Using beryllium‐7 to monitor the relative proportions of interrill and rill erosion from loessal soil slopes in a single rainfall event

Quantifying the relative proportions of soil losses due to interrill and rill erosion processes during erosion events is an important factor in predicting total soil losses and sediment transport and deposition. Beryllium‐7 (⁷Be) can provide a convenient way to trace sediment movement over short timescales providing information that can potentially be applied to longer‐term, larger‐scale erosion processes. We used simulated rainstorms to generate soil erosion from two experimental plots (5 m × 4 m; 25° slope) containing a bare, hand‐cultivated loessal soil, and measured ⁷Be activities to identify the erosion processes contributing to eroded material movement and/or deposition in a flat area at the foot of the slope. Based on the mass balance of ⁷Be detected in the eroded soil source and in the sediments, the proportions of material from interrill and rill erosion processes were estimated in the total soil losses, the deposited sediments in the flat area, and in the suspended sediments discharged from the plots. The proportion of interrill eroded material in the discharged sediment decreased over time as that of rill eroded material increased. The amount of deposited material was greatly affected by overland flow rates. The estimated amounts of rill eroded material calculated using ⁷Be activities were in good agreement with those based on physical measurements of total plot rill volumes. Although time lags of 45 and 11 minutes existed between detection of sediment being removed by rill erosion, based on ⁷Be activities, and observed rill initiation times, our results suggest that the use of ⁷Be tracer has the potential to accurately quantify the processes of erosion from bare, loessal cultivated slopes and of deposition in flatter, downslope areas that occur in single rainfall events. Such measurements could be applied to estimate longer‐term erosion occurring over larger areas possessing similar landforms. Copyright © 2010 John Wiley & Sons, Ltd.     

Evaluation of an interrill soil erosion model using laboratory catchment data

Physically based soil erosion simulation models require input parameters of soil detachment and sediment transport owing to the action and interactions of both raindrops and overland flow. A simple interrill soil water transport model is applied to a laboratory catchment to investigate the application of raindrop detachment and transport in interrill areas explicitly. A controlled laboratory rainfall simulation study with slope length simulation by flow addition was used to assess the raindrop detachment and transport of detached soil by overland flow in interrill areas. Artificial rainfall of moderate to high intensity was used to simulate intense rain storms. However, experiments were restricted to conditions where rilling and channelling did not occur and where overland flow covered most of the surface. A simple equation with a rainfall intensity term for raindrop detachment, and a simple sediment transport equation with unit discharge and a slope term were found to be applicable to the situation where clear water is added at the upper end of a small plot to simulate increased slope length. The proposed generic relationships can be used to predict raindrop detachment and the sediment transport capacity of interrill flow and can therefore contribute to the development of physically‐based erosion models. Copyright © 1999 John Wiley & Sons, Ltd.     

Impact of rainfall pattern on interrill erosion process

The impact of rainfall pattern on the interrill erosion process is not fully understood despite its importance. Systematic rainfall simulation experiments involving various rainfall intensities, stages, intensity sequences, and surface cover conditions were conducted in this study to investigate their effects on the interrill erosion process. Five rainfall patterns designed with the same total kinetic energy/precipitation (increasing, decreasing, rising–falling, falling–rising and constant patterns) were randomly delivered to a pre‐wet clay loam soil surface at a 10° slope gradient. Significant differences in soil losses were observed among the different rainfall patterns and stages, but there was no obvious difference in runoff. Kinetic energy flux (KE_r) was a governing factor for interrill erosion, and constant rainfall pattern (CST) produced nine times greater soil loss than runs with no KE_r. Varied‐intensity patterns had a profound effect on raindrop‐induced sediment transport processes; path analysis results indicated that said effect was complex, interactive and intensity‐dependent. Low hydraulic parameter thresholds further indicated that KE_r was the dominant factor in detaching soil particles, while overland flow mainly contributed to transporting the pre‐detached particles. This study not only sheds light on the mechanism of interrill sediment transport capacity and detachability, but also may provide a useful database for developing event‐based interrill erosion prediction models. Copyright © 2017 John Wiley & Sons, Ltd.     

