The tectonic plate beneath the Pacific Northwest is cracking open, and scientists watching it happen say a section has already broken off entirely.
Findings published in Science Advances map two active tears in the seafloor off Vancouver Island, Canada, using deep-penetrating seismic imaging from the 2021 Cascadia Seismic Imaging Experiment.
The crack in the Explorer plate is 35 kilometres long, drops five kilometres straight down through solid rock, and is still growing. Earthquakes measuring magnitude 6.0 and above have ruptured along it multiple times in the past two decades, confirming the break is live and active. Where the crack has already propagated fully, the rock produces no seismic signal at all, because detached rock no longer grinds against anything. That silent zone is advancing. The plate is not cracking. It is coming apart in sections, one piece at a time, and the process has been running for four million years without anyone being able to see it clearly until now.
This is happening directly beneath the Cascadia Subduction Zone, the fault system that sits offshore from Seattle, Portland, and Vancouver and is capable of generating a magnitude 9.0 earthquake with minutes of warning and a tsunami that would reach the coast before emergency services could evacuate a single coastal town. The subduction zone is where the Juan de Fuca and Explorer oceanic plates grind beneath the North American plate at rates of up to four centimetres per year. That grinding process stores energy in the locked interface between the plates for decades or centuries at a time, then releases it all at once. The last full rupture of Cascadia happened in January 1700 and sent a tsunami across the Pacific that killed people in Japan. The fault has been locked and loading ever since.
The cracking now visible in the seismic images is not on the locked interface. It is happening inside the descending slab itself, the portion of the plate that has already gone beneath North America and is being pulled downward by its own weight. Understanding why it is cracking requires understanding what keeps a subduction zone running in the first place. The force responsible is called slab pull: cold, dense oceanic crust is heavier than the hot mantle it descends into, so gravity drags it downward, and the surface section of the plate gets hauled along behind it. Slab pull is ten times more powerful than any other tectonic force on Earth. It is what makes subduction zones effectively impossible to stop once they have started. The only thing that can overcome slab pull is a portion of incoming plate that is too warm and buoyant to sink, and that is exactly what is arriving at the northern end of Cascadia right now. A mid-ocean ridge, where new plate is being generated and is still young and light, is approaching the trench. The young plate resists sinking. The slab already below is still being pulled down. The plate in between cannot sustain both forces simultaneously, so it is tearing through the middle.
The tear in the Explorer plate sits 30 kilometres past the deformation front, the line on the seafloor where the plate first begins to buckle downward. At that point, the top of the crust drops more than five kilometres in a horizontal distance of roughly two kilometres, a near-vertical cliff face running through solid oceanic rock and mantle to a depth of 40 kilometres. The second tear, in the adjacent Juan de Fuca plate, sits 40 kilometres past the deformation front and is less developed: the plate there is buckling and folding across 10 kilometres of crust rather than snapping clean. The two tears are offset from each other by 20 kilometres, a displacement that matches the accumulated slip of the Nootka Fault Zone, the transform fault running between the Explorer and Juan de Fuca plates, at roughly 20 millimetres per year over approximately one million years. The tears were once a single continuous crack. The Nootka Fault severed them and dragged one section sideways relative to the other as it matured.
The Nootka Fault Zone is itself 20 kilometres wide at the surface and penetrates through the full oceanic crust into the upper mantle. Seismic images show it riddled with a network of fractures that reach 20 kilometres depth, far deeper than normal oceanic faults. Those fractures formed originally at the mid-ocean ridge, parallel to the spreading axis, and were later reactivated when the Explorer plate began resisting subduction approximately four million years ago. Seawater circulated through the open fault network for at least three million years before seafloor sediment sealed the conduits, and that prolonged fluid infiltration chemically weakened the rock over a zone more than 100 kilometres wide. The modern Nootka Fault is the concentrated, matured remnant of that original broad shear zone, now locked into a 20-kilometre-wide band of active left-lateral strike-slip faulting running northeast toward the coast.
