27.03.04 · earth-science / earthquakes-volcanoes

Megathrust earthquakes and the seismic cycle: subduction-zone seismicity, Cascadia 1700, and the 2011 Tohoku event

shipped3 tiersLean: none

Anchor (Master): Reid 1910; Kanamori 1977 J. Geophys. Res. 82:2981; Lay-Kanamori-Ruff 1982; Scholz 1998 Nature 391:37; Satake-Atwater 2003; Simons-Minson-Sladen 2011 Science 332:1421; Bletery-Nocquet 2024

Intuition Beginner

Where one tectonic plate dives beneath another, the boundary is not smooth. The two plates are jagged, and friction locks them together. The upper plate is slowly dragged downward and seaward as the lower plate slides beneath at a few centimeters per year. Elastic strain builds in the rocks, like a spring being wound. When the lock finally breaks, the upper plate snaps back: a megathrust earthquake.

A great megathrust shakes violently for five to ten minutes and reaches magnitude 8 to 9. If the rupture is offshore, the seafloor jerks upward by several meters, and that kick launches a tsunami that crosses the Pacific. The four largest earthquakes ever recorded, all megathrusts, are Chile 1960 at magnitude 9.5, Alaska 1964 at 9.2, Sumatra 2004 at 9.1, and Tohoku 2011 at 9.0.

This unit exists because megathrusts produce the most destructive natural hazards on Earth. The 2004 Indian Ocean tsunami killed about 230,000 people across fourteen countries. The 2011 Tohoku tsunami killed about 20,000 and triggered the Fukushima nuclear disaster. Cascadia, off the Pacific Northwest, last ruptured in 1700 and is now locked and accumulating strain: the next event is overdue.

Visual Beginner

Phase	What happens	Timescale
Interseismic locking	Plates stuck; upper plate drags down, storing elastic strain	Centuries
Coseismic rupture	Lock breaks; upper plate snaps seaward and up; meters of slip in seconds	Minutes
Tsunami generation	Seafloor displacement launches long ocean wave	Minutes
Postseismic relaxation	Afterslip and mantle flow continue at slower rates	Years to decades

Worked example Beginner

On January 26, 1700, at about 9 PM local time, a magnitude 9 megathrust ripped the Cascadia subduction zone from northern California to Vancouver Island. The rupture spanned roughly 1,000 kilometers of fault. The shaking lasted several minutes. The Pacific Northwest coast dropped by about a meter and shifted seaward by up to 20 meters. A tsunami surged across the coast within minutes, drowning forests and estuaries.

That tsunami also crossed the Pacific. In Japan, where written records were kept at temples and merchant houses, the wave arrived the next day with no preceding shaking. Japanese diarists called it an "orphan tsunami." Centuries later, geologists matched the orphan-tsunami records to the Cascadia rupture by computing travel times across the Pacific.

Step 1. The Pacific Northwest "ghost forests," stands of dead cedar trees in coastal Washington and Oregon, record sudden land subsidence. Tree-ring dating of the dead cedars puts their death in the year 1700.

Step 2. Sand layers in coastal tidal marshes, washed inland by the tsunami, give the same date through radiocarbon dating of plant fragments.

Step 3. The Japanese temple records pin the date precisely to January 26, 1700, and the wave height constrains the rupture magnitude to about M 9.

What this tells us: a Cascadia megathrust ruptures roughly every 500 years on average. The last event was 1700, so the next is overdue by decades. Average recurrence is a statistical guide, not a clock.

Check your understanding Beginner

Exercise 1 (easy, multiple choice).

Where do the largest earthquakes on Earth occur?

A. At mid-ocean ridges, where plates pull apart

B. At transform faults, where plates slide horizontally

C. At subduction-zone megathrusts, where one plate thrusts beneath another

D. In continental interiors, far from any plate boundary

Hint

Recall the four largest earthquakes ever recorded: Chile 1960, Alaska 1964, Sumatra 2004, Tohoku 2011. What tectonic setting did they share?

Answer

Option C. All four of the largest instrumentally recorded earthquakes were megathrust events at subduction zones. The long, shallow, gently dipping plate interface can lock over a vast area, accumulate elastic strain for centuries, and release it in a single M 9 rupture. No other tectonic setting provides a fault patch large enough.

Formal definition Intermediate+

A megathrust earthquake is a great earthquake (moment magnitude $M_{w} \geq 8$ ) occurring on the low-angle reverse fault that forms the interplate contact at a subduction zone. The fault, called the megathrust interface, dips at roughly $3 0^{\circ}$ to $6 0^{\circ}$ from the trench down to about $50 km$ depth, where the contact transitions from brittle, seismogenic friction to ductile, aseismic creep.

