<p><strong>Abstract.</strong> Evolution of landscape heterogeneity is controlled by coupled Earth system dynamics, and the resulting process complexity is a major hurdle to cross towards a unified theory of catchment hydrology. The Biosphere 2 Landscape Evolution Observatory (LEO), a 334.5 m² artificial hillslope built with homogeneous soil, may have evolved into heterogeneous soil during the first experiment driven by an intense rainfall event. The experiment produced predominantly

more »

... eepage face water outflow, but also generated overland flow, causing superficial erosion and the formation of a small channel. In this paper, we explore the hypothesis of incipient heterogeneity development in LEO and its effect on overland flow generation by comparing the modeling results from a three-dimensional physically based hydrological model with measurements of total mass change and seepage face flow. Our null hypothesis is that the soil is hydraulically homogeneous, while the alternative hypothesis is that LEO developed downstream heterogeneity from transport of fine sediments driven by saturated subsurface flow. The heterogeneous case is modeled by assigning saturated hydraulic conductivity at the LEO seepage face (<i>K</i>_sat,sf) different from that of the rest (<i>K</i>_sat). A range of values for <i>K</i>_sat, <i>K</i>_sat,sf, soil porosity, and pore size distribution is used to account for uncertainties in estimating these parameters, resulting in more than 20 000 simulations. It is found that the best runs under the heterogeneous soil hypothesis produce smaller errors than those under the null hypothesis, and that the heterogeneous runs yield a higher probability of best model performance than the homogeneous runs. These results support the alternative hypothesis of localized incipient heterogeneity of the LEO soil, which facilitated generation of overland flow. This modeling study of the first LEO experiment suggests an important role of coupled water and sediment transport processes in the evolution of subsurface heterogeneity and on overland flow generation, highlighting the need of a coupled modeling system that integrates across disciplinary processes.</p>