Landscape position,surface hydraulic gradients and erosion processes

Different hydraulic gradients, especially due to seepage or drainage, at different locations on a hillslope profile may have a profound effect on the dominant erosion processes. A laboratory study was designed to simulate hillslope processes and quantify effects of surface hydraulic gradients on erosion for a Glynwood clay loam soil (fine, illitic, mesic Aquic Hapludalf). A 5 m long, 1·2 m wide soil pan was used at 5 and 10 per cent slopes with an external watering tube to vary the soil bed's hydrological conditions. Different combinations of slope steepness with seepage or drainage gradients were used to simulate the hydrologic conditions on a 5 m segment of a hillslope profile. Runoff samples were taken during rainfall-only and rainfall with added inflow. Results showed that, under drainage conditions, interrill processes dominated and rilling was limited. The surface contained scattered crescent-shaped pits after the run. Under seepage conditions, rilling processes dominated and the inflow introduced at the top of the soil pan further accelerated the headward erosion of the rills. Erosion rates increased by as much as 60 times under seepage conditions representative of the lower backslope when compared to drainage conditions that generally occur at the upper backslope. This indicated that rills and gullies on backslopes and footslopes may be catalysed or enhanced by seepage conditions rather than form from flow hydraulic shear stress alone. An understanding of spatial and temporal changes that affect both hillslope hydrology and erosional processes is needed to develop accurate process-based erosion prediction models. This knowledge may lead to different management practices on landscape positions where seepage occurs. © 1998 John Wiley & Sons, Ltd.     

Effects of rainfall patterns on runoff and rainfall-induced erosion

Rainfall-induced erosion involves the detachment of soil particles by raindrop impact and their transport by the combined action of the shallow surface runoff and raindrop impact.Although temporal variation in rainfall intensity(pattern)during natural rainstorms is a common phenomenon,the available information is inadequate to understand its effects on runoff and rainfall-induced erosion processes.To address this issue,four simulated rainfall patterns(constant,increasing,decreasing,and increasing-decreasing)with the same total kinetic energy were designed.Two soil types(sandy and sandy loam)were subjected to simulated rainfall using 15 cm×30 cm long detachment trays under infiltration conditions.For each simulation,runoff and sediment concentration were sampled at regular intervals.No obvious difference was observed in runoff across the two soil types,but there were significant differences in soil losses among the different rainfall patterns and stages.For varying-intensity rainfall patterns,the dominant sediment transport mechanism was not only influenced by raindrop detachment but also was affected by raindrop-induced shallow flow transport.Moreover,the efficiency of equations that predict the interrill erosion rate increased when the integrated raindrop impact and surface runoff rate were applied.Although the processes of interrill erosion are complex,the findings in this study may provide useful insight for developing models that predict the effects of rainfall pattern on runoff and erosion.     

Erodibility of soil and organic matter: independence of organic matter resistance to interrill erosion

The enrichment of organic matter in interrill sediment is well documented; however, the respective roles of soil organic matter (SOM) and interrill erosion processes for the enrichment are unclear. In this study, organic matter content of sediment generated on two silts with almost identical textures, but different organic matter contents and aggregations, was tested. Artificial rainfall was applied to the soils in wet, dry and crusted initial conditions to determine the effects of soil moisture and rainfall and drying history on organic matter enrichment in interrill sediment. While erosional response of the soils varied significantly, organic matter enrichment of sediment was not sensitive to initial soil conditions. However, enrichment was higher on the silt with a lower organic matter content and lower interrill erodibility. The results show that enrichment of organic matter in interrill sediment is not directly related to either SOM content or soil interrill erodibility, but is dominated by interrill erosion processes. As a consequence of the complex interaction between soil, organic matter and interrill erosion processes, erodibility of organic matter should be treated as a separate variable in erosion models. Further research on aggregate breakdown, in particular the content and fate of the organic matter in the soil fragments, is required. Copyright © 2007 John Wiley & Sons, Ltd.     

