Abstract
M.Sc.
The Vaal Dam (South Africa) and its tributaries have been extensively affected by domestic,
mining, agricultural and industrial activities, as well as the release of effluents. These
practices have contributed to large-scale algal blooms that have caused serious ecological,
aesthetic, water purification and water distribution problems.
This study addresses the need to develop a system that enables forecasts to be made
regarding potential changes in the water quality ofthe Vaal Dam, especially with regards to
predicting algal blooms. The primary aim was to develop a simple spreadsheet based model
to predict the occurrence of algal blooms and other water quality changes in the Vaal Dam,
making use of environmental parameters recorded at several sites located upstream of the
Rand Water intake point at the Vaal Dam wall. Accurately forecasting sudden changes in
water quality would enable proactive resource management, ensuring that Rand Water
maintains a high standard of potable water delivered to its customers.
Statistical model equations, to predict the concentrations of various water quality
constituents, were obtained by step-wise regression analysis. These equations were then
entered into MS-Excel spreadsheets. This allowed the input of environmental data and the
subsequent calculation of the predicted value. This also allowed for the manipulation of
various parameters to forecast the effects any changing values will have on the water quality.
These "if-then'' scenarios would be of considerable use in implementing management
measures to achieve the desired water quality. The performance of the model was
statistically tested to determine if it adequately represents the study system.
The models to determine chlorophyll-a concentration and several other water quality
constituents proved to be fairly accurate in representing the study system. However, the
model to predict nitrate concentrations did not perform satisfactorily. The limitations in
model performance were attributed to the low frequency of water quality sampling and the
effects of undetermined variables not represented by the water quality parameters selected
for model development.
The model is compact, does not require specialised software, and is applicable in practice.
The predictive and scenario forecasting abilities make this model useful for the identification,
monitoring and prediction of changes or trends in water quality over time. The benefits
arising from this model will thus contribute to more cost efficient water treatment, improved
response times to algal blooms, improved decision-making and proactive water resource
management.