Tropical cyclones - Extreme Events: A Physical Reconstruction and Risk Assessment - page 105

in Florida (Shelby and Western Lakes), New England (Succotash Marsh) and New
Jersey (Whale Beach) (Fig. 4.4). Good agreement was found between both types
of records when the return probabilities of the prehistoric record were averaged
in the comparison. There were, however, distinct periods of time during the late
Holocene in these regions of the US when the annual landfalling probabilities
of tropical cyclones differed substantially from the average. For example, the
probability of an intense cyclone occurring per year based upon the average
of the 5000 year record from Western Lake, Florida was approximately 0.35%.
Whereas the period between 3400 and 1000 14 CyearsBP at this site had an
annual probability of cyclone occurrence of 0.5%, and the probability for the
last 1000 years was much lower at 0.1% (Liu and Fearn, 2000). Interestingly also
no cyclones of the intensity registered in the prehistoric record have struck the
Western Lake region during the last 130 years. So which annual probability is
themore realistic for risk assessment purposes?
Long-term records often show non-stationarity where periods or regimes exist
that differ from each other in terms of event magnitude and frequency. Long-
term records, therefore, remind us that the assumption that the relationship
between the magnitude and frequency of tropical cyclones remains constant
over time is not always true. Examination of cyclone palaeorecords allows us
to identify the nature of regime changes and, therefore, make more accurate
predictions of future risk. All too often, however, the prehistoric record, and
hence recognition of hazard regimes, is overlooked in risk assessments of tropi-
cal cyclones. The northeast Australian region is a case in point.
When observational bias is removed, the record of tropical cyclones in the
Australian region shows no appreciable change in variability over the last 130
years. The longest complete data set of land falling cyclones occur from northeast
Australia, and here the usual pattern of many more lower magnitude events
occurring compared to higher magnitude ones exists. The record shows that
during this time period only one category 5 and two category 4 cyclones struck
the east coast of Queensland, and each of these occurred prior to AD 1920.
The rarity of these extreme events since this time has led to a relatively
blase attitude towards the severity and consequences of such a hazard in some
regional centres. Indeed, it could be argued that the lack of familiarity with
such extreme events has led to the assumption that they are unlikely to ever
occur within the lifetime of many individuals. This has led to the development
of inadequate policies governing the location of buildings relative to sea level.
Now there are numerous buildings less than a few metres above mean sea level,
and well within the zone of possible storm tide inundation.
Using the historical data set, several studies have analysed the recurrence
intervals of cyclones in this region, and through extrapolation estimated the
frequency of the most extreme events (Harper, 1998; McInnes et al ., 1999). There