Transport model-based method for estimating micropollutant removal efficiency in riverbank filtration

Abstract

Riverbank filtration is a cost-effective and efficient method for drinking water production, using the natural filtration capacity of the river gravelbed. Removal efficiency for organic micropollutants (OMP) in field studies is generally calculated by comparing the concentrations measured in surface water and in the wells either on the same day or with a shift of fixed time interval, neither of which can account for the variability of surface water quality and travel time in the aquifer. The present study proposes a novel method based on travel time distribution determined by a numerical transport model with a hypothesis that it will provide more reliable estimate for OMP removal. The model was developed for two production sites of Budapest Waterworks, Hungary on Danube River. River water and riverbank filtered well water were sampled regularly for one year (158 samples each) and analysed for 41 OMPs (pesticides, pharmaceutical residues and industrial pollutants). Nineteen pollutants were detected in >50% of the well water samples. Median removal rates were 4-97%, while the concentration of five compounds increased in some wells. Removal rates of telmisartan, tramadol, sulfamethoxazole, 4-methyl-benzotriazole, 5-methyl-benzotriazole and desethyl-terbuthylazine correlated negatively to redox potential (|r|=0.456-0.805). Median travel time increased after high flow events resulting in reduced removal of telmisartan, tramadol, 4-methyl-benzotriazole and desethyl-terbuthylazine (|r|= 0.435-0.661). Removal of diatrizoate, iopamidol, tramadol and benzotriazole increased with distance from the shore (148 vs 395 m) by 25%, 28%, 8%, 16%, respectively. Background groundwater contamination increased pesticide concentration in the wells located in agricultural areas 1.5-5-fold compared to river water. The model-based method gave more consistent results compared to traditional calculations for OMP removal efficiency during the sampling campaign and allowed for estimating the impact of various environmental factors.