Abstract

The Southern Ocean is a critically important region for global climate for many reasons, including its key roles in determining the rate of transient climate change; forming/transforming many of the major intermediate and deep water masses; influencing sea level change; etc. It is also a region of low thermal stratification which is marginally stable mainly due haline stratification. Erroneous thermal stratification in climate models would therefore be expected to have a substantial impact on the simulated density stratification, limiting the accuracy of simulated uptakes of heat and carbon.

The Southern Ocean sea surface temperature (SST) biases in HADGEM3 are summarised, together with those documented for IPCC models. The many processes influencing sea surface temperature and expected errors in their representation in climate models, particularly HADGEM3 , are also discussed. We then attempt to systematically diagnose potential causes of the HADGEM3 SST biases. First we compare relevant parameters of the coupled HADGEM3 runs with those from equivalent forced runs, which employ DRAKKAR optmised ocean forcing and have much reduced SST biases. Compared to observational estimates there appear to be significant biases in the wind stress, the wind curl and the downwelling short wave in the regions which the SST biases are largest. Next we present coupled HADGEM3 perturbation experiments which include forcing the ocean using the time-varying DRAKKAR winds; increasing/decreasing the downwelling short wave to the ocean; varying the ocean lateral isopycnal mixing coefficient; and changing the configuration of the ocean vertical mixing parameterisation. We find that using DRAKKAR time varying winds and reduced downwelling short wave appear to reduce the SST biases. However, we also find that the SST biases are also affected by both the ocean lateral isopycnal mixing coefficient and configuration of the ocean vertical mixing parameterisation.

The reduced SST biases in forced ocean experiments together with the errors identified in the coupled winds and downwelling short wave suggest that errors in the atmospheric model could contribute to causing SST biases. However, the SST biases are also affected by changes to ocean model configuration. Further experiments and analyses are in progress to help to better understand the HADGEM3 Southern Ocean sea surface temperature biases.