptions will be. The Cascadia subduction zone poses the largest domestic seismic risk — an eventual M9 megathrust earthquake — but the fundamental physics governing fault slip, magma ascent, and the transition from slow slip to catastrophic rupture remain poorly understood. Key unknowns include what controls whether a fault creeps or locks, what triggers the transition from slow slip events to full rupture, and what precursory signals, if any, precede major eruptions. The problem persists because monitoring networks have historically been split between onshore and offshore instruments, creating a critical observational gap at the most seismically active interface.

Why this matters

Subduction zones have produced every M9+ earthquake in recorded history and the majority of devastating tsunamis. The 2011 Tōhoku M9.1 earthquake caused ~20,000 deaths and the Fukushima disaster despite Japan having the world's most advanced seismic network. The Cascadia subduction zone's last M9 event was in 1700 and recurrence intervals suggest another is overdue. Over 10 million people in the Pacific Northwest live in the hazard zone. ShakeAlert earthquake early warning has coverage gaps at network edges, and regions of sparse station coverage cause missed events and false alerts.

What’s been tried and why it hasn’t worked

Earthquake and volcanic research have been treated as separate disciplines despite being driven by the same tectonic forcing — creating a disciplinary gap that prevents unified hazard models. Existing instrumentation cannot simultaneously capture the full range of spatial and temporal scales relevant to fault processes (nanometers to hundreds of kilometers, milliseconds to centuries). Onshore seismic and geodetic networks provide dense coverage of continental interiors but are sparse or absent on the seafloor where subduction occurs. ShakeAlert, the West Coast early warning system, only began integrating GNSS data in 2024 and still has systematic coverage gaps in the Pacific Northwest. Probabilistic seismic hazard analysis provides long-term statistical forecasts but cannot identify when a specific fault segment will rupture. Slow slip events were discovered in the 2000s and may be related to major earthquake nucleation, but the connection remains theoretical.

What would unlock progress

A cross-disciplinary, shoreline-crossing instrumental array (SZ4D's proposed MultiHazard Array) integrating continuous, high-density seafloor geodetic and seismic monitoring with onshore networks in the Cascadia, Chilean, and Alaska-Aleutian subduction zones. Integration of earthquake, volcanic, and surface process observations into unified models that treat the entire subduction system rather than individual hazards. Machine learning applied to the emerging catalog of slow slip events to identify patterns that may precede major ruptures.

Entry points for student teams

A student team could analyze publicly available slow slip event catalogs from the Pacific Northwest Seismic Network alongside seismicity data to test whether slow slip patterns correlate with subsequent earthquake activity on specific fault segments. Alternatively, a team could design a low-cost seafloor pressure sensor prototype for geodetic monitoring, addressing the key engineering challenge of long-term ocean-bottom deployment (power, biofouling, data transmission). Relevant skills: geophysics, signal processing, sensor engineering, data science.