The Explorer microplate is decelerating. GPS measurements along the British Columbia coast record its convergence rate with North America at approximately two centimetres per year, less than half the four-centimetre-per-year rate of the Juan de Fuca plate immediately to the south. As the Explorer slab pull weakens, the gravitational force that once drove it downward is being transferred to the Juan de Fuca plate instead, accelerating stress loading on that portion of the locked megathrust. The deformation front, the visible underwater ridge where the plate first bends downward at the trench, physically records this shift: it protrudes seaward on the Juan de Fuca side and retreats landward on the Explorer side, a measurable geometric consequence of one plate still pulling hard while the other loses traction. Seismic tremor, the low-frequency vibration generated by slow fluid movement and creep along the locked interface, occurs regularly beneath Vancouver Island on the Juan de Fuca side. On the Explorer side, tremor stops abruptly at the trace of the Nootka Fault. Below that line, the slab has already disconnected from the overriding plate across a section of the margin that spans the full depth of the coupled zone.
What this means for Cascadia’s earthquake hazard is not reassurance. The Juan de Fuca plate is subducting intact, accumulating strain across a locked interface that stretches more than 1,000 kilometres from northern California to southern British Columbia, and it is doing so without any reduction in speed or force. A full rupture of that interface produces a magnitude 9.0 earthquake in under two minutes of initial slip. The coast from Crescent City, California to Victoria, British Columbia receives tsunami inundation within 15 to 30 minutes depending on location. The Explorer plate occupies only the northernmost 75 kilometres of that system, and its structural failure does not unlock or reduce the stress stored in the main locked zone. What it does is introduce structural complexity into the northern end of any future rupture. Whether the slab tears act as barriers that stop a propagating earthquake rupture at the northern margin, or as conduits that redirect rupture energy in unexpected directions, is not yet determinable with existing instrumentation. The current offshore seismic network does not extend far enough beneath the seafloor to resolve the locked zone geometry in the immediate vicinity of the tears.
The analogue for where this ends is preserved off Baja California, where the shattered remains of the ancient Farallon plate lie fossilised on the Pacific Ocean floor. Those fragments, called the Guadalupe and Magdalena microplates, are what Cascadia’s Explorer plate is in the process of becoming: isolated sections of oceanic crust that were progressively pinched off from their parent plate, decelerated, and abandoned as the subduction system around them wound down. The Farallon system produced those remnants over roughly 30 million years of episodic breakup. The Cascadia system is at an early stage of the same cycle. The Explorer plate will fully detach from the subducting slab within approximately one million years, shortening the active Cascadia subduction zone by 75 kilometres, roughly one-twelfth of its current total length.
The Juan de Fuca tear, less developed but confirmed in the new imaging, is running a parallel process at a slower pace. Its buckling morphology, the fold structure visible in the seismic profiles across a 10-kilometre lateral distance, is consistent with a plate that has begun concentrating extensional stress at a single zone but has not yet fractured through. Magnitude 6.3 and 6.4 earthquakes ruptured that zone in 2004 and 2011. The fault extends from approximately 15 kilometres depth down to 40 kilometres, shallower at its upper limit than the Explorer tear, and the seismicity along it is wider and more distributed, consistent with deformation spread across a broader volume of rock rather than concentrated on a single through-going fault plane. The Juan de Fuca plate is not yet tearing. It is beginning to.
Both tears are currently active, both are producing earthquakes, and both are being monitored by seismic networks that were not designed with this structure in mind. The offshore component of the Pacific Northwest monitoring network is limited, and the resolution needed to track tear propagation at depth in real time does not currently exist along this margin.
Source:
huck, B., Boston, B., Carbotte, S.M., Han, S., Bécel, A., Miller, N.C., Canales, J.P., Hutchinson, J., Merrill, R., Beeson, J., Gurun, P., Littel, G., Nedimović, M.R., Savard, G. & Tobin, H. (2025). Slab tearing and segmented subduction termination driven by transform tectonics. Science Advances, 11(39), eady8347. https://doi.org/10.1126/sciadv.ady8347
