Lognormality of trace contaminant concentrations in sewage effluents

It is important to understand the statistical distribution of monitoring data for them to be of value in determining the parameters of environmental models. No such distributional information has been available for many trace contaminants in sewage effluents. This paper applies the data of a major UK sewage works' effluent monitoring programme to determine the validity of the common assumption that data are lognormally distributed. Effluent quality was monitored at 162 wastewater treatment works over 1 year, generating over 3,000 results for each of over 40 substances, including metals, trace organic substances and pharmaceuticals. It is demonstrated that the lognormal assumption is clearly justified for the great majority of substances in the spatial case-for annual average effluent concentrations across different treatment works. In the site-specific, temporal case-for individual determinations of concentration at a single site over an annual period-lognormality is generally supported but not demonstrated so unequivocally for all site/substance combinations. The principal source of uncertainty was lack of sufficient numbers of observations reported to adequately low reporting limits.