MODELING OF INTERRILL EROSION IN THE LOESS PLATEAU OF CHINA

IINTRODUCTIONTheinterrillerosiononafieldplotisaffectedbythekineticenergyoftherainfall,wind,topographyfactors,propertiesofsoilandthecanopy.Theinterrillerosionoccursasthefirstdropimpactsthehillslopes.Theinterrillerosionoccursinallkindsofrainfallandtheamountofthesplasherosion,whichisthemainpartofinterrillerosion,canaccountforagreatpanofthetotalerosionamountinaheavystorm(Baner1990,Glymph1957,QianandWan1986,Zhou1981).Therefore,itisveryimportanttorevealthemechanismtoestimatetheamountofinterri…     

Effects of rainfall,overland flow and their interactions on peatland interrill erosion processes

Interrill erosion processes on gentle slopes are affected by mechanisms of raindrop impact, overland flow and their interaction. However, limited experimental work has been conducted to understand how important each of the mechanisms are and how they interact, in particular for peat soil. Laboratory simulation experiments were conducted on peat blocks under two slopes (2.5° and 7.5°) and three treatments: Rainfall, where rainfall with an intensity of 12 mm h^?1 was simulated; Inflow, where upslope overland flow at a rate of 12 mm h^?1 was applied; and Rainfall + Inflow which combined both Rainfall and Inflow. Overland flow, sediment loss and overland flow velocity data were collected and splash cups were used to measure the mass of sediment detached by raindrops. Raindrop impact was found to reduce overland flow by 10 to 13%, due to increased infiltration, and reduce erosion by 47% on average for both slope gradients. Raindrop impact also reduced flow velocity (80–92%) and increased roughness (72–78%). The interaction between rainfall and flow was found to significantly reduce sediment concentrations (73–85%). Slope gradient had only a minor effect on overland flow and sediment yield. Significantly higher flow velocities and sediment yields were observed under the Rainfall + Inflow treatment compared to the Rainfall treatment. On average, upslope inflow was found to increase erosion by 36%. These results indicate that overland flow and erosion processes on peat hillslopes are affected by upslope inflow. There was no significant relationship between interrill erosion and overland flow, whereas stream power had a strong relationship with erosion. These findings help improve our understanding of the importance of interrill erosion processes on peat. Copyright © 2017 John Wiley & Sons, Ltd.     

Runoff generation and routing on artificial slopes in a Mediterranean–continental environment: the Teruel coalfield,Spain

The aim of this study was to identify the mechanisms of runoff generation and routing and their controlling factors at the hillslope scale, on artificial slopes derived from surface coal mining reclamation in a Mediterranean–continental area. Rainfall and runoff at interrill and microcatchment scales were recorded for a year on two slopes with different substrata: topsoil cover and overburden cover. Runoff coefficient and runoff routing from interrill areas to microcatchment outlets were higher in the overburden substratum than in topsoil, and greater in the most developed rill network. Rainfall volume is the major parameter responsible for runoff response on overburden, suggesting that this substratum is very impermeable—at least during the main rainfall periods of the year (late spring and autumn) when the soil surface is sealed. In such conditions, most rainfall input is converted into runoff, regardless of its intensity. Results from artificial rainfall experiments, conducted 3 and 7 years after seeding, confirm the low infiltration capacity of overburden when sealed. The hydrological response shows great seasonal variability on the overburden slope in accordance with soil surface changes over the year. Rainfall volume and intensities (I₃₀, I₆₀) explain runoff at the interrill scale on the topsoil slope, where rainfall experiments demonstrated a typical Hortonian infiltration curve. However, no correlation was found at the microcatchment level, probably because of the loss of functionality of the only rill as ecological succession proceeded. The runoff generation mechanism on the topsoil slope is more homogeneous throughout the year. Runoff connectivity, defined as the ratio between runoff rates recorded at the rill network scale and those recorded at the interrill area scale in every rainfall event, was also greater on the rilled overburden slope, and in the most developed rill network. The dense rill networks of the overburden slope guarantee very effective runoff drainage, regardless of rainfall magnitude. Rills drain overland flow from interrill‐sealed areas, reducing the opportunity of reinfiltration in areas not affected by siltation. Runoff generation and routing on topsoil slopes are controlled by grass cover and soil moisture content, whereas on overburden slopes rill network density and soil moisture content are the main controlling factors. Copyright © 2002 John Wiley & Sons, Ltd.     

