Subduction zones are the type of convergent boundary responsible for virtually all of the planet’s most destructive earthquake-and-tsunami combinations. At these boundaries, one tectonic plate dives beneath another, and the contact surface between them, called the megathrust fault, accumulates enormous stress over decades or centuries before releasing it in seconds. That sudden release can shift the seafloor upward by tens of meters, shoving the overlying water column into motion and launching a tsunami. Other types of convergent boundaries produce earthquakes too, but only subduction zones routinely generate both hazards at catastrophic scale, and the reasons have as much to do with geometry and ocean depth as with raw seismic power.
How the Megathrust Produces Both Hazards at Once
Subduction megathrust faults are the longest and most powerful fault systems on Earth. They extend from the deep-sea trench, where the downgoing plate first bends beneath the overriding plate, to a depth where rock becomes too hot and soft to break in a brittle fashion. Across that entire surface, the two plates are locked together by friction during quiet periods, storing strain energy the way a compressed spring stores force. When the fault finally ruptures, the overriding plate snaps upward and seaward, generating an earthquake whose shaking can last several minutes and whose energy dwarfs anything produced by other fault types.
The tsunami part of the equation comes from the shallow end of that fault. When slip extends close to the trench, it directly lifts or drops large patches of seafloor, displacing the water above them. During the 2011 Tohoku earthquake in Japan, slip on the shallowest portion of the megathrust reached roughly 62 meters over the nearest 40 kilometers to the trench, producing the devastating tsunami that followed.1Nature Communications. Large fault slip peaking at trench in the 2011 Tohoku-oki earthquake The 2004 Sumatra-Andaman earthquake ruptured more than 1,500 kilometers of megathrust, with slip exceeding 20 meters at shallow depths off northern Sumatra, where the enormous Indian Ocean tsunami originated.2Nature. Plate-boundary deformation associated with the great Sumatra–Andaman earthquake In both cases, it was the shallow, near-trench portion of the fault that mattered most for tsunami generation.
Researchers have confirmed this pattern across many subduction zones. The behavior of the shallow megathrust, though, remains harder to predict than the deeper portions, because frictional properties near the trench vary in ways that current monitoring struggles to capture.3Tectonics. Strain Signals Governed by Frictional‐Elastoplastic Interaction of the Upper Plate and Shallow Subduction Megathrust Interface Over Seismic Cycles
Oceanic-Continental Versus Oceanic-Oceanic Subduction
Subduction zones come in two flavors, and both produce earthquakes and tsunamis. In oceanic-continental subduction, a dense ocean plate dives beneath a lighter continental plate. The west coast of South America is the textbook example: the Nazca Plate slides under the South American Plate, building the Andes and hosting some of history’s largest earthquakes. In oceanic-oceanic subduction, two ocean plates converge and the older, denser one goes under. The Tonga-Kermadec Trench in the South Pacific and the Mariana Trench are examples of this arrangement.
Both types can generate devastating tsunamis because the key ingredient, a large fault surface beneath the ocean floor, is present in each case. The megathrust faults in both settings share the same fundamental mix of behaviors: rapid stick-slip earthquakes, slow-slip events, and steady creep, all controlled by the heterogeneous frictional properties of the fault.4Geosphere. Subduction zone megathrust earthquakes What differs between specific subduction zones is not so much the oceanic-vs-continental distinction but local factors like the amount of sediment being carried into the trench, the roughness of the seafloor on the downgoing plate, and the rigidity of the overriding plate.
Modeling work has shown that along-arc variations in tectonic stress, combined with how stiff or soft the near-fault sediments are at different depths, control both earthquake dynamics and tsunami size.5Nature Geoscience. Stress, rigidity and sediment strength control megathrust earthquake and tsunami dynamics In other words, two subduction zones of the same type can behave very differently depending on what is sitting on and around the fault. That is why hazard assessment has to be done zone by zone rather than by a simple classification.
