Use of low-cost photogrammetry to explore the link between soil microtopography and overland flow on hillslopes

The detachment of soil by flowing water on a hillslope is dependent upon the hydraulics of shallow overland flows. The role of microtopography on controlling the spatial patterns of these flows is poorly understood. This study aimed to improve this understanding by undertaking a series of controlled laboratory tests using an experimental plot, consisting of a soil box, a rainfall simulator and overland flow generator. A particular emphasis was placed on developing low-cost photogrammetric techniques to provide novel, high spatial resolution measurements of surface DEMs, 2D flow velocities and distributions of inundation levels and fluid drag. Overland flow experiments were run over surfaces with differing topographies to further examine the effect of discharge and inundation ratio on the spatial patterns of overland flow velocities, hydraulic resistance and fluid drag. Structure-from-Motion (SfM) was used to measure the microtopography of soils eroded by high intensity rainfall events. An accuracy test showed, for the first time, that SfM is capable of producing highly accurate measurements of microtopographic features at a scale similar to those found on rilled soil surfaces. The results also provided crucial information on how to optimize the acquisition of SFM-imagery in different lightning situations and using different cameras. Analysis of the DEMs of the eroded surfaces revealed the potential to estimate soil surface roughness and the distribution of elevations based on plot slope. A low-cost Particle Tracking Velocimetry system was developed to provide novel, detailed 2D surface vector fields, co-located with the acquired DEMs. The results showed that streamwise and lateral velocities have high spatial variability, revealing the flow was 2D. The distribution shape of streamwise velocities was dependent upon discharge and thus relative submergence; the higher submergences, the spatial variability in streamwise velocity was higher and the distributions were wider, flatter and more skewed. The correlation between these distributions and the surface elevations was not sufficiently high to suggest that overland flow velocity can be predicted well by surface elevations alone. On the other hand, local roughness was better correlated, revealing that local variations in surface elevations are more important in controlling spatial patterns of overland flow velocity. The hydraulic resistance, estimated by the Darcy-Weisbach friction factor, varied strongly with the Reynolds number when the flows were fully turbulent but not when they were close to being transitional, such as in shallow, interrill flows. Surface roughness and relative submergence were found to have a greater or equally strong control as Reynolds number on hydraulic resistance. Drag force varied considerably over eroded soil surfaces, with some areas experiencing forces an order magnitude higher than others. Areas of high drag forces correlated with the walls of rills and with areas in which flows converged from different rills. Thus, the distributions of drag force provide some explanation for why rills expand, deepen and migrate, as well as why erosion rates are so variable over hillslopes. The drag force distributions were well-approximated by a two-parameter Weibull distribution, whose parameters were heavily dependent upon surface roughness and hydraulic resistance. The strength of these relationships revealed the potential to predict the drag force distribution, and thus spatial patterns of hydraulic resistance and erosion, based on these simple parameters. Overall, the study has provided the first detailed study of the effect of microphotography on overland flow patterns, revealing new understanding of the statistical properties of eroded surfaces and the control of roughness on the distribution of 2-D velocities, drag force and hydraulic resistance. These results have important implications for the modelling of overland flow and resultant flow detachment. In doing so, the study has demonstrated the potential of low-cost photogrammetric techniques for providing new understanding of the physics of overland flows, and thus, in the future, driving the development of more physically-based erosion models