Decadal variability of the subtropical gyre and deep meridional overturning circulation of the Indian Ocean
The work presented in this Thesis concerns the large-scale circulation of the Indian Ocean and follows three lines of investigation: (i) decadal variability of the subtropical gyre circulation; (ii) decadal variability of the deep meridional overturning circulation (MOC); and (iii) the influence of diapycnal diffusivity on quasi-steady MOC states. The decadal variability of the subtropical gyre transport over the ocean interior (away from boundary currents) is investigated using hydrographic data from 32°S. Estimates of the relative gyre transports are: 41 ± 5.1 Sv (1 Sv = 106 m3s-1) for 1987, 42 ± 7.0 Sv for 1995 and 58 ± 7.0 Sv for 2002. This represents a 40% increase from 1987 to 2002. The main areas of change in the geostrophic transports are just east of Madagascar Ridge and around Broken Plateau, which is consistent with differences observed in the isopycnal depths in these areas. Maps of contoured velocity suggest that most of the change happened between 1995 and 2002, which supports the transport estimates. The 1987 and 2002 hydrographic data are then combined with a regional model of the Indian Ocean to investigate the impact that changes in conditions near 32°S might have on the deep MOC. The model has lateral open boundaries at 35°S for the Southern Ocean and 122°E for the Indonesian Throughflow. The meridional velocity field dominates over density at the southern boundary (SB) in determining the basin-wide deep circulation on decadal timescales. The initial adjustment of the deep MOC to the first 5-6 years of model integration and shows a large sensitivity to the SB conditions. With ‘best’ estimates of the flow field near 32°S the model shows a 6 Sv and 16 Sv deep MOC for 1987 and 2002, respectively. There are also changes in the zonal structure of the deep circulation. The results suggest that the Indian Ocean exhibits decadal variability in the size and structure of the deep MOC. Furthermore, the apparent inconsistency between previous non-GCM and regional GCM studies may be a result of the lateral boundary conditions, rather than a conflict in the model dynamics. 200-year model integrations suggest that quasi-steady MOC states in the Indian Ocean are reached on century time scales. The size, structure and adjustment time of the quasi-steady deep MOC are controlled by the distribution of diapycnal diffusivity (Kd). The zonal mean distribution of Kd required to support the prescribed deep inflow at the model SB can be estimated using a one-dimensional (1-D) advective-diffusive balance in isopycnal layers. The 18 Sv overturning circulation put forward by Ferron and Marotzke  (FM) collapses when their model configuration is integrated to quasi-steady state under a number of different Kd regimes. With a diagnosed Kd field only 70% of the FM circulation can be supported in quasi-steady state, and the Kd values are an order of magnitude larger than recent observations suggest. The results imply that one can get a good a priori estimate of the Kd-field required to support a quasi-steady model MOC by applying a 1-D advective-diffusive balance in isopycnal layers to the SB conditions. Overall, the research highlights the need to implement improved estimates of (nonuniform) Kd in ocean GCMs when investigating quasi-equilibrium model states.