Comparison of two time-variant forced eddy-permitting global ocean circulation models with hydrography of the Scotia Sea

A comparison between hydrographic observations and output from two realistically forced z-level global ocean circulation models (OCCAM and POCM_4C) in the Scotia Sea, South Atlantic, is described. The study region includes the southern part of the Antarctic Circumpolar Current (ACC) and the northern Weddell Gyre. Despite similar formulations, the models have different strengths and weaknesses. OCCAM simulates well the horizontal circulation around South Georgia but loss of Antarctic Bottom Water distorts the mean circulation in the central Scotia Sea. A poorer bathymetric dataset in POCM_4C means that the circulation is not adequately topographically steered leading to greater zonal flow and a southward shift of the fronts of the southern ACC. In a comparison with sea surface height variability data, OCCAM overestimates and POCM_4C underestimates the maximum values. Both models have higher background variability than the satellite data. Mean monthly model output is compared with a meridional hydrographic section from the study region. The regional water masses at the time of the hydrographic section (April 1995) are recognisably reproduced in both models despite some discrepancies. The surface waters are too saline in OCCAM (by 0.12–0.40) and too warm in POCM_4C (by >2 °C) suggesting problems with the air-sea surface heat and freshwater fluxes used to force both models and the models' vertical mixing parameterisations. Anomalous mixed layer properties in winter lead to inaccurate Winter Water characteristics in both models. Slumping of Circumpolar Deep Water occurs in OCCAM, associated with the loss of the bottom water. Subsurface restoration to climatology at buffer zones prevents this slumping in POCM_4C although the densest waters are not reproduced. The models overestimate the baroclinic transport of the section by up to a factor of two and simulate a significant barotropic component of transport. Overall, both models can be used in this region in ways that utilise their strengths. Further improvements are likely to come from better bathymetric representations, surface fluxes, and bottom water formation processes, elimination of spurious diapycnal mixing, improvement of vertical mixing parameterisations, and higher resolution.

Details