Inverse Modeling of Seasonal Variations in the North Atlantic Ocean

Abstract

An ocean general circulation model (OGCM) of the North Atlantic Ocean is fitted to the monthly averaged climatological temperatures and salinities of Levitus using the adjoint method, representing a significant step forward with respect to previous steady OGCM assimilations. The inverse approach has two important advantages over purely prognostic calculations: (i) it provides an estimate of the North Atlantic circulation and of its seasonal variability, which is optimally consistent with the OGCM dynamics and with the assimilated hydrography; (ii) it provides optimal estimates of the monthly surface heat and freshwater fluxes consistent with the used climatology, which are the most poorly known surface forcing functions. Seasonality is ensured by penalizing field differences between month 13 and month 1 of the forward time integration within each iteration of the adjoint procedure. The primary goal of this work is to estimate large-scale oceanic properties important for climate issues and how they are affected by the inclusion of the seasonal cycle. The resultant meridional overturning displays significant seasonal variations. The surface Ekman cell centered at 35°N reaches a maximum intensity of ?7 Sv (Sv ≡ 106 m3 s?1) in wintertime, while the North Atlantic Deep Water cell reaches a maximum strength of ?19 Sv in summertime. Its annual average is of ?17 Sv, in good agreement with the recent estimate of Schmitz and McCartney. The poleward heat transport exhibits the strongest seasonal variations, reaching its maximum value of 0.85 ? 1015 W at ?25°N in summertime or 0.85 PW (1 PW = 1015 W). The annual average at 25°N is ?0.7 PW, weaker than observational estimates. The dynamical analysis indicates that the wind forcing is the controlling factor for these variations by controlling the time-varying Ekman cell. Comparison with previous steady-state optimizations of Yu and Malanotte-Rizzoli shows that the optimization with seasonal forcing produces three major improvements in the inverse results. First, the inclusion of the seasonal cycle greatly improves the estimated hydrography (temperature and salinity fields) by eliminating the basinwide cold bias in the upper ocean and the warm bias in the deep ocean found in the steady-state inversions. As a consequence, the velocity fields are also significantly improved, with a tight and strong Gulf Stream jet. Second, the monthly optimal estimates of surface heat and freshwater fluxes provide an annual average resembling closely the observational climatological means, a striking contrast to the fluxes estimated in the steady assimilation. Finally, the most important improvement is in the estimate of the poleward heat transport. The annual mean meridional heat transport shows an increase of ?0.2 PW at all latitudes with respect to the steady-state heat transport, thus demonstrating the importance of rectification effects of the seasonal cycle.