The seismic cycle

The megathrust interface cycles through three phases ^{[Reid 1910]}:

Interseismic locking. The interface is stuck (locked). Plate convergence at rate $v$ accrues a slip deficit, the difference between the convergence that would occur if the interface slid freely and the actual slip. Over centuries, the upper plate is dragged downward and seaward, building elastic strain at rate $v$ .
Coseismic rupture. When accumulated stress exceeds frictional strength, the locked patch slips meters to tens of meters in seconds to minutes. Elastic strain built over centuries is released as seismic waves.
Postseismic relaxation. Afterslip on velocity-strengthening patches surrounding the rupture and viscoelastic flow in the mantle below continue the displacement at decaying rates over years to decades.

Seismic moment and moment magnitude

The seismic moment $M_{0}$ is the physical measure of earthquake size, defined as

M_{0} = μ \cdot A \cdot D,

where $μ$ is the shear modulus (rigidity) of the rock around the fault, $A$ is the fault area that slipped, and $D$ is the average slip on that area. For a subduction-zone megathrust, $μ \approx 3 \times 1 0^{10} N m^{- 2}$ , $A$ is of order $1 0^{11} m^{2}$ , and $D$ is of order $1 0^{1} m$ .

The moment magnitude $M_{w}$ , defined by Kanamori in 1977 ^{[Kanamori 1977]}, is

M_{w} = \frac{2}{3} lo g_{10} (M_{0}) - 10.7,

with $M_{0}$ expressed in dyne-centimeters. The logarithmic scale compresses a nine-order-of-magnitude range of moments onto a 0 to 10 magnitude scale and does not saturate for great earthquakes, in contrast to the earlier Richter scale.

Frictional physics and the stability transition

The frictional behavior of the megathrust is captured by rate-and-state friction laws (Dieterich 1979; Ruina 1983; Scholz 1998 ^{[Scholz 1998]}). The coefficient of friction $μ_{f}$ depends on the sliding velocity $V$ and an internal state variable $θ$ :

μ_{f} (V, θ) = μ_{0} + a ln (V / V_{0}) + b ln (θ / θ_{0}) .

The sign of $a - b$ determines stability. Where $a - b < 0$ (**velocity-weakening**), slip is unstable and earthquakes nucleate; this is the seismogenic zone. Where $a - b > 0$ (velocity-strengthening), slip is stable and aseismic; this is the downdip creeping zone. The transition depth, around $30$ to $50 km$ at most subduction zones, sets the downdip width of the seismogenic fault and hence the maximum plausible rupture area.

Tsunami generation and propagation

Vertical seafloor displacement $u_{z}$ above the rupture area displaces the entire water column, launching a long-wavelength wave governed by the linear shallow-water equations. The phase speed is

c = g h,

where $h$ is water depth. At $h = 4000 m$ , $c \approx 198 m/s \approx 710 km/h$ . On approaching shore, $h$ decreases, $c$ drops, and the wave amplitude grows to conserve energy flux, producing run-up heights of tens of meters.

Asperity model

Lay, Kanamori, and Ruff in 1982 ^{[Lay-Kanamori-Ruff 1982]} introduced the asperity model: a megathrust interface contains strongly coupled patches (asperities) surrounded by weakly coupled regions. Rupture of one asperity yields a $M_{w} 7$ to $8$ earthquake; simultaneous rupture of several asperities cascades into a $M_{w} 9$ great earthquake. The 2011 Tohoku event ruptured previously identified asperities plus a deeper slow-slip patch, generating far more slip than anticipated.

Counterexamples to common slips Intermediate+

"Magnitude 9 is impossible on this subduction zone." The maximum magnitude scales with the seismogenic fault area, which is set by subduction-zone geometry (trench length, slab dip, seismogenic depth range). Geometry, not recent history, sets the ceiling. Before 2004, Sumatra was not considered a $M_{w} 9$ candidate; before 2011, neither was Tohoku.
"Earthquake prediction works." It does not. Operational forecasts are probabilistic (Cascadia has a 10 to 14 percent chance of a $M_{w} 8$ + event in the next 50 years), not deterministic. No published method has reliably predicted the time, place, and magnitude of a damaging earthquake ahead of the event.
"The 2004 Indian Ocean tsunami warning system was adequate." It was not. No Indian Ocean tsunami warning system existed in December 2004. The Pacific Tsunami Warning Center detected the earthquake but had no protocol to notify Indian Ocean nations. A functioning system would have saved tens of thousands of lives.
"Cascadia is dormant." It is locked and accumulating slip deficit at roughly $4 cm/yr$ . The locked zone extends from northern California to Vancouver Island, and geodetic measurements show no measurable contemporary creep on the interface.
"Slow slip events are small earthquakes." Slow slip events (SSEs) release seismic moment over days to months, not seconds, producing no damaging shaking. Some SSEs release moment equivalent to a $M_{w} 6.5$ to $7$ earthquake but are detectable only by GPS and strainmeters.

