Evolution of soil surface roughness and flowpath connectivity in overland flow experiments
Introduction
Roughness is one of the major parameters controlling overland flow. The overall roughness effects depend on the scales of the processes involved. For mm to cm scales, soil roughness reduces flow velocity and the roughness effect is usually incorporated in a friction term such as Darcy–Weisbach's, Manning's or Chezy's coefficients Baird et al., 1992, Grayson and Moore, 1992, Scoging et al., 1992. At the decimetre scale, surface clods, ridges, mounds and depressions define the roughness properties. Since water storage in depressions delays overland flow triggering, quantification of storage capacity (i.e., total volume of depressions) has been a subject of numerous researches Langford and Turner, 1972, Mitchell and Jones, 1976, Gayle and Skaggs, 1978, Huang and Bradford, 1990, Mohamoud et al., 1990. When length scales exceed several decimetres, roughness effects become significant for the flow path.
Runoff generation is a spatially distributed process where surface morphology, in both the macro and micro scales, controls the surface flow routing. At the beginning of a rainfall, water infiltrates. If rainfall intensity exceeds infiltration rate, free water remains on the surface and partly fills depressions. In this stage, runoff amount is limited because the water cannot reach the outflow boundary. If rain persists with an intensity higher than the infiltration rate, puddles progressively overflow and either feed adjacent depressions or connect to the outflow boundary, thus initiating runoff. With additional rainwater, more and more depressions are connected and a network of flow paths is eventually formed. For a defined rainfall intensity–infiltration rate ratio, the flux at the outflow is directly related to the drainage area connected to the outflow boundary. This drainage area is a geometrical notion that depends on both the topography and the quantity of water stored on the surface. Describing further runoff generation is complicated by modifications in soil surface conditions and by the numerous processes among which are erosion and sedimentation in interrill areas.
Surface routing of water in interrill areas has not been extensively studied because the difficulty of acquiring data with small-scale resolution was impractical over sufficiently large surface areas. Moore and Larson (1979) showed that flow paths affect puddle-filling and runoff-triggering. While it was commonly assumed that runoff occurred only after depression storage was satisfied, Moore and Larson (1979) showed that some runoff occurred concurrently with depression filling. Later, using digital elevation models (DEMs) with 15-cm×1.3-cm resolution for an area of 0.9 m×1.5 m, Onstad (1984) showed that the rain amount needed to fill all depressions was greater than the storage capacity (i.e., the mean depression volume per area unit). He confirmed that a certain amount of water contributes to the runoff while depressions are still filling. Because overland flow begins before the complete filling of depressions, the storage capacity is not a good predictor for overland flow genesis. Using 2.6-m×1.2-m DEMs with 2.5-cm grid size, Sneddon and Chapman (1989) mapped depressions, their connections and drained areas. It appeared that the outflow of a depression did not only depend on its volume, but also on its drained surface area. As shown by Moore and Larson (1979), the increase of runoff coefficient is not continuous and can even reach transient plateaux, indicating that the growth of the area contributing to runoff is often discontinuous. Indeed, the main process is based on the outflow of individually filled depressions, each of them being characterised by an outflow threshold. The relationship between runoff and added water must thus reflect the nature of this discontinuous evolution.
Recently, Helming et al., 1998a, Helming et al., 1998b performed network geometry analysis on surfaces after simulated rainfall. Their analysis, using techniques developed to characterise river networks, was carried out on 2.8-m×0.6-m DEMs. These authors examined changes in network properties for eroding surfaces at all-connected states assuming that depressions were completely filled before analyses. They showed that flow paths undergo a decrease in sinuosity and gradient and suggested that the flow network structure evolves into a “self-organised” configuration where overall geometry of flow paths is similar to those developed river networks (fractal dimension, Horton's ratios).
This study is designed to quantify the roughness effects on overland flow genesis with a specific focus on relationships between roughness modifications and the development of water flow network in interrill areas. Changes in surface microtopography after a sequence of rainfall events were quantified utilising a laser scanner. A conditioned-walker model was used to study the process of depression filling and the triggering of runoff. This study is expected to enhance understanding of soil roughness effects on the dynamics of surface runoff generation.
Section snippets
Soil and soil box design
Experiments were carried out at the National Soil Erosion Research Laboratory, West Lafayette, IN, US. The soil was a Glynwood clay loam (fine, illitic, mesic Aquic Hapludalf with 22% sand, 49% silt and 29% clay) with a subangular-blocky structure, collected from Blackford County, IN. A wooden box of 2.4×2.4 m2 was used in the rainfall simulation study. The box was filled with a 5-cm layer of pea gravel at the bottom and a 7-cm-thick soil layer on top and separated from the gravel by a piece of
Surface morphology and storage capacity
In the experiments, overland flow was not uniformly distributed on the surface because local heterogeneities in flow depth were caused by the roughness of the surface. Nevertheless, overland flow could be considered as sheetflow. No rill appeared during any experiments and only limited incisions were observed at depression outlets.
All the variograms showed a similar two-straight line trend in a log–log diagram (Fig. 1), with a steep slope at short lag distances, and a gentle slope for lag
Discussion and conclusion
These experiments demonstrated strong differences between the smooth evolution of the variograms, and the contrasted evolution of the storage capacity and runoff characteristics as calculated with the depression-filling model. A variogram quantifies the global organisation of the topography. In our experiments, they show a characteristic correlation length in the range of 20–100 mm, depending on the initial roughness. Above this correlation length, the topography is almost uncorrelated. This is
Acknowledgements
The authors sincerely thank Rorke Bryan and an anonymous reviewer for helpful comments on the first version of this paper.
References (22)
Fluvial landsculpting and the fractal dimension of topography
Geomorphology
(1992)A self-organizing dynamic systems approach to the simulation of rill initiation and development on hillslopes
Comput. Geosci.
(1998)- et al.
Extending overland-flow models to problems of slope evolution and the representation of complex slope-surface topographies
Statistics and Data Analysis in Geology
(1973)- et al.
Laser scanner for erosion plot measurements
Trans. ASAE
(1995) - et al.
A programmable rainfall simulator for field plots
- Gayle, G.A., Skaggs, R.W., 1978 Surface storage on bedded cultivated lands. Trans. ASAE 21(1): 101–104,...
- et al.
Effect of land-surface configuration on catchment hydrology
- et al.
Surface roughness related processes of runoff and soil loss: a flume study
Soil Sci. Soc. Am. J.
(1998) - et al.
Erosional development of small scale drainage networks