Search

The influence of the initial soil surface characteristics and rock fragment coverage on precipitation-driven soil erosion was investigated through laboratory experiments on a 6-m × 1-m soil parcel with mild slope. A sequence of four experiments, consisting of 2-h precipitation followed by 22 h of air drying, was conducted. The precipitation rates used were 28, 74, 74 and 28 mm h-1. In each experiment, one flume was bare while the other had 40% rock fragment coverage. The soil was hand cultivated and smoothed before the first event (named H7-E1). Experiments were designed such that for H7-E1 steady state concentrations were obtained at the end of the erosive event only for the finer sediments. Results suggest that steady state concentrations were mainly controlled by the rainfall intensity. The sediment yields during the transient phase at the beginning of each erosive event were instead sensitive to the initial soil condition, in particular to whether steady-state was reached in the previous event. If steady-state concentrations were reached for a particular size class, that class’s effluent concentration peaked rapidly in the next rainfall event, then declined gradually to its steady-state value. When the stationary behaviour was not reached, the subsequent event produced effluent concentrations that increased gradually to steady state. Steady state concentrations were achieved faster in the presence of stones on the soil surface compared with bare soil condition. The suitability of the Hairsine and Rose model to analyse the sediment concentration of the individual size classes was evaluated. The model could not predict in detail the erosion yields when topsoil was heavily modified by rainfall (H7-E1 and H7-E2), while a satisfactory comparison was observed for the last two experiments (H7-E3 and H7-E4), in which the soil was less modified by raindrop impact. Model application indicates that the contribution of the original soil to the eroded sediment decreased gradually during the sequence of experiments, while that of the deposited layer increased.

Soil erosion is a function of many factors and process interactions. An erosion event produces changes in surface soil properties such as texture and hydraulic conductivity. These changes in turn alter the erosion response to subsequent events. Laboratory-scale soil erosion studies have typically focused on single independent rainfall events with constant rainfall intensities. This study investigates the effect of multiple time-varying rainfall events on soil erosion using the EPFL erosion flume. The rainfall simulator comprises ten Veejet nozzles mounted on oscillating bars 3 m above a 6 m × 2 m flume. Spray from the nozzles is applied onto the soil surface in sweeps; rainfall intensity is thus controlled by varying the sweeping frequency. Freshly-prepared soil with a uniform slope was subjected to five rainfall events at daily intervals. In each 3-h event, rainfall intensity was ramped up linearly to a maximum of 60 mm/h and then stepped down to zero. Runoff samples were collected and analysed for particle size distribution (PSD) as well as total sediment concentration. We investigate whether there is a hysteretic relationship between sediment concentration and discharge within each event and whether the erosion response changes from event to event. Trends in the PSD of the eroded sediment are discussed and correlated with changes in erosion response. Close-up imagery of the soil surface following each event highlight changes in surface soil structure with time. This study enhances our understanding of erosion processes in the field, with corresponding implications for soil erosion modelling.

Soil surface coverage has a significant impact on water infiltration, runoff and soil erosion yields. In particular, surface stones protect the soils from raindrop detachment, they retard the overland flow therefore decreasing its sediment transport capacity, and they prevent surface sealing. Several physical and environmental factors control to what extent stones on the soil surface modify the erosion rates and the related hydrological response. Among the most important factors are the moisture content of the topsoil, stone size, emplacement, coverage density and soil texture. Owing to the different inter-related processes, there is ambiguity concerning the quantitative effect of stones, and process-based understanding is limited. Experiments were performed (i) to quantify how stone features affect sediment yields, (ii) to understand the local effect of isolated surface stones, that is, the changes of the soil particle size distribution in the vicinity of a stone and (iii) to determine how stones attenuate the development of surface sealing and in turn how this affects the local infiltration rate. A series of experiments using the EPFL 6-m × 2-m erosion flume were conducted at different rainfall intensities (28 and 74 mm h-1) and stone coverage (20 and 40%). The total sediment concentration, the concentration of the individual size classes and the flow discharge were measured. In order to analyze the measurements, the Hairsine and Rose (HR) erosion model was adapted to account for the shielding effect of the stone cover. This was done by suitably adjusting the parameters based on the area not covered by stones. It was found that the modified HR model predictions agreed well with the measured sediment concentrations especially for the long time behavior. Changes in the bulk density of the topsoil due to raindrop-induced compaction with and without stone protection revealed that the stones protect the upper soil surface against the structural seals resulting in negligible changes in the bulk density during the erosion event. Since the main process contributing to surface sealing development is the compaction due to the raindrop kinetic energy and associated physico-chemical changes, the protection provided by the stone cover is consistent with the area-averaging approach used in applying the HR model.

Two laboratory flume experiments on the effect of surface rock fragments on precipitation-driven soil erosion yields were carried out. The total sediment concentration, the concentration of seven individual size classes and the flow discharge were measured. Digital terrain models (DTMs) were generated before and after one of the experiments. The results revealed that the rock fragments protected the soils from raindrop detachment and retarded the overland flow, therefore decreasing its sediment transport capacity. Rock fragments were found to affect selectively the different size classes in a manner that changed according to the time scale. For short times, the rock fragment coverage reduced erosion of the finer particles (< 20 µm). For the mid-size classes the protection decreased, while erosion of the larger size classes (> 100 µm) was unaffected. At long times the rock fragment cover decreased the concentration of the individual size classes in proportion to effective rainfall intensity and the area exposed to raindrops. An area-based modification of the Hairsine and Rose (H-R) soil erosion model was employed to analyze the experimental data. The H-R model predictions agreed well with the measured sediment concentrations when high rainfall intensity and low rock fragment cover were used. Predictions were instead less accurate with low rainfall intensity and high rock fragment cover. The DTM results showed that the presence of rock fragments on the soil surface led to increased soil compaction, perhaps due to higher soil moisture content (from greater infiltration) within the rock fragment-covered flumes.