Field measurements of interrill erosion under ­different slopes and plot sizes

Despite numerous studies, the effect of slope on interrill erosion is not clearly established. Several interactions exist between erosion parameters that are not taken into account under experimental laboratory measurements and results need to be validated in the field. The influence of slope steepness (2 to 8 per cent) on soil loss for a crusted interrill area and the detachment and transport processes involved in the interaction between slope, rain characteristics and plot size were investigated. Sediment discharge and runoff rates were measured in bounded plots (1 m2 and 10 m2) under natural and simulated rainfall, allowing the analysis of a combination of detachment and transport processes at various scales in the field. Runoff rate increased from 20 to 90 per cent with increasing slope and rain intensity for both plot sizes, whereas sediment concentration increased from 2 to 6 g l−1 with increasing slope only for the 10 m2 plots. At the 1 m2 scale, erosion was transport‐limited due to the reduced rain‐impacted flow. Interactions between slope angle and rain intensity were observed for detachment and transport processes in interrill erosion. Results show the importance of an adapted experimental set‐up to get reference data for interrill erosion model development and validation. Copyright © 2000 John Wiley & Sons, Ltd.     

Rainfall runoff and erosion in Napa Valley vineyards: effects of slope,cover and surface roughness

The effects of slope, cover and surface roughness on rainfall runoff, infiltration and erosion were determined at two sites on a hillside vineyard in Napa County, California, using a portable rainfall simulator. Rainfall simulation experiments were carried out at two sites, with five replications of three slope treatments (5%, 10% and 15%) in a randomized block design at each site (0%bsol;64 m² plots). Prior to initiation of the rainfall simulations, detailed assessments, not considered in previous vineyard studies, of soil slope, cover and surface roughness were conducted. Significant correlations (at the 95% confidence level) between the physical characteristics of slope, cover and surface roughness, with total infiltration, runoff, sediment discharge and average sediment concentration were obtained. The extent of soil cracking, a physical characteristic not directly measured, also affected analysis of the rainfall–runoff–erosion process. Average cumulative runoff and cumulative sediment discharge from site A was 87% and 242% greater, respectively, than at site B. This difference was linked to the greater cover, extent of soil cracking and bulk density at site B than at site A. The extent of soil cover was the dominant factor limiting soil loss when soil cracking was not present. Field slopes within the range of 4–16%, although a statistically significant factor affecting soil losses, had only a minor impact on the amount of soil loss. The Horton infiltration equation fit field data better than the modified Philip's equation. Owing to the variability in the ‘treatment’ parameters affecting the rainfall–runoff–erosion process, use of ANOVA methods were found to be inappropriate; multiple‐factor regression analysis was more useful for identifying significant parameters. Overall, we obtained similar values for soil erosion parameters as those obtained from vineyard erosion studies in Europe. In addition, it appears that results from the small plot studies may be adequately scaled up one to two orders of magnitude in terms of land areas considered. Copyright © 2000 John Wiley & Sons, Ltd.     

