ARTIFICIAL NEURAL-NETWORK APPROACH FOR MODELING AND PREDICTION OF ALGAL BLOOMS

Following a comparison of current alternative approaches for modelling and prediction of algal blooms, artificial neural networks are introd uced and applied as a new, promising model type. The neural network ap plications were developed and validated by limnological time-series fr om four different freshwater systems, The water-specific time-series c omprised cell numbers or biomass of the ten dominating algae species a s observed over up to twelve years and the measured environmental driv ing variables. The resulting predictions on succession, timing and mag nitudes of algal species indicate that artificial neural networks can fit the complexity and nonlinearity of ecological phenomena apparently to a high degree. (C) 1997 Elsevier Science B.V.