Classification of marine water quality: first attempts using artificial neural networks with measurements from biological effects of pollutants in mussels (Mytilus edulis)
Dr Elisa Ravagnan
International Research Institute of Stavanger, Norway
Date: Friday 13 February 2009
Time: 3pm to 4pm
Location: P335 LT (Poole House)
Abstract: Understanding pollution problems is highly interdisciplinary, combining pollutant chemistry, ecotoxicology, ecology, physical oceanography, estuarine science, sedimentology and mathematical modelling. There is an increasing need, from a political and social point of view, to develop adequate tools for the identification, estimation, comparative assessment and management of the risks posed by chemical pollution present in the marine environment. Risk assessment can not be based entirely on chemical analysis as this in itself will not give any information regarding the health of organisms; for this reason, measurement of biological effects of pollutants (biomarkers) has become of major importance for the assessment of the quality of the marine environment.
Although it is easy to compare variation in the values of a single biomarker, no objective system is available yet to integrate variation of several biomarkers (battery) in order to interpret correctly the health condition of the organisms, and consequently of the environment where they live.
A first attempt to classify marine water quality has been made using Artificial Neural Networks. Feed-forward networks and probabilistic networks have been applied to biomarker data alone or combined with chemical data, to study the capability of the networks in extracting information from non linear and complex databases.
The first results show the potential for using ANNs with ecotoxicological data, even if, in this case, a poor generalization was produced because of scarce data and high individual variability.
Speaker's bio note: Elisa Ravagnan has an MS degree (2000) and a PhD (2004) in Environmental Sciences from University of Venice (Italy). MS final dissertation: "The phenomenon of gelatinous aggregates in the Adriatic Sea: study of its genesis in correlation with climatic factors"; PhD thesis: "Acquisition and processing of physical, chemical and biological measurements of the marine environment. Study and use of modelling techniques to marine ecosystems".
She applied statistical methods and hydrodynamic/ecological models to chemical, physical and biological data collected in marine environments.
She is currently research scientist at the International Research Institute of Stavanger (IRIS, Norway), using statistical methods and modelling tools with ecotoxicological marine data.
