Which Earthquake Cracked This Foundation? A Forensic Workflow for Unknown-Date Seismic Damage

By Enrique Lairet, PE · Oasis Engineering · Forensic engineering notes

Every forensic engineer who works older properties eventually meets a version of this record: a county file, a disclosure, an inspection note that says “unrepaired earthquake damage” — and nothing else. No date. No event. No way to know, from the record alone, which earthquake it means or whether the shaking at that address could plausibly account for the damage in front of you.

I hit exactly this in a recent Washington State matter involving foundation cracking in an older residence. The causation question turned, in part, on whether documented historic earthquake damage could explain what was being attributed to another mechanism entirely. Here is the forensic earthquake investigation workflow that resolved it, and — more useful to my colleagues — the distance mistake I nearly put in writing that would have been a gift to opposing counsel.

Step 1: Derive candidate events from regional seismic history

With no date in the record, you work backward from what the region has actually produced. For the Puget Sound area, any pre-2000s damage note has three serious candidates: the 1949 Olympia earthquake (M~7.0), the 1965 Seattle–Tacoma earthquake (Mw 6.7), and the 2001 Nisqually earthquake (Mw 6.8). Build the candidate list first; then let data, not assumption, pick among them.

Step 2: Model the shaking at the address — not the region

The question is never “was there a big earthquake nearby?” It’s “what did the ground do at this parcel?” The U.S. Geological Survey publishes exactly that: ShakeMap grids model intensity (MMI), peak ground acceleration, and peak ground velocity across the affected region for significant events — including retrospectively reprocessed grids for historic earthquakes, produced with modern methods from archived data (standard USGS practice, and worth disclosing in your report so a reviewer doesn’t mistake a 2020 processing date on a 2001 event for an error).

Run each candidate event, read the modeled values at the subject address, and rank. In my matter, the leading candidate put the site at a modeled intensity approaching MMI VII — “very strong” shaking, the level at which well-built ordinary structures take slight-to-moderate damage and poorly built ones take considerable damage — with peak ground acceleration near 20%g. That is a documented mechanism capable of producing the observed damage class. Community “Did You Feel It?” responses near the site corroborated the intensity band, though with a small local sample I treated them as color, not evidence.

Step 3: The deep-event trap — epicentral distance is not source distance

Here is the part that nearly bit me. The Nisqually earthquake was a deep intraslab event — its source sat roughly 52 kilometers below the surface. For events like this, the distance printed next to “epicenter” is a horizontal measurement to a point on the map, not the distance to the earthquake. The true source-to-site (hypocentral) distance is:

Read that again: the site is not five miles from the earthquake. It is thirty-three. For a deep event, every site within about 50 km of the epicenter is effectively equidistant from the source — which is precisely why Nisqually, for all its magnitude, underperformed on damage relative to a shallow crustal event of the same size.

Step 4: Frame the opinion correctly

Modeled shaking data supports a statement of the form: a documented mechanism capable of producing this damage class existed at this location on this date. It does not support “the earthquake caused this crack” — that requires inspection: crack-edge weathering, displacement patterns, repair history, and the rest of the physical record. Keep the two statements separate in your report, and the data holds up.

One more discipline point: in expert work, cite the USGS primary sources — the event page, the ShakeMap product, the DYFI record — directly. They are the authoritative government record, and they are what survives scrutiny.

The tooling note

Full disclosure: the tool I used to compile the USGS record is one our practice built. QuakeProof geocodes an address, searches the USGS catalog (with an adjustable lookback for historic matters), interpolates the ShakeMap grid at the property coordinates, and returns both distances — labeled — with the deep-event caution generated automatically, magnitude types resolved to the authoritative moment magnitude, and every claim cited to its USGS source. The free checker lists every cataloged event near an address; the full report is $29. It was built for homeowners fighting “those cracks were pre-existing” claim denials, and the same honesty requirements — label what’s modeled, separate event-wide from address-level figures, never overstate — turn out to be what forensic use demands too. For homeowners rather than practitioners, this overview of ten situations where earthquake verification helps is the better starting point.

Case details in this article have been generalized and anonymized to protect client confidentiality. Nothing here is legal advice or an opinion on any pending matter.