Modelling the leaching of imazapyr in a railway embankment.

The use of herbicides on railway tracks is known to present a risk to groundwater, but little is known of the mechanisms influencing leaching through the coarse material used to construct railway embankments. Therefore, in the present study, four different models based on the convection-dispersion equation (CDE) were compared with previously reported field data on the leaching of imazapyr. In particular, the significance of non-equilibrium processes was investigated by comparing different CDE formulations accounting for preferential finger flow, particle-facilitated transport and kinetic sorption. The traditional CDE assuming 'local equilibrium' based on 24 h batch sorption data gave poor results (model efficiency - 1.1). It strongly underestimated leaching of imazapyr in the first 4 months following application, thus confirming the importance of non-equilibrium transport processes. Accounting for short-term sorption kinetics made little difference, giving similar results to the 'local equilibrium' CDE simulation. A simulation accounting for particle-facilitated transport could accurately match this accelerated transport, and also gave the best overall fit to the data (model efficiency 0.76). However, not even this model could match the long-term retention of imazapyr residues observed close to the soil surface more than 1 year after application, and it also underestimated the time of breakthrough to groundwater. This strongly suggests that a long-term retention/sorption process not included in any of the models tested (i.e. sorption hysteresis or bound residues) acted to retard leaching. The formation of 'protected' residues was also indicated by a much slower degradation of imazapyr more than 1 year after application. Industry.