Key theorem with derivation Intermediate+

Theorem (the $M_{w} 9$ ceiling for single-event megathrust earthquakes). Let $μ \approx 3 \times 1 0^{10} N m^{- 2}$ be the typical shear modulus of the upper mantle, let $A_{m a x} \approx 3 \times 1 0^{11} m^{2}$ be the maximum plausible coseismic rupture area (corresponding to a $1500 km$ along-strike rupture on a $200 km$ wide seismogenic fault), and let $D_{m a x} \approx 50 m$ be the maximum plausible average slip (corresponding to a few centuries of accumulated slip deficit at typical convergence rates). Then the moment magnitude of any single megathrust event satisfies

M_{w} \leq \frac{2}{3} lo g_{10} (μ A_{m a x} D_{m a x}) - 10.7 \approx 9.7,

with $M_{0}$ in dyne-centimeters, and in practice no known subduction zone achieves both maxima simultaneously, so the historical ceiling is $M_{w} \approx 9.5$ .

Proof. We work in SI units and convert at the end. By the definition of seismic moment,

M_{0} = μ \cdot A \cdot D .

Substituting the upper bounds,

M_{0} \leq (3 \times 1 0^{10} N m^{- 2}) \cdot (3 \times 1 0^{11} m^{2}) \cdot (50 m) = 4.5 \times 1 0^{23} N m .

Converting to dyne-centimeters, $1 N m = 1 0^{7} dyne cm$ , so

M_{0} \leq 4.5 \times 1 0^{30} dyne cm .

Now apply the moment magnitude definition $M_{w} = \frac{2}{3} lo g_{10} (M_{0}) - 10.7$ :

M_{w} \leq \frac{2}{3} lo g_{10} (4.5 \times 1 0^{30}) - 10.7 = \frac{2}{3} (30 + lo g_{10} 4.5) - 10.7.

Since $lo g_{10} 4.5 \approx 0.653$ ,

M_{w} \leq \frac{2}{3} (30.653) - 10.7 \approx 20.44 - 10.7 = 9.73.

So the absolute single-event ceiling is $M_{w} \approx 9.7$ . To check the historical record against this bound, take Chile 1960 ( $M_{w} 9.5$ ): its published seismic moment is $M_{0} \approx 2 \times 1 0^{23} N m$ , corresponding to $A \approx 800 \times 200 km$ and $D \approx 20 m$ , both below the maxima used above, and yielding $M_{w} = \frac{2}{3} lo g_{10} (2 \times 1 0^{30}) - 10.7 \approx 9.5$ , matching the observed value. The same computation for Sumatra 2004 ( $A \approx 1300 \times 150 km$ , $D \approx 15 m$ ) gives $M_{w} \approx 9.1$ , and for Tohoku 2011 ( $A \approx 500 \times 200 km$ , $D \approx 20 m$ ) gives $M_{w} \approx 9.0$ . $□$

The ceiling reflects three independent physical limits. First, the along-strike length is bounded by the length of the subduction segment before a tear, ridge, or bend in the incoming plate breaks the interface. Second, the downdip width is bounded by the seismogenic depth range where rate-and-state friction is velocity-weakening. Third, the slip is bounded by the elastic strain that the upper plate can store before failing, set by the convergence rate times the inter-event time. The Lay-Kanamori-Ruff asperity framework explains why most subduction-zone earthquakes tap only a small fraction of these maxima: typical ruptures break one or two asperities rather than the entire margin.

Bridge. The $M_{w} 9$ ceiling builds toward 27.03.01 earthquake statistics, where the Gutenberg-Richter law predicts a tenfold decrease in event frequency per magnitude unit, so $M_{w} 9$ events occur roughly once per decade globally. The argument appears again in 27.01.01 plate tectonics, where the convergence rate at each subduction zone, multiplied by the maximum seismogenic area, sets the regional ceiling. The foundational reason no $M_{w} 10$ earthquake is expected is that no known subduction segment is long enough and wide enough and locked enough to host one; this is exactly why Chile 1960, the largest instrumentally recorded event, falls near the practical ceiling.

Exercises Intermediate+

Exercise 3 (medium, multiple choice).

A rate-and-state friction law gives the friction coefficient as $μ_{f} = μ_{0} + a ln (V / V_{0}) + b ln (θ / θ_{0})$ . Which statement about the stability of slip on a fault patch is correct?

A. Patches with $a - b > 0$ are velocity-weakening and produce earthquakes

B. Patches with $a - b < 0$ are velocity-strengthening and creep aseismically

C. Patches with $a - b < 0$ are velocity-weakening and host earthquake nucleation

D. The sign of $a - b$ has no observable consequence

Hint

Velocity-weakening means friction drops as slip accelerates, which is the condition for unstable, runaway slip.

