Extreme events over time - Extreme Events: A Physical Reconstruction and Risk Assessment - page 265

in the case of tropical cyclones along the Great Barrier Reef it is possible that
peaks in the cycles of a number of variables (SST, PDO, SOI) coincide at an
average frequency of 250 years (phase peaks rather than variables with cycles
of identical periods). It is also likely that the cycle variability is considerable
so no one cycle may persist over long periods of time. Rather, the time periods
between phases of increased probability of hazard occurrence may vary. In this
situation a mean cycle period is likely to be the most appropriate measure of
the periodicity of the temporally variable probability with an uncertainty margin
that reflects the natural variability displayed in the long-term record. Despite
thepotential difficulties in employing such an approach, it nonetheless allows
forthe recognition and adoption of the fact that the hazard probability can
change with time and therefore more accurately reflects the return period of
so-called outlier events. Identifying the last major phase of enhanced hazard
activity from the palaeorecord will provide a reasonable estimate of the position
of the present day in the cycle. The likelihood of occurrence of a hazard of given
magnitude over the next 50--100 years can then be assessed. At times this may
suggest the probability of occurrence over a planning period to be lower than
otherwise expected and at other times this probability will clearly be higher.
Adopting such an approach acknowledges that natural hazards and their causal
mechanisms do not occur entirely randomly over time and there is to some
degree an inherent cyclicity or quasi-cyclicity in their occurrence. This then
logically suggests that their probability of occurrence must change over time
also and therefore risks must likewise be temporally variable.
Incorporating palaeorecords into hazard risk assessments
When developing policies, planning authorities usually focus on those
documents that summarise the state of the risk (all three components -- physical
characteristics and community exposure and vulnerability) rather than refer to
the moretechnical scientific papers that deal with the physical characteristics of
the hazard alone. Hence, it is incumbent upon those assessing community risks
to anatural hazard to also address the accuracy of estimates of the frequency
characteristics of the hazard. At the very least the risk assessors need to draw
attention to the level of uncertainty associated with such estimates. By doing
so, and as a corollary, any study addressing community risk could incorporate
variable temporal probabilities and their uncertainty margins into measures of
community exposure and vulnerability.
Planning often requires balancing economic gains and losses against the risk
of the hazard occurring. The life expectancy of many modern urban develop-
ments is between 50 and 100 years. So estimates based upon a short historical