![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Understanding spatial and temporal dependencies in flood risk exposure in the UK |
VerfasserIn |
Linda Speight, Jim Hall, Chris Kilsby, Paul Kershaw |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250053566
|
|
|
|
Zusammenfassung |
In the UK flooding represents a major natural catastrophe risk. The Environment Agency
estimates that one in six houses in England and Wales are at risk of flooding, therefore when
large events occur the consequences can be catastrophic. For example the widespread floods
in the summer of 2007 cost the UK economy over £3 billion and caused the loss of 13 lives.
This is a particular concern for insurance companies who need to be able to accurately
model and understand flood risk to continue to provide insurance cover to property
owners.
The standard means of assessing risk is through Catastrophe (Cat) models. These are
complex process based models which use a mixture of numerical and statistical methods to
provide estimates of expected losses. The disadvantage of Cat models is that due to their
complexity, and commercial sensitivity surrounding their components, it is often difficult for
the end user to fully understand the underlying processes and hence to make informed risk
management decisions.
Illustrating a successful collaboration between academia and the insurance industry we
use a case study of one company’s exposure to demonstrate a methodology for flood risk
assessment at multiple sites nested within a national framework. This nested approach allows
for greater detail to be included at sites of interest and hence results in increased
understanding of the risk driving processes. Following a source-pathway-receptor approach, a
mixture of statistical and physically based methods is used in a systems based model
incorporating the most important random processes associated with flood damage. Within the
system, meteorological inputs are modelled statistically; floodplain inundation and damage
calculations are deterministic; and the consideration of flood defence failure is
probabilistic.
The statistical model of extreme events uses the conditional dependence model of
Heffernan and Tawn (2004). Flows at all gauges in a network are simulated conditional on
one gauge being above a specified threshold. The model is fitted over a range of time periods
allowing for consideration of temporal dependencies, for example in the travel time of a
storm event across the UK or due to lags as a flood peak passes through a river system. Due
to the requirement for concurrent data across the network, the model is fitted to
daily mean flow. This necessitates the use of a conversion method to transfer the
simulated daily mean flows into peak flows for input to the floodplain inundation
model.
The methodology explicitly couples spatial and temporal dependencies at multiple scales.
Through the statistical model of event occurrences we are able to simulate local and
widespread events allowing explicit consideration of the spatial distribution of flood events
and the probability of events affecting large geographical areas. We also include
consideration of the dependencies in flood risk at local scales through detailed consideration
of flood defence systems and failure mechanisms, including a new methodology to
incorporate spatial variation in defence properties. Results will be presented illustrating
the importance of incorporating spatial and temporal dependencies into flood risk
modelling.
The output from the modelling process is greater understanding of risk, and the associated
uncertainties, which can be used to inform decision making. For insurance companies this
may involve changing pricing policies, implementing requirements for mitigation measures
or investing in further analysis. Outside of the insurance industry, the proposed methodology
can be used to assess risk to any spatially distributed receptor and hence can be
used in a wide range of risk management processes involving spatial and temporal
considerations.
Heffernan, J. E. and J. A. Tawn (2004). A conditional approach for multivariate extreme
values. J.R.Statist.Soc.B 66: 497-530. |
|
|
|
|
|