JournalDisplay Article
Wednesday, November 21, 2018

Spatial Transferability of PAH Data of the German ESB by Artificial Neural Networks



Universität Trier, Trier, Germany

The need to have exhaustive data available for environmental assessment is contrary to the local character of the measurement methods for most environmental monitoring programs. Against this background, the spatial transferability of data from the German Environmental Specimen Banking Program (German ESB) was investigated by creating a model that predicts polycyclic aromatic hydrocarbon (PAH) concentrations for sites with missing monitoring data. In particular, we tested if data measured in one representative of a certain ecosystem type may be transferred to further representatives of the same ecosystem type. Modelling was based on real polycyclic aromatic hydrocarbon pollution and on the fundamental assumption that the ecological structure of an ecosystem has a dominant impact on pollutant concentrations. To manage the complexity of processes and factors influencing the pollution of ecosystems, which are far from well-known, artificial neural networks (ANNs) were used to generate a suitable estimation model. It was shown that concentrations of more volatile PAHs could be reliably estimated for sites with unknown concentrations using the ecological structure surrounding these sites and an ANN.



Download the full article below.


Copyright © 2014 by ISEI |Privacy Statement|Terms Of Use