An un-mixing model to study watershed erosion processes

An un-mixing model is formulated within a Bayesian Markov Chain Monte Carlo framework for use within land-use fingerprinting to study watershed erosion processes. The model has two new components: (1) An equation and erosion process parameter are used to weight tracer signatures from each erosion process within a land-use. (2) An extra tracer distribution and episodic erosion parameter are used to represent soil eroded throughout the sampling duration and thus include the episodic nature of erosion. To test specification of these new parameters, the un-mixing model is applied in the 15 km² Jerome Creek Watershed in the Palouse Region of Northwestern Idaho. Erosion processes include surface erosion upon mountain slopes due to logging in the forest land-use and rill/interrill erosion on cultivated slopes and headcut erosion in riparian floodplains of the agricultural land-use (winter wheat/peas rotation and hay pasture). Episodic erosion occurs for the event where the model is applied. A sensitivity analysis shows that the smallest Bayesian credible set results when the new parameters are specified using hydrologic data and process-based models. The un-mixing model predicts that 90% of the eroded-soil originated from the agricultural land-use and 10% originated from the forest land-use. A comparative study is performed that estimates 90.5% and 9.5% of eroded-soil originated from the agricultural and forest land-uses. Successful performance of the un-mixing model highlights future application as a standalone probabilistic tool to monitor watershed erosion processes that exhibit non-equilibrium conditions and provide calibration data for process-based watershed models.     

Loess erosion change modeling during heavy rainfall

A typical gully sub-basin with a complex geomorphological form is used to do a model test of gravity erosion of loess by considering the sequence of slopes in a prototype gully creating a sequence of underlying surface forms in the upper reaches. The results show that the runoff from heavy rainfall is the main external force for the erosion of loess, and also is an important influencing factor to stimulate and intensify the development of gravity erosion. The soil structure and the height of the...     

Effects of rainfall and slope on runoff,soil erosion and rill development: an experimental study using two loess soils

Runoff generation and soil loss from slopes have been studied for decades, but the relationships among runoff, soil loss and rill development are still not well understood. In this paper, rainfall simulation experiments were conducted in two neighbouring plots (scale: 1 m by 5 m) with four varying slopes (17.6%, 26.8%, 36.4% and 46.6%) and two rainfall intensities (90 and 120 mm h^?1) using two loess soils. Data on rill development were extracted from the digital elevation models by means of photogrammetry. The effects of rainfall intensity and slope gradient on runoff, soil loss and rill development were different for the two soils. The runoff and soil loss from the Anthrosol surface were generally higher than those from the Calcaric Cambisol surface. Higher rainfall intensity produced less runoff and more sediment for almost each treatment. With increasing slope gradient, the values of cumulative runoff and soil loss peaked, except for the treatments with 90 mm h^?1 rainfall on the slopes with Anthrosol. With rainfall duration, runoff discharge decreased for Anthrosol and increased for Calcaric Cambisol for almost all the treatments. For both soils, sediment concentration was very high at the onset of rainfall and decreased quickly. Almost all the sediment concentrations increased on the 17.6% and 26.8% slopes and peaked on the 36.4% and 46.6% slopes. Sediment concentrations were higher on the Anthrosol slopes than on the Calcaric Cambisol slopes. At 90 mm h^?1 rainfall intensity, increasingly denser rills appeared on the Anthrosol slope as the slope gradient increased, while only steep slopes (36.4% and 46.6%) developed rills for the Calcaric Cambisol soil. The contributions of rill erosion ranged from 36% to 62% of the cumulative soil losses for Anthrosol, while the maximum contribution of rill erosion to the cumulative soil loss was only 37.9% for Calcaric Cambisol. Copyright © 2014 John Wiley & Sons, Ltd.     

