Introduction - Extreme Events: A Physical Reconstruction and Risk Assessment - page 12

of storm tide simulations actually reflects what would most likely occur, in real-
ity, over a 10 000 year period. When this assumption is based upon only decades
of observational and/or instrumented data, or even a century of data, then it
is clear that it is unlikely to be realistic because it assumes stationarity of the
record. And therefore, the accuracy of this hypothesis is not tested. Problems
arise when planners and policy makers accept the results as realistic or at least
the mostrealistic achievable. The only realistic test of this assumption is to make
comparisons against events that have actually occurred during the time period
supposedly covered by the model runs i.e. 10 000 years. Many modellers might
say that this is not achievable because we do not have 10 000 years of continuous
records of tropical cyclone storm tides. We do, however, have geological records
of the most extreme events over the last 5000 years. It is true that these records
do not present data with the same level of precision as an instrumented record.
However, such records can be used to test the accuracy of model simulations
to within at least an order of magnitude. This has to be a considerably better
option than leaving such assumptions about the levels of risk from the haz-
ard untested. The geological record, for example, can tell us the height of the
1in3000 year storm tide event based upon the elevation of deposits of marine
debris such as corals and shell above normal sea-level and this can be com-
pared to the simulated 3000 year storm tide event. Hence, we do not need to
rely solely upon deduction i.e. the use of simulated events can complement this
approach through retroduction or the examination of a record of events that
really occurred over this length of time. In this sense, the prehistoric record is
more than sufficiently accurate to perform this task.
Prehistoric records focus on the contingency of past phenomena. Contingency
holds that individual events matter in the sequence of phenomena (Baker, 1994).
Changing one event in the past causes the sequence of historical events to be
different from that which is observed to have occurred. One recognised extreme
event only decades before an historical record begins changes the way we would
view that historical record. The fact that this event occurred in time and space
is real and it is registered in the prehistoric record. It is only our assumption
about nature's behaviour that holds that such an event might be an outlier
because these assumptions are formulated in the absence of any knowledge of
these prehistoric events. If they were to have occurred within the observed, and
particularly instrumented, record they would be more readily accepted as a real
event and form part of the assumption of the statistical distribution of events.
Baker notes that such uniqueness to history is alien to the thinking of physicists
who seek timeless, invariant laws presumed to be fundamental for nature. This
same view is held by engineers who often, in terms of hazard risk assessment,
develop conceptual idealisations about nature. Physics is the science devoted to