Answer

Option C. The rate-and-state parameter combination $a - b$ governs frictional stability. Where $a - b < 0$ , friction decreases as sliding velocity increases, producing runaway slip: an earthquake nucleates. Where $a - b > 0$ , friction increases with velocity and slip is stable, producing aseismic creep. The downdip transition from $a - b < 0$ to $a - b > 0$ sets the downdip edge of the seismogenic zone.

Exercise 6 (medium, short answer).

The Lay-Kanamori-Ruff asperity model represents a megathrust as a set of strongly coupled patches (asperities) separated by weakly coupled regions. Use this model to explain why a single subduction zone can produce both $M_{w} 7$ to $8$ earthquakes every few decades and a $M_{w} 9$ earthquake every few centuries.

Hint

Think about what changes when only one asperity ruptures versus when several rupture simultaneously in a cascade.

Answer

Under the asperity model, each isolated asperity rupture releases the moment of a single patch, of order $1 0^{19}$ to $1 0^{20} N m$ , corresponding to $M_{w} 7$ to $8$ . These events recur on the loading timescale of one asperity, typically decades. A great $M_{w} 9$ event requires a cascade: rupture of one asperity dynamically triggers the next, breaking several asperities and the regions between them in a single, long rupture that releases $1 0^{22}$ to $1 0^{23} N m$ . Because the cascade requires that stress has built up on all asperities simultaneously and that rupture dynamics overcome the barriers between them, it occurs far less frequently, on the several-century timescale of full-margin loading. The same fault thus produces a Gutenberg-Richter-like distribution of event sizes from a small number of geometric asperities.

Exercise 7 (hard, short answer).

Before 2011, the Japanese seismic hazard model assigned a low probability to a $M_{w} 9$ event on the Tohoku (Japan Trench) segment, citing the historical record of $M_{w} 7$ to $8$ earthquakes on identified asperities. Describe the specific flaw in this hazard assessment that the 2011 Tohoku event exposed, and explain how an asperity-model perspective could have flagged the underestimation.

Hint

Consider the difference between the historical catalog of individual asperity ruptures and the possibility of a multi-asperity cascade. What does the historical record miss?

Answer

The pre-2011 Japanese hazard model used the historical catalog of recorded earthquakes on the Japan Trench, which contained many $M_{w} 7$ to $8$ events on individual asperities but no instrumental record of a full-margin $M_{w} 9$ event in the past century. The model extrapolated future hazard from this catalog, treating each asperity as independent. The 2011 rupture exposed two flaws. First, the historical catalog of instrumental seismology spans roughly a century, much shorter than the recurrence interval of a $M_{w} 9$ cascade (typically 500 to 1000 years), so absence in the catalog does not imply absence on the segment. Second, paleoseismic and tsunami-deposit studies had identified a previous Tohoku-rupture event in AD 869 (the Jogan earthquake), indicating that the segment was capable of multi-asperity cascades; this evidence was downweighted. From an asperity-model perspective, the presence of multiple adjacent strongly coupled patches on the Japan Trench was the structural warning that a cascade could not be ruled out, regardless of the short instrumental catalog.

Exercise 8 (hard, numeric).

Using the moment magnitude ceiling argument, what is the maximum moment magnitude of a single megathrust event on a fault with along-strike length $L = 1000 km$ , downdip width $W = 150 km$ , and maximum average slip $D = 30 m$ ? Take $μ = 3 \times 1 0^{10} N m^{- 2}$ . Give $M_{w}$ to two decimal places.

Hint

Compute $A = L \cdot W$ , then $M_{0} = μ A D$ , then convert to dyne-centimeters and apply $M_{w} = \frac{2}{3} lo g_{10} M_{0} - 10.7$ .

Answer

$M_{w} \approx 9.29$ . First compute the area: $A = 1000 km \times 150 km = 1 0^{6} m \times 1.5 \times 1 0^{5} m = 1.5 \times 1 0^{11} m^{2}$ . Then $M_{0} = 3 \times 1 0^{10} \times 1.5 \times 1 0^{11} \times 30 = 1.35 \times 1 0^{23} N m$ . Converting to dyne-centimeters, $M_{0} = 1.35 \times 1 0^{30} dyne cm$ . Then $M_{w} = \frac{2}{3} lo g_{10} (1.35 \times 1 0^{30}) - 10.7 = \frac{2}{3} (30 + 0.13) - 10.7 = \frac{2}{3} (30.13) - 10.7 \approx 20.09 - 10.7 = 9.39$ . This falls just below the absolute ceiling of about $9.7$ and is comparable to the largest historically observed events, illustrating that even a moderately large subduction segment with substantial slip approaches the practical ceiling.