Why Continental Collisions Rarely Trigger Tsunamis
Continental-continental convergent boundaries produce powerful earthquakes but almost never cause significant tsunamis. The Himalayan front, where the Indian Plate pushes into the Eurasian Plate, generates earthquakes that can exceed magnitude 7 or 8. The 2015 Nepal earthquake killed thousands. But these faults are buried under land, not ocean. A tsunami requires vertical displacement of the seafloor to move a water column, and when the fault rupture is entirely beneath a continent, there is no ocean water to displace.
There are edge cases. A large earthquake near a coast or beneath a lake can generate local water waves, and landslides triggered by shaking near the sea can push water into motion. But the systematic, repeating cycle of megathrust locking, rupture, and tsunami that defines subduction-zone hazards does not exist at continental collision boundaries. If you are asking which convergent boundary is responsible for the deadliest combined earthquake-and-tsunami events in recorded history, the answer is subduction zones, without exception.
What Makes Some Subduction Earthquakes More Tsunamigenic Than Others
Not every large subduction earthquake triggers a damaging tsunami. Some magnitude-8 events barely ripple the ocean surface, while certain magnitude-7 earthquakes generate waves far out of proportion to their shaking. The difference comes down to where and how the fault slips.
The most tsunamigenic ruptures are those in which a large amount of slip occurs at shallow depth, near the trench. The 2011 Tohoku earthquake illustrated this dramatically: slip peaked right at the trench, producing impulsive tsunami waves that were acutely sensitive to the amount of near-trench displacement.6Journal of Geophysical Research: Solid Earth. A Self‐Consistent Fault Slip Model for the 2011 Tohoku Earthquake and Tsunami The Japan Trench investigation confirmed strong lateral variations in how slip was distributed near the trench, suggesting that local geological features steer where the worst displacement occurs.7PubMed. Investigating a tsunamigenic megathrust earthquake in the Japan Trench
A particularly puzzling category is the so-called “tsunami earthquake,” where ground shaking feels moderate but the resulting tsunami is unexpectedly large. The 1992 Nicaragua earthquake is a well-studied example. Researchers found that the earthquake ruptured very shallow, weak rock near the trench, which produced a slow, long-duration rupture with little high-frequency shaking but large, trenchward-increasing slip. The low rigidity of near-trench rock explains both the deceptive mildness of the shaking and the outsized tsunami.8PubMed Central. Large slip, long duration, and moderate shaking of the Nicaragua 1992 tsunami earthquake caused by low near-trench rock rigidity That finding is important for warning systems: a moderate earthquake felt on shore does not always mean the tsunami threat is moderate too.
The rigidity of the overriding plate also matters at a broader scale. Analysis of wave-speed data from subduction zones worldwide shows that depth-dependent variations in rigidity systematically control how much slip occurs at different depths during an earthquake, offering a potential tool for estimating slip and issuing faster tsunami warnings.9Nature. Upper-plate rigidity determines depth-varying rupture behaviour of megathrust earthquakes
Splay Faults and Amplified Tsunamis
The megathrust itself is not the only fault that matters during a subduction earthquake. Splay faults are secondary faults that branch off the megathrust and cut upward through the overlying wedge of rock and sediment. When earthquake rupture propagates onto a splay fault, it can push a narrow block of seafloor sharply upward, amplifying the tsunami beyond what megathrust slip alone would produce.
Detailed imaging of accretionary wedges has revealed networks of splay faults that could pose significant tsunami hazard. Modeling shows that when multiple splay faults rupture alongside the megathrust, the resulting tsunami consists of distinct wave packets: a tall, sharp crest from the largest splay fault, plus a broader wave from slip on the others. This multi-pulse pattern produces larger coastal flooding and greater run-up distance than a single long-wavelength wave from the megathrust alone.10Journal of Geophysical Research: Solid Earth. Earthquake Rupture on Multiple Splay Faults and Its Effect on Tsunamis
A well-studied example comes from the Nankai Trough off southwest Japan. There, a large splay fault has progressively steepened over geological time, substantially increasing its potential to vertically uplift the seafloor. Researchers concluded that slip on this fault most likely contributed to devastating historical tsunamis, including the 1944 magnitude-8.1 Tonankai earthquake, and that the fault geometry makes this margin particularly prone to generating tsunamis.11PubMed. Three-dimensional splay fault geometry and implications for tsunami generation Work on the Cascadia subduction zone off the Pacific Northwest has similarly highlighted how splay fault structure may influence tsunami sources during future great earthquakes there.12AGU Advances. Structural Controls on Splay Fault Rupture Dynamics During Cascadia Megathrust Earthquakes
Outer-Rise Earthquakes and Local Tsunami Threats
There is one more earthquake source associated with subduction zones that can generate tsunamis, and it does not even occur on the megathrust. Outer-rise earthquakes happen on the incoming oceanic plate just seaward of the trench, where the plate bends downward before diving into the mantle. The bending stretches the top of the plate, reactivating old fractures from when the plate first formed at a mid-ocean ridge. The resulting earthquakes are typically normal-faulting events (the rock pulls apart rather than compresses) and tend to be smaller than megathrust ruptures.
