![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Assimilation of temperature and salinity profile data in the Norwegian Climate Prediction Model |
VerfasserIn |
Yiguo Wang, Francois Counillon, Laurent Bertino, Ingo Bethke, Noel Keenlyside |
Konferenz |
EGU General Assembly 2016
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 18 (2016) |
Datensatznummer |
250132591
|
Publikation (Nr.) |
EGU/EGU2016-13114.pdf |
|
|
|
Zusammenfassung |
Assimilating temperature and salinity profile data is promising to constrain the ocean
component of Earth system models for the purpose of seasonal-to-dedacal climate
predictions. However, assimilating temperature and salinity profiles that are measured in
standard depth coordinate (z-coordinate) into isopycnic coordinate ocean models that are
discretised by water densities is challenging. Prior studies (Thacker and Esenkov, 2002; Xie
and Zhu, 2010) suggested that converting observations to the model coordinate (i.e.
innovations in isopycnic coordinate) performs better than interpolating model state to
observation coordinate (i.e. innovations in z-coordinate). This problem is revisited here with
the Norwegian Climate Prediction Model, which applies the ensemble Kalman filter (EnKF)
into the ocean isopycnic model (MICOM) of the Norwegian Earth System Model. We
perform Observing System Simulation Experiments (OSSEs) to compare two schemes (the
EnKF-z and EnKF-ρ). In OSSEs, the truth is set to the EN4 objective analyses and
observations are perturbations of the truth with white noises. Unlike in previous
studies, it is found that EnKF-z outperforms EnKF-ρ for different observed vertical
resolution, inhomogeneous sampling (e.g. upper 1000 meter observations only),
or lack of salinity measurements. That is mostly because the operator converting
observations into isopycnic coordinate is strongly non-linear. We also study the horizontal
localisation radius at certain arbitrary grid points. Finally, we perform the EnKF-z
with the chosen localisation radius in a realistic framework with NorCPM over a
5-year analysis period. The analysis is validated by different independent datasets. |
|
|
|
|
|