Synopsis top ↑
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.