<p>Sediment connectivity is highly influenced by landscape patchiness. In particular, linear features such as roads, ditches, and terraces, modify landscape patterns and affect sediment transport from hillslopes to surface waters. Connectivity patterns are commonly assessed by spatially-distributed models, which rely on semi-qualitative indices or numerical simulations of soil erosion and sediment transport. However, model-based connectivity assessments are hindered by the uncertainty in model structure and parameter estimation. Moreover, representing linear landscape features is often limited by the spatial resolution of the model input data. Here we demonstrate how a global sensitivity analysis of the WaTEM/SEDEM model can be used to improve our understanding of sediment connectivity in patchy agricultural catchments of the Swiss Plateau. Specifically, we explored model structural connectivity assumptions regarding road drainage and the presence of edge-of-field buffer strips, as well as the uncertainty in the input data, by means of a Monte Carlo simulation and a high resolution 2 m x 2 m DEM. Our results showed that roads are the main regulators of sediment connectivity in ameliorated Swiss landscapes. That is, our sensitivity analysis revealed that assumptions about how the road network (dis)connects sediment transport from cropland to water courses had a much higher impact on modelled sediment loads than the uncertainty in model parameters. These results illustrate how a high-density road network combined with an effective drainage system increases sediment connectivity from arable land to surface waters in Switzerland. Additionally, our approach underlines the usefulness of sensitivity and uncertainty analysis for identifying relevant processes in model-based sediment connectivity assessments.</p>
Derivative‐Based Global Sensitivity Analysis: Upper Bounding of Sensitivities in Seismic‐Hazard Assessment Using Automatic Differentiation