MODELING ALGAL GROWTH AND TRANSPORT IN RIVERS - A COMPARISON OF TIME-SERIES ANALYSIS, DYNAMIC MASS-BALANCE AND NEURAL-NETWORK TECHNIQUES

Algae present considerable problems for river quality managers and wat er suppliers and methods to predict their behaviour, growth and transp ort can assist in operational management. Alternative techniques exist for predicting algal response and three approaches have been compared and applied to data from six sites along the River Thames. These tech niques include time series analysis, dynamic mass balance and growth e quations and neural network approaches. It is shown that neural networ k techniques offer a new approach requiring less intuitive knowledge b ut predictive capability is not improved greatly compared to other app roaches. Neural networks enable models to be developed along all six r eaches of the River Thames.