Generic Artificial Neural Network Framework for Habitat Assessment and Prediction of Australian Stream Systems

Abstract: The Stream Decision Support System (SDSS) is taking advantage of both supervised and non-supervised artificial neural networks (ANNs) for stream assessment and prediction by an integrated approach. Non supervised ANNs were applied for patterning the natural variability in stream macroinvertebrate communities in Queensland. Supervised ANNs were developed for the prediction of the occurrence of stream macroinvertebrates in Victoria based on “clean-water ” approach. Supervised ANNs were also applied for the prediction of taxonomic richness of native macrophytes and macroinvertebrates in the stream system of NSW by means of multi-layer perceptron ANN. The future development of the SDSS and its applicability for environmental management is discussed