The greater Agulhas Current System, is considered to be the largest western boundary current in the Southern Hemisphere, with only the Gulf Stream, and possibly the Kuroshio, considered to be larger globally (Bryden et al., 2005). The Current System is a crucial factor for determining the mean state and variability of the regional marine environment, resources and ecosystems in the region, regional weather, as well as the global climate on a broad range of temporal and spatial scales. Due to an absence of a coherent in situ and satellite-based observing system in the area, modelling and data assimilation techniques are utilised. These both further the quantitative understanding of the ocean dynamics as well as providing better forecasts of this complicated western boundary current system. In this study, we compare two assimilation experiments using the Ensemble Optimal Interpolation (EnOI) data assimilation scheme in a regional implementation of the Hybrid Coordinate Ocean Model (HYCOM). In the first experiment, we assimilate along-track satellite sea level anomaly (SLA) data only, and in the second experiment we assimilate both along-track sea level anomaly (SLA) as well as satellite sea surface temperature (SST) data. The objectives of the study are to investigate the impacts of assimilating SST along with SLA into the regional HYCOM model, with the hopes of improving the model performance. The long term aim of this experiment is to develop a regional ocean prediction system. The additional assimilation of SST along with SLA into the HYCOM model, has improved upon the representation of the SST field across the region by reducing the error. However, with regards to velocity, surface eddy kinetic energies (EKE), as well as subsurface velocities, the updated SST model shows less improvement. A velocity bias can be seen as the reason for underperformance in these aspects. The model still struggles to recreate subsurface water masses, underestimating salinity in the upper 500 m; assimilating T/S profiles in the future could improve on this. The assimilation of SST has improved upon the SST-SSH correlation in the model, as well as the spatial distribution and accuracy. The assimilation of SST along with SLA has had many positive impacts, with unfortunately, a few negatives. The shortcomings of the numerical model will have to be improved upon and additional assimilation variables should be tested in further studies, to provide a solid forecasting system.

Bibliography: leaves 55-62. / To study the coastal hydro-climate in the central Angolan region, in situ temperatrure and satellite derived SST and in situ salinity were used to describe both the SST and SSS fields off Angola during the period 1969-1999 with some gaps. Emphasis was placed upn the Lobito coastal oceanographic station (12° 19S 13° 34E) mean temperature and salinity distributions and the seasonal and interannual temperature and salinity variability. The main aims of this study were thus to identify, quantify and analyze the above parameters and to establish their effects on the appropriate time scales of variability.

Includes bibliographical references (p. 101-108). / The aim of this study is to examine the variability in the wind regime at diurnal, seasonal, interannual and interdecadal time scales. Meteorological measurements including wind speed, wind direction and air pressure collected at Dias Point Lighthouse (260 38.094'S 150 05.612'E) at hourly to eight-hourly intervals from 1960 to 2006 are analysed. Known instrument changes in the time series are validated where possible. Predominant winds at Lüderitz blow parallel to the South to North alignment of the coast. Ekman divergence in response to longshore, equatorward wind stress drives coastal upwelling at Lüderitz, the main centre of upwelling in the Benguela Current System. Wind stress is proportional to the square of the wind speed parallel to the coast and is a proxy for upwelling. A diurnal intensification of wind speeds occurred in all seasons at Lüderitz with a concomitant change in wind direction from south in the early morning to southwest in the afternoon. Pressure changes over the continent due to daytime heating and night-time cooling of the land underlie this variability. Southwesterly winds predominate throughout the year at Lüderitz. Maximum wind stress occurs in the austral summer with a fourfold decrease in wind stress during the austral winter. Highest wind stress was recorded from November to January and lowest wind stress from May to July. The wind mixing index, a measure of turbulent mixing calculated from total wind speed cubed, follows the same seasonal pattern indicating the predominance of southerly winds. The wind minimum at Lüderitz is caused by weakened pressure gradients due to the latitudinal northwesterly shift in the position of the South Atlantic Anticyclone in winter combined with a pressure increase over the continent.

