The Weddell Sea is one of the key regions of the Southern Ocean with respect to climate as most of the Antarctic Bottom Water (AABW) that occupies the world ocean deepest layers is likely to originate from this region. This study applies the Optimum Multiparameter water mass analysis to the Weddell deep waters in order to investigate their distribution and variability. The dataset used is based on the WOCE repeat sections in the area (SR04 and A12) from 1984 to 1998. The mean water mass distribution is consistent with previous knowledge of the region, along with high interannual variability. Regarding the temporal variability, it seems that the years of maximum Weddell Sea Deep Water (WSDW) contribution correspond to the lowest levels of Weddell Sea Bottom Water (WSBW), and vice versa. In order to identify possible forcing mechanisms for such variability, the water mass temporal anomalies were compared with oceanic and atmospheric modes of variability in that region such as the Southern Annular Mode (SAM). An apparent correlation between the SAM index temporal gradients and WSBW anomalies indicate that the Weddell Sea export of dense waters to the world ocean may be linked to that index on several time scales.

We examine Weddell Sea deep water mass distributions with respect to the results from three different model runs using the oceanic component of the National Center for Atmospheric Research Community Climate System Model (NCAR-CCSM). One run is inter-annually forced by corrected NCAR/NCEP fluxes, while the other two are forced with the annual cycle obtained from the same climatology. One of the latter runs includes an interactive sea-ice model. Optimum Multiparameter analysis is applied to separate the deep water masses in the Greenwich Meridian section (into the Weddell Sea only) to measure the degree of realism obtained in the simulations. First, we describe the distribution of the simulated deep water masses using observed water type indices. Since the observed indices do not provide an acceptable representation of the Weddell Sea deep water masses as expected, they are specifically adjusted for each simulation. Differences among the water masses’ representations in the three simulations are quantified through their root-mean-square differences. Results point out the need for better representation (and inclusion) of ice-related processes in order to improve the oceanic characteristics and variability of dense Southern Ocean water masses in the outputs of the NCAR-CCSM model, and probably in other ocean and climate models.