However, they are close to the ocean surface and can still displace the seafloor enough to produce damaging waves locally. Research along the Middle America Trench in Central America found that outer-rise normal faults there could produce coastal wave heights above 2 meters in many locations, with maximum heights reaching about 8 meters. The study noted that while these faults are unlikely to generate ocean-spanning tsunamis, they can produce significant damage near the source.13Tectonophysics. Tsunamigenic potential of outer-rise normal faults at the Middle America trench in Central America This is a blind spot for communities that associate tsunami risk only with great megathrust earthquakes.
Volcanic Tsunamis at Convergent Margins
Subduction zones build volcanic arcs, and those volcanoes occasionally produce tsunamis of their own, through mechanisms entirely distinct from fault slip. The January 2022 eruption of Hunga Tonga-Hunga Ha’apai in the South Pacific was the most dramatic recent example. With a volcanic explosivity index possibly equivalent to VEI 5, it was the largest seaborne eruption in nearly a century and a half, and it produced ocean-wide tsunamis never before documented in the Pacific instrumental record.14Geoscience Letters. Tonga volcanic eruption and tsunami, January 2022: globally the most significant opportunity to observe an explosive and tsunamigenic submarine eruption since AD 1883 Krakatau
Subsequent research showed that the rapid collapse of the submarine caldera during the eruption enhanced the resulting tsunami. The speed and style of caldera formation, rather than the eruption blast alone, played a key role in generating hazardous waves, highlighting a capacity for relatively small but steep submarine caldera collapses to produce dangerous tsunamis.15Nature Geoscience. Rapid submarine caldera collapse during the 2022 climactic eruption of Hunga volcano (Tonga) Volcanically generated tsunamis have been called a “blind spot” in Pacific tsunami hazard planning, and the 2022 event forced a reassessment of how much attention volcanic sources deserve alongside the megathrust earthquakes that dominate traditional risk models.
How Slow Slip Events Complicate the Picture
Between large earthquakes, subduction megathrusts do not simply sit still. Many zones experience slow-slip events, episodes in which parts of the fault creep over days to weeks rather than rupturing in seconds. These events release energy gradually, without generating destructive shaking or tsunamis themselves, but they are closely watched because they may reveal what is happening on the locked portions of the fault.
Seafloor pressure gauges deployed at the Hikurangi subduction margin off New Zealand captured direct evidence of slow-slip events occurring within 2 kilometers of the trench, in the very zone where the largest tsunamis originate.16PubMed. Slow slip near the trench at the Hikurangi subduction zone, New Zealand That observation matters because it shows the shallow megathrust is not uniformly locked. Some patches creep silently while neighboring patches stay stuck, and the interaction between the two may influence when and where the next big rupture begins.
Off the coast of Mexico, seafloor geodesy revealed the first observed shallow slow-slip events in that region. One slow-slip event migrated from the trench toward the eventual earthquake source before a large rupture occurred. Researchers also detected episodic near-trench deformation, which they termed “slab-pull surges,” before three regional earthquakes of magnitude 7 or greater, raising the possibility that these signals could serve as precursors observable at other subduction zones.17PubMed Central. Seafloor geodesy unveils seismogenesis of large subduction earthquakes in Mexico This is still early-stage science, and no one is reliably predicting megathrust earthquakes yet, but the growing network of seafloor instruments is slowly filling in the picture of how these faults behave between their catastrophic moments.
