Stochastic study of the temperature response of the upper ocean to uncertainties in the atmospheric forcing in an Atlantic OGCM

International audienceThe impact of errors in atmospheric forcing on the behaviour of ocean models is a fundamental issue for ocean modellers and data assimilation and one that has yet to be fully addressed. In this study, we use a stochastic modelling approach with 50 7-months (September-March) primitive equation eddy permitting (1/4°) integrations. We investigate the response of the oceanic circulation to atmospheric uncertainties, focusing principally on their impact on the upper oceanic temperature field. The ensemble is generated by perturbing the wind, atmospheric temperature and incoming solar radiation of the ERA40 reanalysis. Each perturbation consists of a random combination of the 20 dominant EOFs of the difference between the ERA40 and NCEP/CORE reanalysis datasets. The ensemble standard deviation of various interfacial and oceanic quantities is then examined in the upper 200 m of three distinct regions of the North Atlantic: in the Gulf Stream, in the Northern Tropical band and in the North East Atlantic. These show that even a very small perturbation of the atmospheric variables can lead to significant changes in the ocean properties and that regions of oceanic mesoscale activity are the most sensitive. The ocean response is driven by vertical diffusivity and eddy activity. The role of subsurface currents is also crucial in carrying the eddy signal away from the regions of mesoscale activity. Finally, the decorrelation time scale of the mesoscale activity is critical in determining the amplitude of the oceanic response