Traditional understanding of mixed-layer (ML) dynamics in the African sector of the Southern Ocean suggests that seasonal summer stratification and subsequent reduction in ML depth (MLD) is determined by the onset of a positive net heat ux. The impact of physical forcing mechanisms on the intra-seasonal variability of the ML is still relatively unknown. Recent research in the North Atlantic has highlighted the role that sub-mesoscale ML eddy dynamics has on ML stratification. It is now understood that large horizontal density gradients drive sub-mesoscale eddy formation which have been shown to result in the early onset of spring phytoplankton blooms at high latitudes. To date these ML eddies have been researched primarily in models with few observational studies available. To test the ML eddy hypothesis in the Subantarctic Zone (SAZ) we use high-resolution (~3km, 4-hourly) glider measurements between austral spring to late summer.

Bibliography: pages 64-67. / Saldanha Bay is located lOOkm north of Cape Town, along the south western coast of South Africa, at a latitude of approximately 33° S. In 1975 major harbour works, including the construction of an iron ore jetty, divided Saldanha Bay into two distinct bays. The bay to the north of the jetty has become known as Small Bay, while the bay to the south is commonly referred to as Big Bay. Big Bay is connected at its southern end to the shallow Langebaan Lagoon system, and to the west with the Benguela upwelling system (see figure 1.1). The oceanography of the shelf outside the bay is dominated by the coastal upwelling system (Shannon 1985). The equatorward winds which predominate for much of the year, drive an offshore flux of surface water, which is replaced near the coast by nutrient-rich water from deeper layers. These winds are the result of an interplay between the South Atlantic High Pressure Cell, a thermal low that forms over southern Africa in summer, coastal lows, and eastward moving extra tropical cyclones (Shannon 1985). In summer, the band of extra tropical cyclones associated with the jetstream moves to the south of its winter position, while the South Atlantic High intensifies and moves approximately 6° to the south (Preston-Whyte and Tyson 1988), creating a pressure gradient over the Benguela region. The presence of the thermal low over the adjacent subcontinent increases the existing pressure gradient, and enhances the equatorward air flow. This seasonal modulation of upwelling favourable winds results in an upwelling season that extends from about September to March (Shannon 1985). During this period, there is also synoptic modulation, provided by wind relaxation or reversal events. These events are either associated with the passage of a cyclone to the south of the continent, or with a coastal low passing along the coast (Shannon 1985).

Includes abstract. / Includes bibliographical references (leaves 49-58). / A research survey was conducted, consisting of six transects between 38°- 46°S and 38 - 41°45'E, during the austral autumn of 2007. The aim of the survey was to investigate the physical, chemical and biological dynamics of the Subtropical Convergence (STC), in the SW Indian sector of the Southern Ocean. Satellite data was obtained and in-situ data were collected. Mixed layer depth (MLD), geostrophic velocities, density and Brunt Vaisala frequencies were calculated. The STC meandered across the survey area between 41 ° - 42°15'S. The total integrated Chl-a ranged from 12.8 to 40.1 mg Chl-a/m2. The most significant correlation between biological and physical data was that of Chl-a and MLD (r=-0.374, n=45, p=0.013) over the entire survey region. Phosphate and nitrate (r=0.8779, n=45, p=0.001) measured over the survey region were strongly correlated, as expected. The surface currents showed cyclonic motion between 38° - 39°E and 38 - 42°S, with the exception of an eddy-like feature between 39.5°and 40.5"S and generally anti-cyclonic motion to the east of 39°E.