A simulation model for unified interrill erosion and rill erosion on hillslopes

A mathematical model was developed for simulating runoff generation and soil erosion on hillslopes. The model is comprised of three modules: one for overland flow, one for soil infiltration, and one for soil erosion including rill erosion and interrill erosion. Rainfall and slope characteristics affecting soil erosion on hillslopes were analysed. The model results show that the slope length and gradient, time distribution rainfall, and distribution of rills have varying influence on soil erosion. Erosion rate increases nonlinearly with increase in the slope length; a long slope length leads to more serious erosion. The effect of the slope gradient on soil erosion can be both positive and negative. Thus, there exists a critical slope gradient for soil erosion, which is about 45° for the rate of erosion at the end of the slope and about 25° for the accumulated erosion. Copyright © 2005 John Wiley & Sons, Ltd.     

Runoff and sediment loss responses to rainfall and land use in two agricultural catchments on the Loess Plateau of China

Soil erosion is a severe problem hindering sustainable agriculture on the Loess Plateau of China. Plot experiments were conducted under the natural rainfall condition during 1995–1997 at Wangdongguo and Aobao catchments in this region to evaluate the effects of various land use, cropping systems, land slopes and rainfall on runoff and sediment losses, as well as the differences in catchment responses. The experiments included various surface conditions ranging from bare soil to vegetated surfaces (maize, wheat residue, Robinia pseudoacacia L., Amorpha fruticosa L., Stipa capillata L., buckwheat and Astragarus adsurgens L.). The measurements were carried out on hill slopes with different gradients (i.e. 0 ° to 36 °). These plots varied from 20 to 60 m in length. Results indicated that runoff and erosion in this region occurred mainly during summer storms. Summer runoff and sediment losses under cropping and other vegetation were significantly less than those from ploughed bare soil (i.e. without crop/plant or crop residue). There were fewer runoff and sediment losses with increasing canopy cover. Land slope had a major effect on runoff and sediment losses and this effect was markedly larger in the tillage plots than that in the natural grass and forest plots, although this effect was very small when the maximum rainfall intensity was larger than 58·8 mm/h or smaller than 2·4 mm/h. Sediment losses per unit area rose with increasing slope length for the same land slope and same land use. The effect of slope length on sediment losses was stronger on a bare soil plot than on a crop/plant plot. The runoff volume and sediment losses were both closely related to rainfall volume and maximum intensity, while runoff coefficient was mainly controlled by maximum rainfall intensity. Hortonian overland flow is the dominant runoff process in the region. The differences in runoff volume, runoff coefficient and sediment losses between the catchments are mainly controlled by the maximum rainfall intensity and infiltration characteristics. The Aobao catchment yielded much larger runoff volume, runoff coefficient and sediment than the Wangdongguo catchment. Copyright © 2001 John Wiley & Sons, Ltd.     

Modeling aggregate size distribution of eroded sediment resulting from rain-splash and raindrop impacted flow processes

Soil susceptibility to detachment and transport sub-processes of erosion is generally controled by the aggregate breakdown mechanism. Measuring particle size and aggregation to the estimate erodibility potential of soils is important under erosive rainfall conditions. The Aggregate Size Distribution (ASD) is one of the most important determinants of soil structure along with soil organic matter content for describing the efficiency of applied, sustainable management strategies. This study aimed to compare the performances of three different aggregate size distribution models to predict the characteristic aggregate size parameter (median diameter, D50) for eroded sediment from interrill erosion processes of Rain- Splash Transport (RST) and Raindrop Impacted Flow Transport (RIFT). The ASDs of 1143 collected sediment samples from the RST and RIFT processes were measured and modeled by the Log-normal, Fractal, and Weibull approaches. The D50 value, as a characteristic parameter for aggregate size distributions, derived from the cumulative ASD curve was compared for soils from different land use types and different slope and rainfall intensity conditions. The performance of each model was evaluated using the Mean Square Error (MSE) and Coefficient of Determination (R^2). The Weibull approach was the most accurate model showing the best fit with the lowest MSE values (0.0002 ≤MSE≤ 0.0048) and having the greatest R2 values (0.936≤ R^2≤ 0.998) when compared with the Log-normal and Fractal models. Herewith, for semi-arid land use and soil, specific shape and scale parameters for the Weibull distribution, the respective ASDs were successfully re-generated for modeling the eroded sediment of the simulated RST and RIFT interill processes.     

