 |
| Titel |
Atlantic Hurricanes and Associated Insurance Loss Potentials in Future Climate Scenarios |
| VerfasserIn |
Christoph C. Raible, Sabine Kleppek, Mark Wüest, David N. Bresch, Akio Kitoh, Hiroyuki Murakami, Thomas F. Stocker |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250054299
|
|
|
|
|
|
| Zusammenfassung |
| Atlantic tropical cyclone (TC) activity is a major source of substantial damage along the U.S.
coastline and the Caribbean. The recent trend and the potential future changes of TC activity
are currently under debate.
In this study climate model output is combined with a loss model of the Swiss Reinsurance
Company to assess changes of potential losses of the end of the 21st century. We use
time-slice experiments of two atmospheric general circulation models (AGCMs), the
ECHAM5 model (MPI, Hamburg, Germany) and the MRI/JMA model (MRI, Tsukuba-city,
Japan). For each model set up a control simulation for present day conditions and a scenario
simulation for the period 2070-2100 (SRES A2 for ECHAM5 and A1B for MRI/JMA) are
available. Both models are used with a high horizontal resolution which is suitable
to resolve TCs. TCs are individually identified by our TC detection and tracking
method.
The modeled tracks of TCs are evaluated by comparing the CTRL simulations with
best track data. This comparison shows that the number of intense storms is still
underestimated which could be traced back to the model resolution. In order to
account for this underestimation, we apply a statistical calibration leading to a better
representation of the TC intensities. The results of both models after the calibration show an
increased number of major hurricanes in the scenario simulation, but the number
of major hurricanes which make landfall is still underestimated in the MRI/JMA
model.
Despite the biases, the simulated hurricanes for the present and the future are used as input
for the loss model of the Swiss Reinsurance Company. This loss model estimate a
probabilistic event set and combine this with average vulnerability curves for all risk types
and locations. The resulting loss frequency curve for the MRI/JMA model shows that, e.g.,
a 10-yr event in the CTRL simulation will become an 80-yr event in the SCEN
simulation, because of less frequent landfalls in the SCEN simulation, whereas the
10-yr event of the ECHAM5 CTRL simulation will become a 6-yr event in the
future. On regional scale, e.g., single U.S. states, even stronger deviations between
the different models are found. Overall, we show that a combination of current
state-of-the-art models and loss models is possible, but the results still lacks robustness due
to the high uncertainty of projected changes in hurricane strength, location and
frequency. |
|
|
|
|
|
|