Spatial variability of ecosystem exposure to home and personal care chemicals in Asia

Abstract It is well recognized that there are currently limitations in the spatial and temporal resolution of environmental exposure models due to significant variabilities and uncertainties in model inputs and parameters. Here we present the updated Pangea multi-scale multimedia model based on the more spatially resolved, catchment-based hydrological HydroBASINS dataset covering the entire globe. We apply it to predict spatially-explicit exposure concentrations of linear alkylbenzene sulphonate (LAS) and triclosan (TCS) as two chemicals found in homecare (HC) and personal care (PC) products in river catchments across Asia, and test its potential for identifying/prioritizing catchments with higher exposure concentrations. In addition, we also identify the key parameters in the model framework driving higher concentrations and perform uncertainty analyses by applying Monte Carlo simulations on emissions and other non-spatial model inputs. The updated combination of Pangea with the HydroBASINS hydrological data represents a substantial improvement from the previous model with the gridded hydrological dataset (WWDRII) for modelling substance fate, with higher resolution and improved coverage in regions with lower flows, with the results demonstrating good agreement with monitored concentrations for TCS in both the freshwater (R2 = 0.55) and sediment (R2 = 0.81) compartments. The ranking of water basins by Predicted Environmental Concentrations (PECs) was similar for both TCS and LAS, with highest concentrations (Indus, Huang He, Cauvery, Huai He and Ganges) being one to two orders of magnitude greater than the water basins with lowest predicted PECs (Mekong and Brahmaputra). Emissions per unit volume of each catchment, chemical persistence, and river discharge were deemed to be the most influential factors on the variation of predicted PECs. Focusing on the Huang He (Yellow River) water basin, uncertainty confidence intervals (factor 31 for LAS and 6 for TCS) are much lower than the variability of predicted PECs across the Huang He catchments (factors 90,700 for LAS and 13,500 for TCS).