Modelling cyanobacteria (blue-green algae) in the River Murray using artificial neural networks

In recent times, an apparent increase in the frequency and intensity of blooms of cyanobacteria (blue-green algae) in the River Murray (Australia) has caused widespread concern. When present in large numbers, they can cause serious problems for domestic, industrial, agricultural and recreational users of water, as they can produce toxins and impart undesirable tastes and odours to water. It is important to understand the relationship between the incidence of algal populations and the prevailing environmental conditions in order to prevent algal blooms from occurring. In this paper, artificial neural networks (ANNs) are used to model the incidence of a specific genus of cyanobacteria (Anabaena sp.) in the River Murray at Morgan, with the dual objectives of forecasting algal concentrations to give prior warning of impending blooms and to identify the factors that affect the blooms of Anabaena. The model inputs include weekly values of turbidity, colour, temperature, flow and the concentrations of total nitrogen, as well as soluble and total phosphorus. The results obtained are very promising as the model was able to forecast most major variations in Anabaena concentrations (timing and magnitude) for an eight-year period two weeks in advance. A sensitivity analysis carried out on the model inputs indicated that all input variables are important, with no one variable being dominant.