Predicting critical source areas for diffuse herbicide losses to surface waters: Role of connectivity and boundary conditions

Field studies have shown that diffuse herbicide losses to surface waters often originate from a limited part of a catchment only. These critical source areas (CSA) are characterized by the occurrence of fast-flow processes like surface runoff or macropore flow to tile drains. Moreover, topographic barriers often interrupt the overland flow into the adjacent brook and lead to internal sink areas. From there artificial drains can create a shortcut to the stream network. We tested the predictability of CSA with a modified version of the hydrological Soil Moisture Distribution & Routing model (SMDR). The study area was a small agricultural catchment in Switzerland, in which herbicide losses to surface waters had been experimentally investigated. The small-scale structures, relevant for surface connectivity, were derived using terrain analysis algorithms based on a high-resolution digital elevation map. The analysis showed that surface runoff from a large part (66%) of the study area cannot reach the stream network. Based on prior parameter estimates, the simulated discharge agreed well with the measured values for spring 2000 (Nash–Sutcliffe coefficient of 0.88). Surface saturation was mainly predicted for Gleysol areas, whereby topography was the main driver, and not soil properties. The spatial distribution of CSA and predicted runoff processes agreed well with measured herbicide losses. The main losses were predicted for connected areas prone to surface runoff; additional drained sinks (reinfiltration areas on unconnected areas) were identified as potential risk areas. The predictions are subject to substantial uncertainties. A local sensitivity analysis revealed that the model is most sensitive to parameters describing the lower boundary of the soil (deep percolation and depth of the soil). The spatial extent and the efficiency of the drainage system are also crucial for the spatial variability of herbicide losses.