Simulating SST Variability by Forcing a Coupled Model With Observed Wind Stress

Abstract Ocean general circulation models (GCMs) forced with observed surface atmospheric variability are widely used to simulate and investigate interannual variations in sea surface temperature (SST). However, imposing atmospheric thermodynamic forcing is physically inconsistent as SST usually drives air temperature and humidity. To overcome this problem, the present study forces a coupled atmosphere‐ocean GCM with observed wind stress while retaining thermodynamic air‐sea coupling. This ocean hindcast system proves a useful tool to assess the role of wind stress in observed SST variations year to year. The model hindcast is highly correlated with observed SST variability in the equatorial oceans, consistent with the dominance of Bjerknes feedback. In particular, the hindcast reproduces the El Niño‐Southern Oscillation (ENSO) realistically, with a correlation exceeding 0.9. In contrast, the hindcast skill is low in eastern subtropical oceans, where under background trade winds, stochastic atmospheric forcing excites thermodynamically coupled variability. This suggests that the Pacific meridional mode acts mainly as a stochastic forcing on ENSO rather than being forced by ENSO. The role of wind stress feedback in ENSO growth can be examined by modifying wind stress forcing. A hindcast experiment without concurrent wind stress anomalies reveals that subsurface temperature anomalies in the western Pacific drove the growth of the 2023 and 2009 El Niños as well as the 1998 La Niña, with the classic Bjerknes feedback playing a secondary role.