Advanced results Master

Reid's elastic rebound theory

Harry Fielding Reid, studying survey triangulation across the San Andreas fault before and after the 1906 San Francisco earthquake, formulated elastic rebound in 1910 ^{[Reid 1910]}. The crust deforms elastically under slow tectonic loading; an earthquake is the sudden release of the stored elastic strain when the fault slips. Reid's five observations remain the load-bearing empirical foundation of earthquake mechanics: the rupture released strain accumulated over decades; the displacement was discontinuous across the fault; the strain field far from the fault was unchanged; the elastic modulus of the rock was unchanged; and the rupture propagated at finite speed.

Kanamori's moment magnitude and the energy-moment relation

Hiroo Kanamori in 1977 ^{[Kanamori 1977]} introduced the moment magnitude scale $M_{w} = \frac{2}{3} lo g_{10} (M_{0}) - 10.7$ by relating seismic moment to radiated energy through the apparent stress. The scale does not saturate at $M_{w} 8$ as the surface-wave magnitude $M_{s}$ does, because it is built on the static physical quantity $M_{0} = μ A D$ rather than the amplitude of a specific phase. Every great earthquake recorded since 1900, when ranked by $M_{w}$ , falls within the predictions of the moment-magnitude ceiling argument: Chile 1960 at $9.5$ , Alaska 1964 at $9.2$ , Sumatra 2004 at $9.1$ , Tohoku 2011 at $9.0$ .

The Lay-Kanamori-Ruff asperity model

Lay, Kanamori, and Ruff in 1982 ^{[Lay-Kanamori-Ruff 1982]} proposed that subduction-zone megathrusts host discrete strongly coupled asperities. Each asperity has a characteristic size and rupture recurrence, and great earthquakes occur when rupture cascades across multiple asperities. The model explains the observed Gutenberg-Richter-like distribution of magnitudes on a single subduction zone, the variability of rupture extent from event to event on the same margin, and the apparent segmentation of subduction boundaries. The 2011 Tohoku rupture, which propagated across asperities previously considered independent, confirmed the cascade mechanism.

Dieterich, Ruina, and Scholz: rate-and-state friction

James Dieterich in 1979 and Andy Ruina in 1983 developed the rate-and-state friction law $μ_{f} (V, θ) = μ_{0} + a ln (V / V_{0}) + b ln (θ / θ_{0})$ , in which friction depends on sliding velocity $V$ and a state variable $θ$ that evolves with slip and contact time. Christopher Scholz in 1998 ^{[Scholz 1998]} synthesized this framework into a theory of earthquake nucleation. The sign of $a - b$ partitions the fault into velocity-weakening (seismogenic, unstable) and velocity-strengthening (aseismic, stable) patches; earthquake nucleation occurs on the velocity-weakening asperities; the downdip edge of the seismogenic zone is the depth at which $a - b$ changes sign, typically $30$ to $50 km$ . Slow slip events occur on the transition region, where $a - b \approx 0$ .

Cascadia 1700 reconstruction: Atwater, Yamaguchi, and Satake

Brian Atwater and Eiji Yamaguchi in 1991 identified the Pacific Northwest ghost forests as evidence of sudden coastal subsidence in 1700. Kenji Satake, Kelin Wang, and Atwater in 2003 ^{[Satake-Atwater 2003]} inverted Japanese orphan-tsunami records for the source. Combining the tsunami arrival time in Japan with the wave heights recorded at six coastal sites, they constrained the January 26, 1700 rupture to $M_{w} \approx 9.0$ , with rupture length about $1100 km$ and average slip about $20 m$ . The reconstruction is the canonical example of merging paleoseismology, dendrochronology, and historical written records across the Pacific.

Simons and the Tohoku early science

Mark Simons and collaborators in 2011 ^{[Simons-Minson-Sladen 2011]} used GPS-coseismic displacements, seafloor geodesy, and teleseismic waveform inversion to estimate the Tohoku rupture. They found peak slip exceeding $50 m$ on the shallow part of the megathrust near the trench, far exceeding pre-event hazard estimates. The shallow slip, previously thought unlikely on a supposedly creeping trench interface, was the principal driver of the large tsunami. The result prompted a global re-evaluation of shallow megathrust frictional properties.

Bletery and Nocquet on the modern locked zone

Quentin Bletery and Jean-Mathieu Nocquet in 2024 ^{[Bletery-Nocquet 2024]} reanalyzed the Cascadia locked zone using two decades of GNSS observations and a fully three-dimensional dislocation model. They found the interface is weakly locked, with a small effective friction coefficient, and that the seismogenic patch is narrower and more segmented than earlier models assumed. The result reframes the Cascadia hazard: a full-margin $M_{w} 9$ cascade remains possible, but a sequence of $M_{w} 8$ to $8.5$ asperity ruptures may be more probable than previously estimated.

