Impact of the Benguela coastal low-level jet on the southeast tropical Atlantic SST bias in a regional ocean model

Abstract

The current generation of coupled ocean–atmosphere climate models, which are widely used for studying the dynamics of the climate system and to make projections of future climate, suffers from erroneously warm (exceeding 5 °C) sea surface temperature (SST) simulation in the eastern boundary upwelling systems (EBUS) in the Atlantic and Pacific oceans. While recent improvements in the horizontal resolution of coupled model components helped to alleviate this issue in the North and South Pacific and North Atlantic oceans, the warm bias in the Southeast Tropical Atlantic (SETA) region still remains above 5 °C. Recent studies have highlighted the importance of inaccuracies in the near-coastal winds, especially in the Benguela low-level coastal jet (BLLCJ), in the formation of this warm SST bias. In addition, unique ocean circulation features like the poleward flowing coastal current and the presence of a strong coastal SST front were argued to contribute to the complexity in simulating realistic SST in the SETA region. In this study, we investigate how ocean model resolution and the spatial structure and strength of the BLLCJ affect the SST warm bias in the SETA. We conducted a suite of ocean model experiments with varying horizontal resolutions (81–3 km) and different atmospheric forcing products. We found that the accuracy in the magnitude of BLLCJ’s alongshore component and in the spatial structure of BLLCJ are the primary factors in reducing the warm SST bias in SETA and the ocean model horizontal resolution is of only secondary importance. A weaker alongshore component of the BLLCJ leads to a weaker upwelling through Ekman transport and fails to generate a downwind equatorward surface jet in the ocean. When the core of the BLLCJ is too far offshore, the broad wind stress curl zone gives rise to anomalous poleward depth integrated flow through Sverdrup balance and weakens the wind stress curl induced Ekman pumping near the coast. Both the presence of a poleward current, which can transport warm equatorial waters farther south, and the absence of the equatorward surface jet contributes to the warm SST bias. Heat budget analysis shows that the contribution from the anomalous poleward flow is a bigger contributor to the warm bias than the weak upwelling. With more realistic, high-resolution winds, an ocean model resolution increase from 27 to 9 km slightly reduces the extent and magnitude of SST bias. However, no such improvement occurs with coarse resolution less realistic winds. Sensitivity experiments using atmospheric forcing fields from different sources confirm that the major contribution to the warm SST bias is in fact from the erroneous winds rather than other atmospheric forcing fields like net shortwave and downward longwave radiation. Further studies are needed to understand whether eddy heat advection or mean flow heat advection primarily cause the warm SST bias.