You must have a javascript-enabled browser and javacript and stylesheets must be enabled to use some of the functions on this site.
 
   

 

Synergy between ocean observations and numerical simulations : CLIPPER heritage and DRAKKAR perspectives

Thierry Penduff(1) , Bernard Barnier(1) , Anne-Marie Treguier(1) , and Pierre-Yves Le Traon(2)

(1) CNRS, BP53, 38041 Grenoble, France
(2) IFREMER, BP70, 29280 PLOUZANE, France

Abstract

Research on oceanic circulation, variability and underlying processes has been stimulated since the WOCE years by combined examinations of observations and model simulations. This synergy is achieved in data assimilation studies, but also whenever observations are used with unconstrained, high-resolution, multiyear simulations forced by atmospheric reanalyses. Between 1995 and 2003, the French CLIPPER project has developed and run Atlantic models based on the OPA8 code at 1°, 1/3° and 1/6° resolution to study various aspects of the ocean dynamics over the period 1980-2000+. Its successor DRAKKAR is a community modelling program that partly aims at improving model-observation synergies, in an extended context: European collaboration (France, Germany, Russia, Finland), wider domain (Atlantic-Nordic Seas to global), increased resolution (1/4° to 1/12°, with local grid refinement capabilities), longer period (1950-present), richer and improved physics (OPA9+LIM sea-ice, parameterizations).

The evaluation of numerical solutions against available observations is an example of model-data synergy. CLIPPER solutions were evaluated by several authors who extracted model counterparts of real data colocated in space (and possibly time) and treated both datasets identically. For example, lagrangian time/space eddy scales (Lumkin et al, 2002), deep subtropical zonal flows (Treguier et al, 2003a), Agulhas rings properties (Treguier et al, 2003b), interannual variations of water mass characteristics (CLIPPER Group, 2001) and of basin-scale eddy distribution (Penduff et al, 2004) were validated and studied by building and comparing colocated datasets (drifters, ADCP, BRAVO timeseries, and altimeter fields, respectively). More quantitative validation studies may involve model-data correlations as done by Illig et al (2004) for tropical sea level anomalies, or require the development of synthetic misfit (or model skill) estimates, as proposed by Penduff et al (2005) for current meter data. Within the OST/ST framework, the DRAKKAR group wishes to improve validation methods and make them more quantitative, by generating synthetic datasets colocated with various observations (e.g. ARGO, T/P, Jason, etc) and defining appropriate skill estimates.

High-resolution satellite products should help improve the forcing of ocean models, in combination with atmospheric reanalyses. The CLIPPER group made use of satellite SSTs (Reynolds and Smith, 1994) for air-sea flux corrections, and observed that scatterometer winds may substantially improve the simulations at low latitudes. Reanalysed atmospheric variables are being used through bulk formulae to force the global and regional DRAKKAR models. The merging of recent satellite datasets (e.g. winds, radiative fluxes) with reanalyses is currently investigated, and their impact on oceanic simulations will be evaluated.

In turn, realistic simulations provide a dynamical context to interpret observations or design observing systems. Among other examples, CLIPPER-derived synthetic datasets proved useful for estimating the representativeness of hydrographic transects (Treguier et al, 2005), diagnosing the spatial and temporal scales of surface salinity for remote sensing applications (Molines et al, 2001), evaluate the design of the ARGO array (Guinehut et al, 2002) or its combination with satellite data for temperature analyses (Guinehut et al, 2004). One of DRAKKAR objectives is to pursue such collaborative investigations in order to characterize the representativeness of (possibly combined and/or future) observation systems, the observability of small-scale processes or climat indexes. To reach this goal, different kinds of synthetic datasets (satellite, in-situ, up to kilometric resolutions) will be designed, generated, distributed and analysed in collaboration with the OST/ST and observationalist communities.

 

Workshop presentation

Full paper

 

                 Last modified: 07.10.03