 |
| Titel |
Partitioning the covariance between space and time to reconcile the controversy in model evaluations |
| VerfasserIn |
Fubao Sun, Michael Roderick, Graham Farquhar |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250111143
|
| Publikation (Nr.) |
 EGU/EGU2015-11226.pdf |
|
|
|
|
|
| Zusammenfassung |
| There is an ongoing debate on “whether past performance of hydrologic/climate models has any guarantee of future skill”. That debate can be extended to a more general argument on "can we predict inter-annual variations when models have reasonable performance over intra-annual timescales, e.g., daily, monthly?". To contribute to that debate we start with our recently developed analytical approach that partitions the variance between space and time [Sun et al. 2010, GRL]. We show that the same framework can be used to answer the question by incorporating the covariance. This approach allows the multiple representations that are needed for handling a climate model ensemble. The covariance partitioning scheme can accommodate variations at various space and time scales. We show that the argument arose originally because of incorrect handling of intra- versus inter-annual variations. |
|
|
|
|
|
|