Dependency of simulated tropical Atlantic current variability on the wind forcing

Abstract. The upper wind-driven circulation in the tropical Atlantic Ocean plays a key role in the basin-wide distribution of water mass properties and affects the transport of heat, freshwater, and biogeochemical tracers such as oxygen or nutrients. It is crucial to improve our understanding of its long-term behaviour, which largely relies on model simulations and applied forcing due to sparse observational data coverage, especially before the mid-2000s. Here, we apply two different forcing products, the Coordinated Ocean-ice Reference Experiments (CORE) v2 and the Japanese 55-year Reanalysis (JRA55-do) surface dataset, to a high-resolution ocean model. Where possible, we compare the simulated results to long-term observations. We find large discrepancies between the two simulations regarding the wind and current field. In the CORE simulation, strong, large-scale wind stress curl amplitudes above the upwelling regions of the eastern tropical North Atlantic seem to cause an overestimation of the mean and seasonal variability in the eastward subsurface current just north of the Equator. The wind stress curl of JRA55-do forcing shows much finer structures, and the JRA55-do simulation is in better agreement with the mean and intraseasonal fluctuations in the subsurface current found in observations. The northern branch of the South Equatorial Current flows westward at the surface just north of the Equator. On interannual to decadal timescales, it shows a high correlation of R=0.9 with the zonal wind stress in the CORE simulation but only a weak correlation of R=0.35 in the JRA55-do simulation. We also identify similarities between the two simulations. The strength of the eastward-flowing North Equatorial Counter Current located between 3 and 10° N covaries with the strength of the meridional wind stress just north of the Equator on interannual to decadal timescales in the two simulations. Both simulations present a comparable mean, seasonal cycle and trend of the eastward off-equatorial subsurface current south of the Equator but underestimate the current strength by half compared to observations. In both simulations, the eastward-flowing Equatorial Undercurrent weakened between 1990 and 2009. In the JRA simulation, which covers the modern period of observations, the Equatorial Undercurrent strengthened again between 2008 to 2018, which agrees with observations, although the simulation underestimates the strengthening by over a third. We propose that long-term observations, once they have reached a critical length, need to be used to test the quality of wind-driven simulations. This study presents one step in this direction.