Compaction and cover effects on runoff and erosion in post-fire salvage logged areas in the Valley Fire,California

Runoff and erosion processes can increase after wildfire and post-fire salvage logging, but little is known about the specific effects of soil compaction and surface cover after post-fire salvage logging activities on these processes. We carried out rainfall simulations after a high-severity wildfire and post-fire salvage logging to assess the effect of compaction (uncompacted or compacted by skid traffic during post-fire salvage logging) and surface cover (bare or covered with logging slash). Runoff after 71 mm of rainfall across two 30-min simulations was similar for the bare plots regardless of the compaction status (mean 33 mm). In comparison, runoff in the slash-covered plots averaged only 22 mm. Rainsplash in the downslope direction averaged 30 g for the bare plots across compaction levels and decreased significantly by 70% on the slash-covered plots. Sediment yield totalled 460 and 818 g m⁻² for the uncompacted and compacted bare plots, respectively, and slash significantly reduced these amounts by an average rate of 71%. Our results showed that soil erosion was still high two years after the high severity burning and the effect of soil compaction nearly doubled soil erosion via nonsignificant increases in runoff and sediment concentration. Antecedent soil moisture (dry or wet) was the dominant factor controlling runoff, while surface cover was the dominant factor for rainsplash and sediment yield. Saturated hydraulic conductivity and interrill erodibility calculated from these rainfall simulations confirmed previous laboratory research and will support hydrologic and erosion modelling efforts related to wildfire and post-fire salvage logging. Covering the soil with slash mitigated runoff and significantly reduced soil erosion, demonstrating the potential of this practise to reduce sediment yield and soil degradation from burned and logged areas.     

The influence of subsurface moisture on rill system evolution

Rill development studies have focused almost exclusively on surface erosion processes and critical threshold hydraulic conditions. Characteristic rill features, such as arcuate headcuts and knickpoints, are morphologically similar to the ‘theatre-headed’ valleys which have been associated with ‘sapping’ processes at various scales. This paper reports on laboratory experiments designed to identify linkages between surface flow hydraulics, subsurface moisture conditions and rill development. Experiments were carried out in a 16·57 m² flume under simulated rainfall with soil samples up to 0·15 m depth in which moisture conditions were monitored by miniature time-domain reflectometer probes. Tests showed complex responses in which some rill incision reflected surface flow conditions, but major rill system development with markedly enhanced sediment yield was closely associated with high soil moisture contents. It was not possible to measure seepage forces directly, but calculation and observation indicate that these were less important than reduction in soil strength with saturation, which resulted in increased effective runoff erosivity. This caused concentrated undercutting along the water table at rill walls, while slightly stronger surface layers above the water table formed microscarps. These retreated along the water table into interrill surfaces, producing residual pediment transport slopes. The microscarps eventually disappeared when the water table reached the surface, eliminating differential soil strength. The experiments showed complex relationships between surface and subsurface erosional processes in evolving rill systems, strongly influenced by soil moisture dynamics. The very small topographic and hydraulic head amplitudes indicate that seepage forces and ‘sapping’ were minimal. The dominant effect of soil moisture was reduction of soil strength with saturation, and increased runoff entrainment. Experimental conditions were not unusual, either for agricultural fields or natural hillslopes, and the intricate interrelationship of surface and subsurface erosion processes observed is probably not uncommon. Attempts to link specific morphologic features at rill scale to dominance of surface or subsurface processes alone are therefore unlikely to be successful or reliable. © 1998 John Wiley & Sons, Ltd.