'1 in 100 Year Event'....So Why So Many?

Science describes the natural world around us.  Engineering is necessarily heavily reliant on measurement of that natural world.  Consequently, an experienced engineer develops an affinity with statistical measures and their interpretation.  One such measure is the “1 in 100-year event”.

This is a headline which is becoming more common as our environment changes.  The terminology is often misinterpreted and misunderstood when used in a general sense.  It is often used to describe a weather-related event such as: rainfall, storms, flooding, seismic activity.

Interpretation:  Statistically speaking, on average over a long time period, we expect a 100-year recurrence interval between similar events (where ‘expect’ refers to statistical expectation).

Misinterpretation:  This does NOT mean that, if the last such event occurred last year, then the next such event will occur in 99 years’ time.

As with any statistical measure, an amount of data is gathered and combined to give a meaningful description of a certain event.

A ‘1-in-100-year event’ indicates that, in any one year, we expect the event to occur with a probability of 1% (1 out of 100)*. A common terminology used is Annual Exceedance Probability (AEP).  This is the probability of the event occurring in any one particular year.  In our example this is 1%^.

*Similarly, a 1-in-50-year event has AEP 2%, a 1-in-20 year event has AEP 5%.

^As working probabilities the AEP for a 1 in 1-year event is 63%, and the AEP for a 1 in 2-year event is 39%.

Issues which can affect the accuracy of the statistic are many.  One such issue is that the amount of data is not large enough; generally, the more data, the more reliability.  A second issue occurs when the statistic is used as a future predictor of the event.  The underlying assumption is that the past is an accurate predictor of the future; however, this is not necessarily the case (e.g. climate change, human activity). 

A third issue occurs when the effect of area is ignored; i.e. a rainfall measurement is taken at a specific location where the AEP may be 1%.  Perhaps what we really want to know is the probability of the event occurring at any location in a larger area (e.g. a city, or a state).  Intuitively we sense that the AEP for that event in the larger area will be greater than for the single location.

As with any statistical measure, its accuracy is only ever as good as its interpretation.  The interpretation of the 1-in-100 measure needs to be treated with caution.

Websites of interest:

[1] http://www.bom.gov.au/water/designRainfalls/rainfallEvents/why100years.shtml

[2] https://coastadapt.com.au/sites/default/files/infographics/What_is_a_100-year_flood.pdf

[3] https://water.usgs.gov/edu/100yearflood.html

[4] https://albertawater.com/how-are-flood-maps-created-in-alberta/what-is-a-1-100-year-flood

[5] https://www.src.com.au/earthquakes/seismology-101/earthquakes-in-australia/

Brendan Nielsen