 |
| Titel |
The Mediterranean Sea 1985-2007 re-analysis: validation results |
| VerfasserIn |
Mario Adani, Srdjan Dobricic, Nadia Pinardi |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250039387
|
|
|
|
|
|
| Zusammenfassung |
| Re-analyses are different from analyses because they are consistent for the whole period since the oceanic state estimates are produced without changes in the modelling assumptions and they are usually done with systems which are more advance then the available systems at the time of the observations collection. A fundamental part of a re-analysis system is the data assimilation scheme which minimizes the cost function penalizing the time-space misfits between the data and the numerical solutions, with the constraint of the model equations and their parameters. In this work we will compare ocean circulation estimates provided by pure simulation, a system in which the assimilation scheme is based on a sequential algorithm: Optimal Interpolation (OI) and a three-dimensional variational scheme (3dvar).
The OGCM used in this work is based on OPA 8.1 code (Madec et al. 1998), which has been implemented in the Mediterranean Sea by Tonani et al.(2008). The model has 1/16th horizontal resolution and 71 unevenly spaced vertical levels. The present model formulation uses a realistic water flux with river runoffs which improves the realism of the simulation. One re-analysis is produced with the Reduced Order Optimal Interpolation (ROOI) (De Mey and Benkiran, 2002) and the other with OceanVar (Dobricic and Pinardi, 2008). The observational data sets assimilated for both reanalysis are:
• the historical data archive of MedATLAS (Maillard et al., 2003) which contains vertical in situ profiles of temperature and salinity from bottles, XBT, MBT and CTD sensors
• temperature and salinity profiles collected in the framework of MFSPP and MFSTEP projects
• CLS along track satellite sea level anomaly data from ERS1, ERS2, Envisat, Topex/Poseidon, Jason1 satellites (Pujol and Larnicol,2005)
Reanalyzed daily mean fields of Sea Surface Temperature (SST) from Medspiration (Marullo et al., 2007) and the Delayed-Time operational product of CNR-ISAC have been used to relax the model SST.
The Mean Dynamic Topography of (Dobricic, 2005) has been used for both experiments.
The model is forced with a combined dataset of ECMWF analysis when available and ERA-15. The precipitations are monthly mean climatology of the NCEP re-analysis (Kistler et.al 2001), the river runoff data are monthly mean climatology from the Global Runoff Data Centre (GRDC) and from Raicic (1996) for the minor Adriatic Sea rivers.
The assimilation schemes help in reducing the spin up time of the model by acting as a forcing inside the water column. Both re-analyses show significantly better results then the simulation reducing both bias and root mean square error even though the structure of the error remains almost the same of the simulation: the largest error for tracers is confined in the thermocline especially in summer, highlighting a problem in the mixing parameterization; the majors error for SLA is confined in the most dynamically active areas. Satellite altimetry observations result in a fundamental dataset to constrain model solution and since its homogeneity in the sampling they permits a consistent assessment of the model behaviour along the years which it is not possible from in-situ observations whose sampling is extremely inhomogeneous both in time and space.
This study describes the development of modelling and data assimilation tools for the production of re-analysis for the entire Mediterranean Sea. In order to carry out a re-analysis two major steps were undertaken in this work. In the first, the general circulation model was upgraded to have the correct air-sea water fluxes. In the second, two assimilation schemes, one new and the other consolidated, were compared to show their impact on the quality of the re-analysis. The general circulation model used in this study is shown to be capable of reproducing quite accurately the ocean dynamics of the Mediterranean Sea. The results have shown that the model solution is in agreement with data and observations, even though some parameterizations of the model should be improved (i.e. heat flux and mixing processes). The new implementation of a realistic water flux, proposed in this study, has improved the model solution so that re-analysis is possible. The study of the re-analysis produced shows that both products are sufficiently accurate for appropriate climate studies. Both assimilation schemes show good capabilities in correcting the solutions provided by the dynamical model. Moreover it has been shown the ability of both systems in retaining this information and projecting it in the future. Eventually, even for very complex non linear systems, with millions of prognostic variables, the equality between the Sequential Kalman Filter Approach and the Variational one as been demonstrated. |
|
|
|
|
|
|