Introduction - Extreme Events: A Physical Reconstruction and Risk Assessment - page 10

Current observational
record
Currently adopted
relationship
Frequency
Frequency
Figure 1.5. Non-stationarity in the time series.
community vulnerability and levels of exposure. If we believe that the high-
magnitude event is an outlier then we will believe that the chances of an event
of similar size occurring during the next planning cycle, or the near future, are
low (i.e. as set out in Table 1.1). For example, if we perceive the outlier event
to be a 1 in 500 year event (0.2% AEP) then we will draw the conclusion that
its chances of occurring in the next 50 years are less than 10%. However, if we
find from a longer-term record that this magnitude event is really more likely
to be a 1 in 100 year event (and indeed it did occur only a little over 100 years
ago in reality) then its chance of occurring in the next 50 years is about 39%.
Alternatively, it could be an event that signals the onset of a change in hazard
behaviour. Either way, when we assume it was a 1 in 500 year event, and ignore
thepossibility that it is part of a normal series of events for a regime that was
not recognised due to the brevity of the observational record, then we will have
placed potentially large numbers of people and property at higher than pre-
dicted levels of risk. Each change of phase in hazard behaviour, i.e. change of
slope in the magnitude--frequency curve, can be referred to as a hazard regime
(Nott, 2003). These regimes can be likened to alternating periods of variable sta-
tionarity. Recognising the possibility that these regimes may exist for a hazard
at any location can only help us to gain a more realistic view of the nature of
the hazard.
Reality versus reasonableness
Mathematics and statistics have dominated our approach to assessing
risk from natural hazards. The strength of these methods is in their predictive