Synthesis. The seismic-cycle picture builds toward a unified mechanical theory of subduction-zone seismicity in which Reid's elastic rebound, Kanamori's moment, the Lay-Kanamori-Ruff asperity geometry, and the Dieterich-Ruina-Scholz friction law are layered, not alternative, descriptions. The foundational reason great megathrusts reach the $M_{w} 9$ ceiling is that the seismogenic interface, locked for centuries, accumulates a slip deficit equal to the convergence rate times the recurrence time; this is exactly the content of the moment ceiling argument. The central insight is that putting these together, the same frictional physics explains both the centuries-long interseismic lock and the seconds-long coseismic rupture, with the stability transition at depth controlling the maximum rupture width. The pattern appears again in 27.01.01 plate tectonics, where the global distribution of $M_{w} 9$ events traces the longest, fastest-converging subduction zones, and generalises to every convergent margin on Earth and to the slow-slip and tremor phenomena at the downdip transition; the bridge is the recognition that every phase of the cycle is governed by one rate-and-state friction law evaluated at different slip velocities.

Full proof set Master

Proposition (interseismic slip-deficit accumulation). Let $v$ be the plate-convergence rate at a subduction zone, and let $T$ be the recurrence interval between great megathrust events on a given asperity. Then the average coseismic slip $D$ on that asperity, averaged over many cycles, satisfies $D \approx v \cdot T$ .

Proof. Define the slip deficit $S (t)$ as the cumulative plate convergence at time $t$ since the last coseismic event, minus the cumulative aseismic slip on the asperity over the same interval. By definition of locked, the aseismic slip on the asperity is negligible during the interseismic period, so $d S / d t = v$ . Integrating from $0$ to $T$ gives $S (T) = v \cdot T$ . At the coseismic event, elastic strain accumulated from the slip deficit $S (T)$ is released as average slip $D$ on the asperity. Over many cycles, by conservation of moment release on the asperity, the average $D$ equals the average $S (T)$ , giving $⟨ D ⟩ = v \cdot ⟨ T ⟩$ . For Cascadia with $v \approx 4 cm/yr$ and $⟨ T ⟩ \approx 500 yr$ , this gives $⟨ D ⟩ \approx 20 m$ , consistent with the Satake-Atwater reconstruction of the 1700 event. $□$

Proposition (tsunami phase speed in the shallow-water limit). For a long-wavelength surface gravity wave in an ocean of depth $h$ , with wavelength $λ ≫ h$ , the phase speed is $c = g h$ .

Proof. In the linear, inviscid, shallow-water equations, the pressure is hydrostatic, $p = ρ g (h + η)$ , where $η (x, t)$ is the surface elevation perturbation. Linearizing about the rest state $η = 0$ , $v = 0$ , the equations of motion reduce to

\frac{\partial η}{\partial t} + h \nabla \cdot v = 0, \frac{\partial v}{\partial t} = - g \nabla η,

where $v$ is the depth-averaged horizontal velocity. Taking the time derivative of the first equation and substituting the divergence of the second,

\frac{\partial ^{2} η}{\partial t ^{2}} - g h \nabla^{2} η = 0,

which is the wave equation with phase speed $c = g h$ , independent of wavelength in the shallow-water limit. Substituting $g = 9.8 m s^{- 2}$ and $h = 4000 m$ gives $c \approx 198 m s^{- 1} \approx 710 km h^{- 1}$ , the deep-ocean tsunami speed observed following the 2004 Sumatra and 2011 Tohoku events. $□$

Proposition (moment-magnitude ratio formula). The ratio of seismic moments of two earthquakes whose moment magnitudes differ by $Δ M_{w}$ is $M_{0, 1} / M_{0, 2} = 1 0^{\frac{3}{2} Δ M_{w}}$ .

Proof. From the definition $M_{w} = \frac{2}{3} lo g_{10} (M_{0}) - 10.7$ , subtract the two equations for events 1 and 2:

Δ M_{w} = M_{w, 1} - M_{w, 2} = \frac{2}{3} (lo g_{10} M_{0, 1} - lo g_{10} M_{0, 2}) = \frac{2}{3} lo g_{10} (M_{0, 1} / M_{0, 2}) .

Solving for the ratio,

lo g_{10} (M_{0, 1} / M_{0, 2}) = \frac{3}{2} Δ M_{w}, M_{0, 1} / M_{0, 2} = 1 0^{\frac{3}{2} Δ M_{w}} .

