|
Titel |
An estimate of the reliability of climate model trends |
VerfasserIn |
Geert Jan van Oldenborgh, Francisco Doblas-Reyes |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250058036
|
|
|
|
Zusammenfassung |
In weather and seasonal forecasting, a reliable probability forecast is
defined as one in which the forecast probability is equal to the observed
probability, verified over a large number of cases. An ensemble forecast
can only be interpreted as a probability density function (PDF) if the
ensemble is reliable. Here we apply this concept to the CMIP3 multi-model
ensemble of climate model simulations that is widely used for the
construction of climate scenarios. We verify hindcasts of the temperature
trend over 1950-2010 against observed trends. As there is only a single
verification time, the spatial dimension is used to collect the data for
the verification. For each grid point the percentile of the observed
trend in the PDF of the ensemble is computed. If the trend hindcasts would
reliable, the resulting reliability histogram would be flat. However, like
many weather and seasonal forecast ensembles, the climate change ensemble
is seen to be overconfident: the observed trend is outside the range of
modelled trends in many more grid points than would be expected by chance.
To obtain calibrated forecasts from the ensemble the ensemble spread will
have to be increased to obtain a reliable PDF. |
|
|
|
|
|