Includes bibliographical references (leaves 41-46) / The ability of the Hadley Centre Atmospheric Model version 3 (HadAM3) to capturerainfall variability patterns over Eastern Africa and the western tropical Indian Oceanduring the rainy seasons (October to December (OND) and March to May (MAM)) isassessed against data derived from National Centre for Environmental Prediction (NCEP)model and Global Precipitation Climatology Project (GPCP). The vector windsclimatology at 850hPa and 200hPa reveals some comparable patterns in both HadAM3 andNCEP. However, there are indications that the Somali Jet appears earlier and the monsooneasterly winds over the western Indian Ocean are weaker in HadAM3 than in NCEP.Empirical Orthogonal Functions (EOF) and Principal Component Analysis (PCA) derivedfrom the GPCP precipitation and NCEP Outgoing Longwave Radiation (OLR) andgeopotential anomalies were also compared to those from the HadAM3 model.

Includes bibliographical references (leaves 62-71). / Knowledge of the chemical composition of marine macroalgae is important to understand their nutritional value for abalone as well as their potential as a source of protein, carbohydrate and lipid for commercial use. This study investigates the seasonal variations in chemical composition and nutritive value of Ecklonia maxima collected from various kelp beds near commercial abalone farms on both the west and south west coasts of South Africa. It has been suggested by numerous abalone farmers that west coast kelp is not as good as the south west coast for abalone feed.

Two studies of the quantification of water masses around southern Africa are described in this thesis. The first, a pilot study, is concerned with the volumetric analysis of historic hydrographic data from the region off the South African west coast. It uses data that have been collected with various instruments over a considerable period of time and with relatively poor vertical and horizontal resolution. Nevertheless meaningful and significant results have been obtained: (i) The T/S characteristics of the upwelling domain and the open-ocean are significantly different and show no overlap. Upwelled water is derived from different parts of the South Atlantic Central Water mass. (ii) Upwelled water, irrespective of its particular origin the Central Water mass, eventually is heated to a limit of 23°C. (iii) The volume of water in each bivariate T/S class of Central Water, which is available for upwelling, are constant. This implies that the intensity of upwelling may be estimated from the salinity of the upwelled water only. The second or main study is concerned with the Agulhas Retroflection area. This is located in the broader South-East Atlantic/South West Indian ocean region, which is identified as a serious gap in a previously published fine-scale volumetric census of the world ocean. Recently collected, high quality hydrographic data make it possible to rectify this. This is the principle objective of this thesis. Some results include: I. The subtle differences between the Central water of South Atlantic Ocean origin and that of the South Indian ocean were addressed. These differences are real and distinguishable as is shown using both historic data and high quality hydrographic data. 2. For the first time, reliable quantitative estimates for the warm and thermocline water masses are produced. The warm, saline surface water of the Agulhas Retroflection area contributes very little to the overall volume of the upper 1500 m of the water column. 3. The low salinity water from south of the Subtropical Convergence is spread widely (in T/S space) on the less saline side of the three-dimensional volumetric diagrams and has very little volume (less than 1% of the total volume). 4. The more saline, higher volume section of the Central Water in this region corresponds to South West Indian Central Water. 5. Twenty five percent of the total volume of water in this region is contained in only 21 fine-scale bivariate classes. 6. Fifty percent of the water (contained in 71 classes) has a temperature below 3°C. If the lower volume. non-ranked classes were added then almost seventy percent (67,26%) of all the water in the Retroflection area would be colder than 3°C. The results of this census are compared with those of the world ocean census. Although the volume per bivariate class interval is not of the same order of magnitude, the results are remarkably similar. The water masses of the Retroflection area are less diverse than those of the world ocean.

The motivation for this thesis arises from the apparent global warming problem. It is common knowledge that the ocean is an important heat sink and that to predict global warming trends it is important to measure changes in the ocean temperature. Measurements of ocean temperature are subject to large local variability associated with mesoscale features. In order to measure global warming trends a method is required to determine large scale averages, over large oceanic ranges, in order to remove the inherent variability associated with mesoscale features, such as currents and mesoscale eddies. This requirement can be met by acoustic thermometry.