 |
| Titel |
Understanding independence |
| VerfasserIn |
James Annan  , Julia Hargreaves |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250134667
|
| Publikation (Nr.) |
 EGU/EGU2016-15413.pdf |
|
|
|
|
|
| Zusammenfassung |
| In order to perform any Bayesian processing of a model ensemble, we need a prior over the ensemble members. In the case of multimodel ensembles such as CMIP, the historical approach of ``model democracy'' (i.e. equal weight for all models in the sample) is no longer credible (if it ever was) due to model duplication and inbreeding. The question of ``model independence'' is central to the question of prior weights. However, although this question has been repeatedly raised, it has not yet been satisfactorily addressed.
Here I will discuss the issue of independence and present a theoretical foundation for understanding and analysing the ensemble in this context. I will also present some simple examples showing how these ideas may be applied and developed. |
|
|
|
|
|
|