“…Our work attempts to do so through assembly of a large dataset comprised of 940 highway runoff events, which report a total of eight separate physicochemical parameters (TSS, Cu, Zn, Pb, Cr, TN, TKN and TP), as well as five of the parameters widely understood to govern the pollutant concentrations in runoff (rainfall depth [RD], rainfall intensity [RFI], average annual daily traffic [AADT], antecedent dry period [ADP] and seasonality [season]). Acknowledging these uncertainties, the predictive utility trained ANNs applied in the context of complex, nonlinear systems cannot be understated, and this makes their use worth investigating (Kisi et al, 2013;Trenouth and Gharabaghi, 2015a).…”