Return periods can get a bad press. It is no surprise that they are often misunderstood by the public. Things to do with rarities have probably not been helped by the National Lottery. The odds of an entry winning the jackpot are about 14 million to one, yet there are winners in most weeks. With river flooding, much of Sussex sees nothing for decades and then the fleet sails in. Confusing.
However, the scorn poured on return periods by many professionals is disappointing. An objective description of event rarity - albeit approximate - is surely important. How else can one distinguish the vulnerable from the unlucky? While there is fair claim that annual exceedance probability is the more precise expression of rarity, quoting the return period in years avoids the scope to misread decimals or to misconstrue percentage probabilities.
The technical point that seems to get missed is that misleading return periods arise from mistaken logic. One can only ascribe a rarity to a fully specified feature or impact:
the maximum water level at a particular site
the peak flow in a particular river reach
the 1-day rainfall at a particular site
the maximum depletion in a reservoir if operated in a stated way
the insured loss in a given year for a given portfolio of flood risks.
The mistake comes from measuring rarity in terms of one feature, and transferring the assessment to another.
Yes, doing it right can be quite a challenge: one that has yet to be resolved for drought severity assessment. It's so much easier to pick out a feature that makes an event look rare, and say: This was a whopper. Didn't we cope well? The drawback comes from such assessments disguising the level of risk to which an individual or community is exposed. When a second or third extreme occurs in quick succession, it's the return period concept that gets put on the ducking-stool, when it ought to be the shallow thinker.