Exploration of model errors in terms of Sea Surface Height and Temperature in a ¼° model of the North Atlantic
Nadia Ayoub(1) , Marc Lucas(1) , and Pierre De Mey(2)
LEGI - BP53x,
38041 Grenoble Cedex ,
(2) CNRS/LEGOS, 18 avenue Edouard Belin, 31401 Toulouse Cedex, France
The knowledge of model errors is a fundamental issue as it is needed to evaluate the quality of a GCM simulation and to formulate data assimilation problems. Understanding the errors covariance structure is also critical for the interpretation of surface data, such as altimetry or satellite temperature observations, in terms of subsurface signal on both the tracers and dynamical fields. Despite the considerable development in data assimilation techniques stemming from the availability of altimetric data since the ERS1 and TOPEX/POSEIDON missions, and in spite of the increasing realism of numerical models, the estimation and formulation of model errors remain a vexing methodological issue as well as a major drawback to forecasting systems.
This study aims to contribute to the exploration of model errors in the surface layers within a realistic modelling framework of the North Atlantic circulation. The focus is on errors induced by uncertainties in the atmospheric forcing fields. The approach is based on an ensemble method. Our objective is to understand the sensitivity of the superficial heat content to the forcing and the relationships between errors on surface fields and on subsurface variables that are not directly observed. The emphasis is on sea surface height and temperature (SSH and SST) data since satellite observations provide very complete datasets and are thus part of most data assimilation systems. Furthermore, correlation between errors in SSH and SST are expected to be significant since both fields contain the signature of heat content and dynamical processes. We are interested in seasonal scales. Since the sensitivities are expected to depend on the ocean state, separate analyses are done for winter and summer conditions. The ocean model is OPA9 in a 1/4 deg. configuration over the North Atlantic, stemming from the DRAKKAR project. The model uses bulk formulae for the air/sea exchanges. ERA40 reanalysis fields are used for the atmospheric forcing for the years 1984-2000. Our work consists of generating an ensemble of atmospheric forcing perturbations. This is followed by an analysis of the ensemble of integrations. The objectives are threefold:
- 1/ to estimate error covariances in SSH, SST and sea surface salinity (SSS),
- 2/ to characterise the impact on the mixed layer properties and heat budget,
- 3/ to identify the space-time structure of the impact of local SST and SSH perturbations on other model variables.
The results presented are for a 6-month run in 1993.