Variability and climate change signals in the Southern Ocean in the CSIRO and Antarctic CRC coupled ocean-atmosphere model.

To investigate the variability and potential climate change impacts in the Southern Ocean I use output from the c10, c15 and c16 model runs of the CSIRO and Antarctic CRC coupled ocean-atmosphere model. 300 years of the output is reported here, after a spin up of approximately 1000 years. Modifications over the earlier versions of this model (c10 and c15), including improved topography in waters south of Australia make the c16 model run more suitable for examining natural variability in the southern ocean. We use potential temperature, salinity and velocity (21 depths, 66 longitudes and 29 latitudes in the Southern Hemisphere Ocean) and surface heat flux (66 longitudes and 29 latitudes in the Southern Hemisphere). Hovmoeller diagrams of potential temperature anomalies along a streamline show circumnavigating propagation. At 400m, the strongest anomaly timescales are 4-5 years (interannual anomalies) and take approximately 20 years to travel around the globe. The anomaly strength is not uniform along this streamline and the strongest anomalies occur in the longitude band 100 to 200¬¨‚àûE. Below 1000m, decadal variability is the dominant signal and persists to at least 2000m. Scatterplots show a clear relationship between ACC current speed and phase speed throughout the water column. This relationship is strongest at intermediate depths (500 to 1500m) where anomalies are most strongly advected and least subject to other influences such as convection, mixing and mid-ocean ridges. Using HEOF analysis revealed two dominant modes of variability in the Southern Ocean. First, a mode characterised by zonal wavenumber 3, interannual anomaly in temperature and salinity on shallow to intermediate depth and density surfaces. Interannual variability of wavenumber 2 or 3 has been described in studies using observations or models of the Southern Ocean. Also, significant correlation between temperature and pressure anomalies occurred on density surfaces. The pressure pattern is 180vÄv¿ out of phase with the temperature changes on density surfaces. This is similar to a mechanism proposed by (White et al., 1998) for the ACW. The second mode is characterised by zonal wavenumber 2, decadal variation at all intermediate to deep depths and density surfaces. Overall, the decadal signal is the dominant feature of the Southern Ocean, with more total energy in this mode throughout most of the water column. Variability in several atmospheric properties including wind stress, heat and salinity fluxes is examined. The heat flux and meridional wind component showed dominant modes of interannual variability with zonal wavenumber 3. These two properties have been shown to play a significant role in determining SSTs in the Southern Ocean. The results indicates that heat flux and meridional wind are important for creating zonal wavenumber 3 interannual variability in resultant SSTs. The second mode, consists of a decadal signal with dominant zonal wavenumber 2. This suggests that slower timescales may occur in the ocean due to natural filtering of anomalies such that interannual signals are absorbed and only the longer term decadal signals remain. The change in dominance from a signal with spatial structure zonal wavenumber 3 to zonal wavenumber 2 is more complex. Exploring the possible reasons for a zonal wavenumber 2 structure required investigation of convection regions in the Southern Ocean. An estimate of mixed layer depth revealed two dominant regions of wintertime convection. The first begins in the southeast Indian Sector of the Southern Ocean and continues to below Australia. The second region is the Pacific Ocean to the east of Drake Passage. These two regions are important for generation of SAMW and AAIW respectively. In essence, these two regions of convection act like highways in a myriad of surburban streets whereby anomalies at the surface can quickly access the deeper depths (compared with anomalies at the surface in adjacent regions of low convection). A simple model is used to test the assumption (Chapter 4). Forcing this model with interannual, zonal wavenumber 3 heat flux anomalies and applying a convection system of two convective regions, we obtain a resultant subsurface anomaly with zonal wavenumber 2. The hypothesis that the distribution of convection regions influences the zonal wavenumber structure of anomaly was shown to be reasonable. For the third aim, I compared warming and deepening signals seen in the transient model run with a control run. There is a consistent pattern of cooling on isopycnals in shallow/Mode waters and warming on isopycnals in AAIW in the transient run. This coincides with a general deepening of density surfaces. The fingerprint experiment reveals statistical significance in these results indicating that the simulation of CO‚Äövává changes indeed produces changes to Southern Hemisphere water masses with the described spatial pattern. Qualitative comparison with studies of observed changes show a similar pattern of change.