Coastal water quality.
This research focuses on the pathogenic pollution of coastal recreational waters. Pollution of this resource can have serious social and economic implications. The health of the public could be compromised and there may be associated adverse impacts on the tourism industry. A section of coastline along the Durban Bight and including some of the nation's premier bathing beaches, was used for a case study. The water quality condition of the beaches was evaluated against both local and international marine recreational water quality standards. Most of Durban's bathing beaches were found to have good water quality. However beaches situated close to stormwater drains regularly experience poor water quality conditions. The relationships between beach water quality, the pollution sources and environmental factors such as rainfall were quantified. A weak correlation was found between rainfall and beach pathogenic pollution levels. No correlation was found between successive fortnightly beach samples indicating that the time scales of coastal dispersion processes are significantly shorter than the beach monitoring period. The research also indicates a need to update the SA marine water quality standards. The exclusive use of Escherichia coli (E.coli) as the indicator of faecal pollution is inconsistent with international trends towards the use of Enterococcus, which is a more robust pathogen indicator for marine environments. The main aim of the research was to develop a model to predict the water quality conditions of beaches. The Coastal Water Quality Model (CWQM) is intended to serve two functions: firstly to provide daily estimates of pathogenic pollution levels for beach management (e.g. closure under poor water quality conditions), and secondly to provide decision-makers with a tool for predicting the effects of changes on future water quality conditions. The CWQM was formulated as a stochastic state-space lumped advection diffusion model. A Kalman Filter was used for state estimation. Parameter estimation using the Extended Kalman filter was investigated but found to be unsatisfactory due to large input uncertainties and sparse measurements. An alternative statistical fitting procedure was therefore used for parameter estimation. The model was shown to produce accurate predictions of pathogenic pollution for the case study site. To further demonstrate it's utility. it was used to evaluate options for improving the poor water quality at Battery Beach. The results show that a constructed wetland could be effective in this case.