The dynamics of soil erosion during repeated rainfall events was studied, in particular focusing on the effect of initial soil characteristics and surface shielding by rock fragments. A sequence of four 2-h erosive events (named H7- E1, E2, and E3, respectively) separated by 22 h of air drying was performed using a 6-m long laboratory rainfall simulator and erosion flume. A loamy soil was used in all experiments. The surface was hand cultivated before the first event. Rainfall intensities of 28, 74, 74 and 28 mm h-1 were considered. The erosion flume was divided into two identical 1-m wide sections, one of which was covered with rock fragments. Results showed that steady-state behavior is mainly controlled by the rainfall intensity. Soil initial conditions, in particular whether steady state was reached during the previous event, control the sediment yields during the initial transient phase of the erosive event. If quasi-steady behavior was reached for a particular sediment size class, that class’s effluent concentration peaked rapidly in the next rainfall event, then declined gradually to its steady-state value. In contrast, if the sediment concentrations were still varying at the end of a rainfall event, the subsequent event produced effluent concentrations that increased gradually to steady state. The surface rock fragments reduced the time needed to achieve the steady state, compared to bare soil conditions. The Hairsine and Rose erosion model was adopted to analyze the measured sediment delivery. A satisfactory comparison was observed for the two experiments in which the soil was only slightly modified by raindrop impact (H7-E3 and H7-E4). On the contrary, the model could not predict accurately the erosion yields of the first two rainfall events (H7-E1 and H7-E2), during which the soil surface was heavily compacted and a surface seal developed. Furthermore, the model could not reproduce in detail the sediment concentrations of the individual size classes, when partially steady conditions were obtained during the event.

Process-based erosion modelling has proven to be an efficient tool for description and prediction of soil erosion and sediment transport. The one-dimensional Hairsine-Rose (HR) erosion model, which describes the time variation of suspended sediment concentration of multiple particle sizes, accounts for key soil erosion mechanisms: rainfall detachment, overland-flow entrainment and gravity deposition. In interrill erosion, it is known that raindrop splash is an important mechanism of sediment detachment and therefore of sediment delivery. In addition, studies have shown that the mass transported from a point source by raindrop splash decreases exponentially with radial distance and is controlled by drop characteristics and soil properties. Here we test experimentally and numerically the HR parameter consistency at different transversal widths for soil erosion in the presence of splash. To achieve this, soil erosion experiments were conducted using different configurations of the 2 m × 6 m EPFL erosion flume. The flume was divided into four identical smaller flumes, with different widths of 1 m, 0.5 m, and 2 × 0.25 m. Total sediment concentration and the concentrations for the individual size classes were measured. The experimental results indicate that raindrop splash dominated in the flumes having the larger widths (1 m and 0.5 m). This process generated a short time peak for all individual size classes. However, the effect of raindrop splash was less present in observed sediment concentrations of the collected data from the smaller width flumes (0.25 m). For these flumes, the detached sediment was controlled by the transversal width of the flume. An amount of detached sediment adhered to the barriers instead of being removed in the overland flow. Moreover, the experimental results showed that the boundary conditions affect the concentration of the mid-size and the larger particles. The one-dimensional Hairsine-Rose model was used to fit the integrated data and to provide parameter estimates for each flume. The analytical results agreed with the total sediment concentrations but not the measured sediment concentrations of all individual size classes. The observed sediment concentrations for the individual size classes could be predicted only when the initial sediment concentration was adjusted and a new calculation of the settling velocities was used. This new settling velocity calculation was conducted by taking the effect of raindrop splash on the deposition force of the particles into account.

The parameter consistency of the one-dimensional Hairsine-Rose (H-R) erosion model under conditions of significant rainfall splash was examined. To account for the splash characteristic length scale and its interaction with the transverse erosion width, experiments were carried out using erosion flumes of the same length (6 m), but different widths, with sediment concentrations measured at the flume exits. Total sediment concentration and the concentration of seven size fractions (< 2, 2-20, 20-50, 50-100, 100-315, 315-1000 and > 1000 µm) were measured at high rainfall intensity (60 mm/h) and with a gentle slope (2.2%). The conditions employed ensured that erosion was predominantly precipitation-driven. The experimental results showed that raindrop splash affected particularly the sediment breakthrough from the wider flumes (flumes 1 and 2, 1- and 0.5-m wide, respectively). However, the raindrop splash effect was less significant in observed sediment concentrations from the narrower flumes (flumes 3 and 4, both 0.25-m wide). For these flumes, the detached sediment was affected by the transversal width of the flume in that an amount of detached sediment adhered to the barriers instead of being removed in the overland flow. The one-dimensional H-R model was fitted to the experimental results and good agreement was found, in particular for the finer size classes. The data for the coarser grain sizes were more scattered, suggesting sediment motion by mechanisms other than as a suspension in the overland flow (e.g., rolling along the soil surface). The optimized parameters indicated that the shield layers (where the shield consists of redeposited eroded sediment) of the wider flumes (1 and 2) developed within 5-10 min from the start of the experiment, whereas in the narrower flumes (3 and 4) they never fully developed. The optimized detachment rates were consistent with previous findings, but the estimated thickness of the deposited layer was too small to provide complete protection of the original soil against raindrop detachment, indicating that the shield was not uniform. The experimental design allowed us to investigate directly the effect of flow non-uniformity on soil erosion by inclusion of an offset drainage point in flume 4. The observations taken during and after the experiment, as well as surface elevation data, confirmed the noticeable impact of non-uniform flow on the erosion process.