Predicting the probability of concentrated flow occurrence using a dynamic approach on high-resolution soil surface data

Concentrated flow erosion, one of the major mechanisms of erosion by water, is often an indicator of accelerated soil degradation and off-site sediment delivery. While flow hydraulics in a concentrated flow pathway is a well-studied mechanism, much less is known about the probability of flow concentration and the spatial extent of concentrated flow networks. In this study, we develop a methodology to study concentrated flow processes from three-dimensional soil surfaces by applying virtual rainfalls followed by flow routing and channel cross section analyses in digital space. Channel width estimates derived from a proposed computer vision approach were consistent with previously published channel width prediction models. The approach was successfully tested to quantify the probability of concentrated flow occurrence on 6 m × 2 m rangeland erosion plots. An equivalent flow path area (Afpe) was defined and used to calculate the probability of flow concentration (Pfc) with a logistic model predicting the odds that concentrated flow areas exceed a threshold fraction of Afpe. The study also provides the first conceptual framework to compute and predict the spatial extent of the concentrated flow network as the number of flow paths (NFP). Pfc and NFP were positive functions of runoff rate, bare ground, canopy cover (NFP only), and clay content while rock cover had an attenuating effect on Pfc. These results are consistent with previously published functions for the probability of flow concentration.