 |
| Titel |
Multi-year forecast of Atlantic tropical cyclone activity using EC-Earth |
| VerfasserIn |
L.-P. Caron, C. G. Jones, F. J. Doblas-Reyes, M. Caian |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250069501
|
|
|
|
|
|
| Zusammenfassung |
| Seasonal forecasts of Atlantic hurricane activity are now routinely performed by
various groups and have become relatively successful at forecasting basin wide
activity a few months before the official start of a hurricane season. Part of the skill
in predicting the level of tropical cyclone activity for a given season comes from
the ability to predict the behaviour of ENSO, which strongly impacts (through
teleconnections) hurricane formation over the Atlantic. Over the same basin, the ~100 years
of available hurricane records also show variations at the decadal timescale. This decadal
fluctuation appears linked to an oscillation in North Atlantic sea surface temperatures
(SSTs). Coined the Atlantic Multi-decadal Oscillation (AMO), a successful multi-year
forecast of this oscillation offers the potential to produce a skillful multi-year forecast
of hurricane activity. Such forecast could help mitigate against (potentially huge)
hurricane-related losses through improved preparedness and improved insurance
schemes.
Using EC-Earth, a coupled global atmosphere-ocean model, we perform a series of
ensemble decadal re-forecasts at 5 year intervals between the 1965-2005 period and
investigate the ability of these re-forecasts at capturing observed variations in North Atlantic
SSTs (in essence, the AMO) as well as other large-scale fields known to impact cyclogenesis.
We also use an automated procedure to track the tropical cyclones produced in these
re-forecasts, which then allows direct comparison with the actual number of tropical cyclones
that formed over the equivalent period.
Preliminary analysis shows that EC-Earth re-forecasts manage to capture variations in
large-scale fields relatively well, especially variations in the AMO, which suggests a
potential for skillful multi-year forecast of Atlantic tropical cyclones. However,
direct comparison of simulated and observed TC numbers does not offer the same
level of skill. Tropical cyclone numbers in the re-forecasts are biased low, and this
appears to result not only from the low-resolution of the simulation (~1.125°), but
also from a drift in simulated SSTs. This drift brings surface temperature below
the 26°C threshold required for tropical cyclone formation, effectively shutting
down storm formation over large parts of the tropical Atlantic. Potential solutions to
circumvent these limitations are currently being considered and will be discussed. |
|
|
|
|
|
|