A unit increase in $M_{w}$ corresponds to a factor $1 0^{3/2} \approx 31.6$ in seismic moment. Hence Chile 1960 ( $M_{w} 9.5$ ) released about $1 0^{3/2 \cdot 0.5} \approx 5.6$ times the moment of an $M_{w} 9.0$ Tohoku event. $□$

Connections Master

Plate tectonics and continental drift 27.01.01. The megathrust interface exists only at convergent plate boundaries, where one plate subducts beneath another at a few centimeters per year. The convergence rate, slab dip, and trench length of each subduction zone are the geometric inputs that set the seismic-cycle period and the maximum plausible rupture area, so the moment-magnitude ceiling derived in this unit descends directly from the plate-tectonic geometry established in the survey unit.
Earthquakes, volcanoes, and geologic hazards 27.03.01. The chapter survey introduces seismic moment, the Gutenberg-Richter law, and the basic fault taxonomy. This unit deepens those foundations: $M_{0} = μ A D$ becomes the quantitative engine of the magnitude ceiling, the Gutenberg-Richter distribution emerges from the asperity cascade model, and the convergent-boundary thrust fault becomes the specific geometry hosting the largest events on Earth.
Stratospheric ozone depletion 27.07.05. Both units treat global-scale geophysical phenomena reconstructed from sparse, distributed observations: ozone depletion from Dobson spectrometers and satellite overpasses, Cascadia 1700 from Japanese orphan-tsunami records and Pacific Northwest ghost forests. Each couples a small, well-understood physical mechanism (catalytic chlorine chemistry; coseismic elastic rebound) to large-scale transport (stratospheric vortex dynamics; plate convergence and tsunami propagation), and each is governed by an international policy framework (Montreal Protocol; Pacific tsunami warning system).
Adaptive optics and interferometry 28.09.02 pending. Modern megathrust observation relies on the same high-precision geodetic and remote-sensing instrumentation as observational astronomy: GNSS displacement time series, satellite radar interferometry, and seafloor pressure gauges all exploit long-baseline interferometric techniques developed for telescopes. The Cascadia locked zone and the modern Tohoku slip inversion would not have been possible without the geodetic precision that descends from the interferometric and adaptive-optics methods catalogued in that unit.

Historical & philosophical context Master

Harry Fielding Reid, a professor at Johns Hopkins, was appointed to the State Earthquake Investigation Commission after the April 18, 1906 San Francisco earthquake. His 1910 report, The Mechanics of the Earthquake ^{[Reid 1910]}, compared triangulation surveys of the San Andreas fault zone conducted in the 1850s and 1880s with a post-earthquake resurvey, finding that the crust had deformed elastically before the rupture and snapped back during it. Reid inferred that the fault had failed when the accumulated elastic strain exceeded the frictional strength of the rock, and that strain had built up at the slow rate of relative plate motion. This elastic-rebound picture remains the conceptual basis of every modern earthquake model, including the seismic cycle formalized in this unit.

Hiroo Kanamori in 1977 ^{[Kanamori 1977]} introduced the moment magnitude $M_{w}$ to address the saturation of surface-wave magnitudes at about $M_{s} 8.2$ . By anchoring the scale to the seismic moment $M_{0} = μ A D$ , a static physical quantity that does not saturate, Kanamori made it possible to compare earthquakes across the entire observed range, from microseismic events to great megathrusts. The Lay-Kanamori-Ruff asperity framework ^{[Lay-Kanamori-Ruff 1982]} followed, organizing subduction-zone seismicity into discrete strongly coupled patches.

The Cascadia 1700 reconstruction is a high point of multi-method geophysical inference. Brian Atwater's 1987 and 1991 fieldwork in Pacific Northwest estuaries identified sudden coastal subsidence dated to 1700, and dendrochronology of ghost-forest cedars pinned the death year. Kenji Satake's 2003 inversion of Japanese temple and merchant records of the orphan tsunami inverted for the source, recovering $M_{w} \approx 9.0$ and an $1100 km$ rupture. The 2024 Bletery-Nocquet reanalysis of the locked zone ^{[Bletery-Nocquet 2024]} extends this lineage: modern GNSS geodesy now resolves the contemporary accumulation of the slip deficit that will drive the next Cascadia event.

Bibliography Master

@book{Reid1910,
  author = {Reid, Harry Fielding},
  title = {The Mechanics of the Earthquake},
  series = {The California Earthquake of April 18, 1906},
  volume = {2},
  publisher = {Carnegie Institution of Washington},
  year = {1910},
}

@article{Kanamori1977,
  author = {Kanamori, Hiroo},
  title = {The energy release in great earthquakes},
  journal = {Journal of Geophysical Research},
  volume = {82},
  number = {20},
  pages = {2981--2987},
  year = {1977},
}

@article{LayKanamoriRuff1982,
  author = {Lay, Thorne and Kanamori, Hiroo and Ruff, Larry},
  title = {The asperity model and the nature of large subduction zone earthquakes},
  journal = {Earth-Science Reviews},
  volume = {18},
  number = {3},
  pages = {1--71},
  year = {1982},
}

@article{Scholz1998,
  author = {Scholz, Christopher H.},
  title = {Earthquakes and friction laws},
  journal = {Nature},
  volume = {391},
  pages = {37--42},
  year = {1998},
}

