Modelling non-point source pollution of rivers in the UK and Colombia

Mathematical fate models have been developed and validated to simulate the transport of contaminants in temperate regions but little is known about their applicability in the tropics. Different models were applied to simulate brominated flame retardants in Colombia and the UK and to identify differences in model application and drivers of emissions in both regions. Emissions of decabromodiphenyl ether (decaBDE) in Colombia and the UK were estimated and suggested large releases to wastewater textile back-coating and waste management stages. Emission data were used to study the partitioning of the flame retardant with a fugacity approach. Fugacity results from Colombia were in agreement with sediment concentrations from the literature for the outlet of the River Magdalena. GREAT-ER was also applied to simulate decaBDE emissions in the Calder catchment; the model showed good potential for the simulation of the flame retardant. Monitoring of polybrominated diphenyl ethers (PBDEs) in sediments in the Calder showed that decaBDE represented the vast majority of PBDEs analysed (>90%) with increasing concentrations moving downstream. A modelling framework with field-scale models using MACRO was developed to simulate transport of six contrasting herbicides targeted by a management programme in the Wensum catchment in eastern England. The catchment-scale model SPIDER was also used for comparison. Preferential flow was the main driver of pesticide transport to water. A fairly good simulation of the flow was achieved (model efficiency, E = 0.6 for MACRO and 0.4 for SPIDER) but variability in pesticide simulations was observed due to uncertainties in input parameters. In-stream processes had little effect on pesticide simulations from either model. Modelling showed that most of the observed reductions in pesticide transport to the river (ca. 80% decrease between 2006 and 2011) can be explained by changes in weather and flow in the catchment during the study period, but an influence on management practices cannot be excluded. AnnAGNPs was applied to simulate triazine loss to the River Cauca from sugarcane, maize and sorghum in the Cauca Valley of Colombia. Runoff was the main driver of pesticide emissions to water. Satisfactory simulation and validation of the hydrology was achieved after little calibration (E = 0.7). A fairly good simulation of pesticides was generally achieved, but some patterns in the measured data could not be simulated. Use of grab samples resulted in uncertainty in measured concentrations. Implementing best management practices was predicted to result in a 78% reduction in triazine losses, whilst replacing triazine herbicides resulted in an 87% reduction when expressed as a proportion of the total pesticide applied. Uncertainty analyses of sensitive input parameters were carried out for the applied models. Their impact on simulations was chemical- and situation-specific. Recommendations for future research are provided to improve modelling of chemical fate in contrasting situations.