Convergent Plate Boundaries

By the end of this lecture

• Describe subduction-zone anatomy and the Chilean Mariana continuum
• Use age, rate, and sediment to classify a margin — and test the maximum-magnitude hypothesis
• Interpret the depth-varying rupture domains (seismicity, tsunami, strong motion)
• Explain why young, slow Cascadia is nonetheless -capable

Subduction zones host every

Takeaway — Every great earthquake is on a megathrust — but the giants scatter across margins of very different age and rate.

A rule that data dispelled

• Ruff–Kanamori (1980): old + fast ⇒ strong coupling ⇒ great earthquakes
• Sumatra 2004 (old, slow) →
• Tōhoku 2011 (where was thought "impossible") →
• Age and convergence rate do not set the maximum magnitude

What controls how large a subduction earthquake can be — and why were the obvious parameters wrong?

The slab is a cold, dense sinker

• Cold, thick, old lithosphere → negative buoyancy
• Basalt → eclogite transition adds density at depth
• Slab pull is the dominant plate-driving force
• This is why old slabs looked like the big-quake recipe

Two end-member modes

Takeaway — Coupling and dip set the style: Chilean (shallow, coupled, shortening) Mariana (steep, weak, extension).

The thermal parameter

• Combines age , convergence rate , and dip
• A measure of how cold the slab stays as it descends
• Large → deep seismicity, the "strong-coupling" expectation
• Hold onto it — the data will test it

What limits the largest earthquake?

• Maximum magnitude is set by rupture area (), not slab age
• Wide + long + strongly locked ⇒ great earthquake
• The controls are geometric

Depth-varying rupture domains

Takeaway — Shallow domain A → tsunami; downdip domain C → strong ground motion. A margin's hazard depends on which domains it has.

The recipe, tested against data

Takeaway — Age & rate (top) don't order ; seismogenic width () & flatness () do (Wirth 2022).

What actually controls

• Seismogenic width & dip — strongest; needs width km, needs km
• Downdip curvature — flat megathrusts rupture biggest
• Secondary: sediment smoothing, roughness, fluids, upper-plate strain
• Working assumption: any mature megathrust may be

Dip controls the back-arc

Takeaway — Steep dip () → back-arc extension; shallow dip () → upper-plate shortening.

Three margins, three verdicts

• Chile — old-ish, fast, coupled → recipe works ( in 1960)
• Mariana — very old but steep, narrow zone → no great EQ
• Cascadia — young, slow → recipe says "small"... yet (1700)
• One scheme can't fit all three ⇒ the controls were wrong

Cascadia: the -capable exception

Takeaway — Young, warm, slow, sediment-rich, smooth — yet locked, and it ruptured in January 1700.

Zooming out: the convergence spectrum

Takeaway — Buoyancy of the incoming material sets the mode: ocean subducts; arcs & continents collide (Taiwan, Himalaya).

AI literacy — the confident, outdated answer

• Ask an AI: "how do age and rate set the maximum subduction earthquake?"
• It may recite the dispelled 1980 rule — fluently and confidently
• Grade it: Does it cite Sumatra / Tōhoku? Does it give the modern controls?
• You catch this with domain knowledge, not better prompting

Concept check

1. Compute for Tōhoku vs. Mariana — which slab is colder? Which is bigger?
2. Rupture area and length for ( GPa, m, km)?
3. Name two "high-recipe" margins with no great earthquake — why not?
4. Why is Cascadia regarded as -capable? Cite two second-order controls.

Interactive classification: notebooks/subduction_parameter_space.ipynb