The Cascadia Problem
The Cascadia subduction zone, stretching from northern California to British Columbia, has not produced a great earthquake since January 1700. That event, estimated at roughly magnitude 9, sent a tsunami across the Pacific that was recorded in Japan. For most of the twentieth century, scientists did not realize Cascadia was capable of such events, because the written historical record in the Pacific Northwest was too short to capture one. It was geological evidence, primarily layers of tsunami sand in coastal marshes and drowned forests along the coast, that revealed the threat.
Simulations of earthquakes ranging from magnitude 7.5 to 9.2 on the Cascadia megathrust show that events of magnitude 8.5 or larger initiating in the middle segments of the zone can produce coastal tsunami heights comparable to those from the largest expected rupture. The concave shape of the Pacific Northwest coastline focuses tsunami energy between about 44° and 45° latitude in Oregon, making that stretch of coast disproportionately vulnerable.18Geophysical Research Letters. Relative Tsunami Hazard From Segments of Cascadia Subduction Zone For Mw 7.5–9.2 Earthquakes Splay fault dynamics at this margin add further uncertainty about what the next tsunami might look like.12AGU Advances. Structural Controls on Splay Fault Rupture Dynamics During Cascadia Megathrust Earthquakes
Paleotsunamis and the Long View
Written records only go back so far, and some of the most dangerous subduction zones have recurrence intervals that span centuries. Geological deposits of past tsunamis, preserved as sand sheets in coastal marshes and lagoons, have become an essential tool for understanding how often the largest events occur and how big they get.
Along the Nankai Trough in Japan, tsunami deposit studies have identified nine major Tokai segment earthquakes over the past 1,300 years, with recurrence intervals ranging from 90 to 265 years.19Quaternary Science Reviews. Tsunami deposits refine great earthquake rupture extent and recurrence over the past 1300 years along the Nankai and Tokai fault segments of the Nankai Trough, Japan That variability is itself informative: megathrust earthquakes do not repeat on neat schedules. Sometimes a segment ruptures alone, producing a large but regionally limited event. Other times, ruptures cascade across multiple segments in a single enormous earthquake. The 1707 Hōei event on the Nankai Trough, for instance, ruptured both the Nankai and Tokai segments simultaneously, producing one of the largest earthquakes in Japanese history.
This geological perspective now directly influences public policy. Histories of earthquakes and tsunamis inferred from geological evidence aid in anticipating future catastrophes, and this natural warning system has shaped building codes and tsunami planning in the United States, Canada, and Japan, particularly where geology demonstrates the past occurrence of events larger than those in written or instrumental records.20Annual Review of Earth and Planetary Sciences. Long-Term Perspectives on Giant Earthquakes and Tsunamis at Subduction Zones The Cascadia subduction zone is the clearest example: absent paleotsunami evidence, the region would still be considered low risk.
Improving Warnings With Seafloor Monitoring
Traditional tsunami warning relies heavily on seismometers on land, which estimate earthquake magnitude and location within minutes but cannot directly measure what the seafloor actually did. That gap is why tsunami earthquakes, with their deceptively mild shaking, remain so dangerous. A coastal population that feels only gentle rolling may not realize a major wave is coming.
New approaches aim to close this gap. Seismic array techniques have shown promise for faster and more accurate tsunami forecasting. Testing on the May 2023 magnitude-7.7 Loyalty Islands earthquake demonstrated that integrating array processing with existing warning systems significantly improved forecast accuracy and timeliness, with simulated tsunami signals showing over 70 percent coherence with actual coastal observations.21SpringerLink (Natural Hazards). Improving regional tsunami early warning with seismic array techniques
Meanwhile, seafloor pressure sensors and GPS-acoustic transponders deployed directly on the megathrust are beginning to capture fault behavior in real time. These instruments revealed the slow-slip events and slab-pull surges discussed earlier, and they could eventually feed into operational warning systems that detect not just the earthquake but the actual seafloor displacement that drives a tsunami. For subduction zones like Cascadia, where the coastline sits close to the trench and warning times could be as short as 15 to 20 minutes, every improvement in detection speed matters.