Can we use a simple modelling tool to validate stormwater biofilters for herbicides treatment?

This study proposes a new stormwater biofilter validation approach, using a process-based model of micropollutant removal in stormwater biofilters. The model performance was assessed against in-situ challenge tests conducted on a field biofilter under challenging operational conditions for removing four herbicides (atrazine, simazine, prometryn and glyphosate). Two-site adsorption kinetics were used on the laboratory results to estimate parameters; the estimated K_oc (soil organic carbon-water partitioning coefficient) corresponded well with literature values, while f_e (instantaneous adsorption fraction) and α_k (kinetic adsorption rate) differed from the literature. The agreement between modelled outflow concentrations and in-situ challenge tests was good for prometryn (Nash-Sutcliffe coefficient, E = 0.60) and moderate for glyphosate (E = 0.45), with up to 20% over-prediction of peak outflow concentrations. Poor performance were found for atrazine and simazine (E = 0.30). The prediction uncertainties were bigger after long dry periods, which was attributed to complex processes (biodegradation and evaporation) not captured in either the laboratory column experiments or the model.