Continental scale inverse modeling of common organic water contaminants in European rivers Alberto Pistocchi * , 1 , Dimitar Marinov, Saudade Pontes, Bernd M. Gawlik European Commission, Joint Research Centre, Institute for Environment and Sustainability, via E. Fermi, I-21020 Ispra (VA), Italy article info Article history: Received 6 April 2011 Received in revised form 5 October 2011 Accepted 26 October 2011 Keywords: Inverse modeling of contaminants Emission factors Pan-European assessment Personal care products Pharmaceuticals Diffuse pollution abstract The paper presents an analysis of measured riverine concentrations of 16 common organic water contaminants. From observed concentrations we back-calculate emissions and chemical half lives through a simple inverse model. The analysis does not allow identifying a single half life/emission factor combination, but a set of combinations which are Pareto-optimal (or “non-dominated”). The approach is shown to provide a rational basis for the screening of chemicals in rivers: with reference to the 16 chemicals considered here, estimated emission factors and half lives are consistent with the ones re- ported in other studies. For more precise estimates, prior knowledge about either emission factors or half lives is necessary. For the considered chemicals, loads to European seas can be subsequently estimated with an uncertainty usually within a factor of 2. The approach can be proposed for the inventorying of catchment-specific chemical pollutant emissions required for European environmental policies. Ó 2011 Elsevier Ltd. All rights reserved. 1. Introduction Among the many chemicals emitted to the environment, phar- maceuticals, personal care products and other substances widely used in households and diffuse activities are increasingly an issue of concern because of their potential adverse impacts on ecosystems (European Environment Agency; Cunningham et al., 2006; Fent et al., 2006; Kidd et al., 2007; Kostich and Lazorchak, 2008) and possibly human health (Kostich and Lazorchak, 2008; Schwab et al., 2005; Brooks et al., 2009). Their environmental fate and impacts have been characterized in the last years through a number of studies aimed at understanding their sources, environmental behavior, mass balances and exposure routes (Brooks et al., 2009; Ashton et al., 2004; Bendz et al., 2005; Benotti and Brownawell, 2009; Carballa et al., 2007; Tixier et al., 2003; Heberer, 2002; Heberer and Feldmann, 2005; Jonkers et al., 2009; Kasprzyk-Hordern et al., 2009; Löffler et al., 2005; Schowanek and Webb, 2002). In order to carry out meaningful risk assessment studies on water contaminants, it is essential to know what concentrations wildlife may be exposed to. As a complement to monitored concentrations, water quality models are used on purpose to generate predicted environmental concentrations (PECs) from estimated point or diffuse emissions of chemicals to the environ- ment (e.g. Heberer, 2002, Johnson and Williams, 2004; Coetsier et al., 2009; Bound and Voulvoulis, 2006; Tauxe-Wuersch et al., 2005; Johnson et al., 2007 , 2008; Ort et al., 2009; Sumpter et al., 2006; Wind et al., 2004; Feijtel et al., 1998). Several models have found broad application in Europe, among which the GREAT-ER model (Feijtel et al., 1998). Modeling studies have shown that concentrations of water contaminants such as pharmaceuticals in waste water treatment plant (WWTP) effluents and in surface waters can be predicted with reasonable accuracy when realistic data on chemical emissions and water discharge are available. However, the accuracy of PECs, both in absolute values and spatial distribution, depends critically on the assumed contaminant emissions, besides other variables such as chemical removal effi- ciency in distinct WWTPs (in turn depending on the treatment technology, management and even local weather conditions), physico-chemical properties, the variability of hydrologic condi- tions in rivers, etc. Emissions are usually estimated based on chemical use data, which are in turn rather uncertain, particularly when mass fluxes such as improper disposal of chemicals by the users assume a high importance. Besides their importance in modeling, emissions are a key policy variable as, for most chem- icals, restriction is the only viable control action towards reducing environmental concentrations and risks. We may argue that, when accurate emission estimates are not available, any model application is likely to suffer from uncertainties * Corresponding author. E-mail address: alberto.pistocchi@gecosistema.it (A. Pistocchi). 1 Present address: GECOsistema srl, viale G.Carducci, 15 e 47023 Cesena, Italy. Contents lists available at SciVerse ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/locate/envpol 0269-7491/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2011.10.031 Environmental Pollution 162 (2012) 159e167