Why spend on tsunami towers when Cascadia last broke in 1700?
The last full-margin Cascadia earthquake was January 26, 1700. Tree rings, buried coastal forests, and Japanese records of an “orphan tsunami” pin the date. Magnitude was about 8.7 to 9.2. That is 326 years ago. The question is fair: why put tens of millions into towers, hospital refuges, and seafloor instruments for an event no one living has seen?
The geology does not treat 1700 as a one-off.
Paleoseismic records—onshore subsidence, tsunami sand sheets, and offshore turbidites—show at least 19 great (M ≥ 8) ruptures along most of the Cascadia coast in the last ~10,000 years. Full-margin events average about 500 years apart. Individual gaps range from a couple of centuries to nearly a thousand. Southern segments break more often. “Average 500 years” is not a schedule. It is a long, messy record.
USGS hazard models put the chance of another M~9 in the next 50 years at about 10 percent if you ignore how long it has been since 1700, and about 15 percent if you use a time-dependent model that counts the 326 years already elapsed. Those are not “tomorrow” numbers. They are also not zero. The plates have not stopped. Seafloor GNSS-A sites off Oregon and Washington show the shallow megathrust is nearly locked. Strain is still stacking. When it lets go, a local tsunami can reach Westport, Ocean Shores, Seaside, or Astoria in 10 to 20 minutes. Distant Pacific warning systems do not help that timeline. High ground is too far for many people on the spit and in the marina.
That is the case for the current projects, not a claim that the next rupture is due on a calendar.
Westport breaks ground October 3 on a vertical evacuation tower in the marina district. Capacity is at least 1,000 people. Construction target is summer 2027. Project cost is about $13.7 million, mostly a federal grant plus city and state match. It is the third such structure on the Washington coast after Ocosta Elementary and the Shoalwater Bay tower in Tokeland.
In Astoria, Columbia Memorial Hospital just unlocked $14.57 million in FEMA funds for a resilient expansion that includes a refuge for about 1,900 people, plus critical systems placed above inundation. The hospital itself is a much larger rebuild; the federal slice pays for the foundation and vertical refuge that keep a coastal hospital usable after a Cascadia event.
Science spending is smaller and different. USGS is adding seafloor GNSS-A sites so locking and strain are measured on the fault itself instead of inferred from land GPS. UW and partners have more than $10 million for a Cascadia underwater observatory tied to the existing cabled array—seismometers and pressure gauges that can feed early warning and map how the plate is loading. Oregon issued 2026 guidance for vertical evacuation structures because most of its coast still has no refuge and natural high ground is out of reach in the first 15 minutes.
None of this proves the next M9 is imminent. Recurrence is irregular. Partial southern ruptures are more common than full-margin ones. Money can be wasted on poorly sited towers or grant theater. The honest argument for the spend is narrower: if the locked plate does rupture, minutes matter more than centuries, and 1700 is evidence the plate does rupture—not proof that it is finished.