Soil Erosion

The papers contained in this volume were amongst those presented to a conference on "The role of laboratory and field experiments in soil erosion research and modelling of hillslope development" held at the Scarborough Campus of the University of Toronto in April, 1989. The conference was held in association with the formal opening of a new Soil Erosion Laboratory at the campus. The conference was intended to draw together people active in soil erosion and hillslope geomorphic research to examine recent progress and to discuss priorities and approaches for future research. While many active researchers were inevitably unable to attend, the conference did attract representatives from more than a dozen countries and many different disciplines. The papers presented a good cross-section of different approaches to soil erosion and hillslope research, including field monitoring, field and laboratory experiments and theoretical modelling, and touched on almost all the topics of current importance, with examples from many different environments. Traditional, long-established methods of soil erosion research were reflected by Fullen's1 study of erosional trends on agricultural soils in central England using monitored, instrumented erosion plots under natural rainfall. Dunne1, on the other hand, carried out a detailed study of rill-interrill process on experimental field plots under simulated rainfall and applied the results in a model of long-term slope evolution in the southern Kenyan grasslands. These experiments demonstrated the critical role of vegetation in the hydrological and hydraulic response of semi-arid hillslopes. This was also reflected by large-scale simulated rainfall experiments carried out on rocky hillslopes at Walnut Gulch, Arizona, by Abrahams, Parsons & Luk. The studies by Dunne and by Abrahams et al. both focussed closely on hydraulic conditions in shallow overland flow under intense rainfall, and this topic was returned to in a number of experimental laboratory studies using simulated rainfall. Kinnell studied the interaction of rainfall intensity, flow depth and flow velocity on sediment entrainment in a small laboratory flume, while Guy & Dickinson used rather similar methodology in demonstrating the difficulty of accurate predictions of hydraulic thresholds for sediment entrainment using the Shields criterion. Torri, Biancalani & Poesen also found problems in using the Shields criterion for shallow overland flow, particularly in the situation where coarse gravels rest on steep, cohesive rill beds of much finer texture. In an examination of conditions for the incipient movement of rock fragments, they used empirical results from small flume experiments to test a model by NADEN (1987) which examines incipient gravel instability in terms of average flow velocity and turbulent velocity fluctuations. Apart from focussing on the Shields criterion, Guy & Dickinson also emphasized the importance of accurate assessment of relative roughness for prediction of sediment entrainment in shallow flows. This problem was also addressed in papers by Govers1 and Rauws which reported experimental laboratory studies from Leuven and Caen. BRYAN also examined critical hydraulic thresholds in a study of rill initiation, headcut and knickpoint development in a large laboratory flume under simulated rainfall, and provided further information on the processes of cyclic rilling and colluviation described by Bryan & Poesen (1989). De Ploey examined the problem of knickpoint and headcut development at field scale in a study of thalweg gullying on loess soils in Europe, which further developed some of the theoretical concepts first presented in a theoretical model of headcut retreat by De Ploey (1989). The papers which examined patterns and processes of erosion in a framework of shallow flow hydraulics were paralleled by a series of papers which examined time-dependant variations in soil resistance to erosion. Young, Romkens & McCool approached this problem within the framework of the Universal Soil Loss Equation, emphasizing that the K or soil erodibility factor is not, as originally conceived, a fixed soil parameter, but a highly dynamic one with marked seasonal and annual variations. A technique for appropriate modification of the K factor was presented. Several papers demonstrated the critical effect of short-term changes in surface crust and seal evolution on soil erodibility. Le Bissonnais described experiments on the structural degradation of silty French soils and demonstrated the critical influence of antecedent moisture conditions and wetting kinetics. These controls, together with rainfall energy, were also emphasized in experiments carried out in the southern United States by Romkens, Prasad & Parlange. Luk, Dubbin & Mermut described a detailed study of crust evolution under changing rainfall intensity, in which crust development was identified by thin section analy- sis and related to complex fluctuations in crust strength. Farres1 also incorporated many of these considerations in a conceptual discussion of the relationship between pedogenesis and soil erodibility. Most discussions of soil erodibility still treat the erosion process primarily in physical terms. Gerits & De Lima departed from this approach in a laboratory flume study of the erosional response of saline and sodic marine marls from southeastern Spain. In these experiments significant changes in erodibility were related to physico-chemical thresholds which reflected progressive change in the balance of runoff electrical conductivity and soil sodium absorption ratio during storm events. This paper also addressed another under-researched topic, the influence of wind characteristics on rainfall erosivity and the generation of overland flow. Several of the papers presented briefly examined the implications of the empirical results obtained for soil erosion models, but three papers were entirely devoted to discussion of new models. Kirkby presented a sediment budget model which examines the implication of the interaction of surface crusting and rill scouring, and the cyclic rill incision and colluviation identified by Bryan & Poesen (1989). The model supported the occurrence of cyclic rilling at both flume and field scale, but was unable to generate completely discontinuous channels. This was tentatively attributed to the relative simplicity of the model which ignores flow convergence effects and unsteady flow conditions. Rose, Hairsine, Proffitt & Misra presented a model rather similar to that for bed armouring in streams, in which there is continuous interchange of material between the flow and the bed. This may produce effects similar to armouring when cohesive topsoil transported by interrill processes or from rill bank collapse blankets more erodible bed material, leading to rapid fluctuations in rillflow sediment concentrations. Band described a numerical simulation experiment in which the dominant grain size of the surface soil was incorporated as a power function of gradient into a sediment transport equation. Simulations with fixed and constant initial and boundary conditions failed to produce hillslope profiles with forms uniquely characteristic of a given process. Instead profile evolution was complex with similar forms being attained at several different stages in hillslope development. While most of the papers presented at the conference focussed on either empirical observation or modelling of soil erosion processes, several papers addressed the problem of assessment of long-term erosion rates and the development of appropriate study methodologies. Bullock & Neal used a combination of air photo analysis, rainfall records and contemporary soil erosion observations to examine the expansion of saline seepage scalds in New South Wales, Australia, and demonstrate a period or "catastrophic" expansion related to exceptional rainfall conditions in the early 1950s. Sutherland & De Jong examined some of the technical difficulties involved in using caesium-137 to assess wind redistribution of soils in Saskatchewan, Canada, while studies by Pennock and by Kachanoski also used caesium-137 to identify local patterns of soil erosion on agricultural soils in southern Ontario, Canada, which demonstrated the major influence of local topography and tillage patterns. The papers presented and the accompanying discussions indicated many directions for future reaearch. It is clear that understanding of the processes of soil entrainment in sheet and rillwash under intense rainfall is still very incomplete. However, study methods are improving rapidly and swift progress in both field and laboratory experiments can be anticipated. It will be important to link the improved data available to much more detailed study of short-term temporal and spatial changes in surface soil conditions, and of selective transport and depositional processes. These must address not only changes in soil physical characteristics, but also shortterm changes in soil chemistry. Rather limited attention has been paid to the significance of soil horizons of differing charactristics in the soil erosion process, whether these originate in pedogenesis or in short-term crust evolution on homogeneous soils. Extension of experimental testing to multi-layer soils, and to a much wider range of soil types is a clear research priority. This in turn will present formidable new challenges in the development of three-dimensional models which incorporate highly dynamic soil properties. While the development of models will be complex and difficult, it provides the only hope of bridging the gap between very detailed short-term process experiments and prediction of long-term soil erosion and hillslope development patterns.