A Statistical Concept, Not a Literal Prediction Calendar
A code's wind, seismic, or flood hazard map often references a specific "return period" — a 50-year wind speed, a 475-year seismic event, a 100-year flood level. It's a common and understandable misreading to interpret these numbers as a literal prediction ("this event happens exactly once every 475 years"), but the actual statistical meaning is different: a return period is the inverse of the annual probability of that hazard level being exceeded in any given year, not a literal, regularly spaced recurrence interval.
What the Number Actually Represents
A 475-year return period event corresponds to roughly a 1-in-475 (about 0.21%) chance of being exceeded in any single year — which means, purely from that annual probability, such an event could in principle occur in consecutive years, or not occur for a thousand years, with the 475-year figure representing a long-run statistical average interval rather than a scheduled recurrence. This distinction matters for structural design, since a design life calculation (discussed elsewhere in this series) combines this annual probability with the structure's intended service life to arrive at the overall probability of that hazard level being exceeded at least once during the structure's lifetime.
Why Different Hazard Types Often Use Different Return Periods
Codes frequently specify a longer return period (implying a lower annual probability, and therefore a more severe design event) for seismic hazard than for ordinary wind hazard, reflecting both the generally higher consequence of under-designing for a major earthquake and genuine differences in how reliably each hazard type's statistical behavior can be characterized from available historical data — a deliberate calibration choice, similar in spirit to the importance factor discussed elsewhere in this series, reflecting how much risk a code is willing to accept for a given hazard type and building category.
Quick FAQ
Does a structure's design life automatically match the return period used in its design?
No — they're related but distinct concepts; the return period feeds into a probability-of-exceedance calculation over whatever design life the structure actually has, meaning a longer design life structure faces a higher cumulative probability of experiencing its design-level hazard event at some point, even using the identical return period as a shorter design life structure.
Can return period hazard maps become outdated the same way the climate load assumptions discussed elsewhere in this series can?
Yes, closely — return period maps are built from historical hazard data and statistical models, meaning they face the exact same potential gap between historical calibration and future conditions discussed in this series' climate load content, which is part of why hazard maps periodically get revised as more data and improved climate understanding accumulate.
Final Thoughts
A 'once in 475 years' event isn't a scheduling promise — it's a probability statement about any given year, which is precisely the distinction that matters when combining it with a structure's actual design life to understand the real, cumulative chance that structure will face its design-level hazard at some point during its service.