Urban runoff pollution can essentially be characterised by fluid quantities and pollutant concentrations. It has been possible to construct models accounting for variations in runoff quantities with some success. However, although several pollutant storage and transport mechanisms have been postulated there still remains substantial unexplained variation in pollutant concentrations. Through a series of well established multivariate pattern recognition techniques the present study has aimed at disclosing the underlying structure of systematic variations in the event mean concentrations (EMC) of pollutants in combined sewers during rainfall. The statistical methods that have been applied to the pollutant concentration variables are factor analysis, cluster analysis, distribution analysis and correlation analysis. The event mean runoff data considered includes eleven pollutant variables originating from five combined sewer catchments in Denmark and in the Netherlands. The combined results of the analyses support earlier findings that EMCs are best described by bimodal or mixture distributions, and further suggest that event based pollutant modelling could be improved through a recognition of these characteristics.

You do not currently have access to this content.