Inverse modelling of herbicide transport during transient flow in vegetated weighable lysimeters

  

Metolachlor, terbuthylazine, nicosulfuron and prosulfuron are among the typical crop protection products used on maize plantations. Leaching of those lead to a decrease of groundwater and surface water quality. Often measured individual pesticides and their metabolite concentrations in drinking water are exceeding the limit established by the European Council in drinking water. In our work we investigate contamination potential of these four herbicides to groundwater bodies using lysimeter studies and numerical modelling with HYDRUS-1D. The four herbicides were applied at specific times in the vegetation phase over a period of three years on two lysimeters located in Wielenbach, Germany. The studied lysimeters contain soil cores dominated by sandy gravel (Ly1) and clayey sandy silt (Ly2) and are both vegetated with maize. To identify governing transport and fate processes in the unsaturated zone of the lysimeters and determining dynamics and rates of these, different model structural approaches were compared. In a first step we have characterized soil hydraulic and transport parameters of each soil core by investigating stable water isotopes (δ¹⁸O and δ²H). For Ly2, model performance was improved by considering immobile water in a dual-porosity approach whereas for Ly1 a single-porosity approached seemed to yield satisfying results. This might be explained by a higher fraction of fine particles in Ly2 which can be available for water storage. Based on these findings reactive transport parameters were fitted also for root water and chemical plant uptake, sorption, and biodegradation.

It was found that sorption plays a significant role in herbicide transport whereas root water uptake and chemical plant uptake is of minor influence. Metabolite formation was observed; however, biodegradation seems to show minor influence, only, which is also reflected by measured carbon isotopes (slight δ13C increase). Non-equilibrium and non-linear sorption were compared leading to no significant difference in model results. This was especially surprising for the charged herbicides prosulfuron and nicosulfuron. Measured herbicide concentration peaks in seepage water seemed to be connected in time with higher amounts of precipitation events indicating the influence of preferential flow. Such influences could be considered in a dual-permeability flow model setup which however was not available for stable water isotope modelling in HYDRUS-1D. Contrary to our expectations, the coarser soil of Ly1 did not lead to an increase in leached herbicides which might be explained by a higher organic matter content and thus higher sorption.