@article{SatakeAtwater2003,
  author = {Satake, Kenji and Wang, Kelin and Atwater, Brian F.},
  title = {Fault slip and seismic moment of the 1700 Cascadia earthquake inferred from Japanese tsunami descriptions},
  journal = {Journal of Geophysical Research: Solid Earth},
  volume = {108},
  number = {B11},
  pages = {2535},
  year = {2003},
}

@article{SimonsMinsonSladen2011,
  author = {Simons, Mark and Minson, Sarah E. and Sladen, Anthony and Ortega, Francisco and Jiang, Junle and Owen, Susan E. and Meng, Lingsen and Ampuero, Jean-Paul and Wei, Shengji and Chu, Risheng and Helmberger, Donald V. and Kanamori, Hiroo and Hetland, Eric and Moore, Gareth A. and Webb, Frank H.},
  title = {The 2011 magnitude 9.0 Tohoku-Oki earthquake: rupturing the boundary between the {Pacific} and {North American} plates},
  journal = {Science},
  volume = {332},
  number = {6036},
  pages = {1421--1425},
  year = {2011},
}

@article{BleteryNocquet2024,
  author = {Bletery, Quentin and Nocquet, Jean-Mathieu},
  title = {The Cascadia subduction zone is weak and weakly locked},
  journal = {Nature},
  volume = {627},
  pages = {552--557},
  year = {2024},
}

@article{Dieterich1979,
  author = {Dieterich, James H.},
  title = {Modeling of rock friction: 1. Experimental results and constitutive equations},
  journal = {Journal of Geophysical Research: Solid Earth},
  volume = {84},
  number = {B5},
  pages = {2161--2168},
  year = {1979},
}

@article{Ruina1983,
  author = {Ruina, Andy},
  title = {Slip instability and state variable friction laws},
  journal = {Journal of Geophysical Research: Solid Earth},
  volume = {88},
  number = {B12},
  pages = {10359--10370},
  year = {1983},
}

@article{AtwaterYamaguchi1991,
  author = {Atwater, Brian F. and Yamaguchi, Eiji},
  title = {Sudden, probably coseismic submergence of Holocene trees and grass in coastal Washington State},
  journal = {Geology},
  volume = {19},
  number = {7},
  pages = {706--709},
  year = {1991},
}

@book{Scholz2002,
  author = {Scholz, Christopher H.},
  title = {The Mechanics of Earthquakes and Faulting},
  edition = {2},
  publisher = {Cambridge University Press},
  year = {2002},
}

@book{SteinWysession2009,
  author = {Stein, Seth and Wysession, Michael},
  title = {An Introduction to Seismology, Earthquakes, and Earth Structure},
  edition = {2},
  publisher = {Wiley-Blackwell},
  year = {2009},
}

Prerequisites

27.03.01

Tier anchors

beginner: Tarbuck and Lutgens, Earth Science (15e), Ch. 5; USGS earthquake basics; Schulz, 'The Really Big One' (New Yorker, 2015)
intermediate: Stein and Wysession, An Introduction to Seismology (2e); Lay and Wallace, Modern Global Seismology; Scholz, The Mechanics of Earthquakes and Faulting (3e)
master: Reid 1910; Kanamori 1977 J. Geophys. Res. 82:2981; Lay-Kanamori-Ruff 1982; Scholz 1998 Nature 391:37; Satake-Atwater 2003; Simons-Minson-Sladen 2011 Science 332:1421; Bletery-Nocquet 2024

References

Reid, H. F. — The Mechanics of the Earthquake, Vol. 2 of The California Earthquake of April 18, 1906 (Carnegie Institution, 1910) · Full report; elastic rebound theory
Kanamori, H. — The energy release in great earthquakes, J. Geophys. Res. 82, 2981-2987 (1977) · Full text; moment magnitude Mw defined
Lay, T., Kanamori, H. and Ruff, L. — The asperity model and the nature of large subduction zone earthquakes, Earth-Science Reviews 18, 1-71 (1982) · Full text; asperity model
Scholz, C. H. — Earthquakes and friction laws, Nature 391, 37-42 (1998) · Full text; rate-and-state friction and nucleation
Satake, K., Wang, K. and Atwater, B. F. — Fault slip and seismic moment of the 1700 Cascadia earthquake inferred from Japanese tsunami descriptions, J. Geophys. Res. 108, 2535 (2003) · Full text; Cascadia 1700 reconstruction
Simons, M., Minson, S. E., Sladen, A. et al. — The 2011 magnitude 9.0 Tohoku-Oki earthquake: rupturing the boundary between the Pacific and North American plates, Science 332, 1421-1425 (2011) · Full text; Tohoku early science
Bletery, Q. and Nocquet, J.-M. — The Cascadia subduction zone is weak and weakly locked, Nature 627, 552-557 (2024) · Full text; modern locked-zone update

Estimated time

beginner: 25m
intermediate: 50m
master: 85m