The classical model of surface runoff generation is by an infiltration excess mechanism in which rainfall intensity exceeds the local infiltration capacity of the soil for a sufficient period of time for any depression storage at the soil surface to be satisfied such that downslope flow is initiated. This will not occur where the permeability of the soil is high in comparison with expected rainfall intensities, and even when this is not the case there may be an initial period when all the rainfall infiltrates before bringing the surface to saturation (the time to ponding). There have been many infiltration equations proposed, either empirical or based on various approximate solutions to Darcy's law, with the aim of predicting times to ponding, infiltration capacities, and the production of surface runoff. All require the specification of some parameter values that measurements suggest may vary dramatically even within a single soil type. Spatial variability of soil characteristics may therefore be important in the production of surface runoff, since infiltration excess runoff will start first in areas of low infiltration capacity. Where runoff flows downslope onto areas of higher infiltration capacity as run-on, there may be further infiltration of part or all of the water before it reaches a stream channel.
Infiltration capacities of the soil can also be greatly enhanced by the presence of macropores [cracks, root channels, animal burrows see Beven and Germann (1982)] or reduced by the presence of surface crusting resulting from the redistribution of fine particles by raindrop impact (Rômkens et al, 1990) such that the properties of the bulk soil matrix alone may not be a good indication of actual infiltration rates.
Surface runoff may be found in areas of high infiltration capacity soils if the rainfall falls on soil that is saturated to the surface or if return flow from the subsurface occurs onto a saturated surface. This may not, in fact, require saturation of the whole soil profile. The buildup of a perched water table, for example, due to a permeability break between horizons in the soil profile, might result in saturation to the surface and the consequent generation of surface runoff. Surface runoff produced in this way has been studied, for example, by Bonell et al. (1981).
Saturation is most likely to occur where upslope contributing areas are large, such as in valley bottoms and hillslope hollows, and where effective hydraulic gradients are small. Convergence of flow lines will tend to increase the likelihood of saturation; steep slopes will tend to decrease it. The topographic index, a/s, discussed above, reflects these counteracting tendencies. High values of the index (high contributing areas, low slope) will indicate, other things being equal, a higher likelihood of saturated conditions occurring; low values of the index indicate the reverse.
It is not only runoff source areas on the hillslopes that exhibit dynamic extension during storm periods. Many studies have emphasized the role of extension of the ephemeral channel network in the generation of storm runoff (e.g., Hewlett, 1974). The area of the channel itself and the immediately adjacent riparian area can, in some catchments in which the storm hydrograph represents a volume equal to only a small proportion of the incident rainfall (a small runoff coefficient), be the most important source area of storm runoff.
4 EFFECT OF HETEROGENEITY Heterogeneity of Hillslope Forms
Topography is an important control on runoff generation in catchments where down-slope flows are an important control on the runoff response. The topographic index provides one very simple approach to identify likely runoff production areas. This will be particularly true for saturation excess runoff production or subsurface storm-flows where topographic convergence is important. It may also be true for infiltration excess runoff production where a catenary relationship between soil textural properties and topographic position has developed due to translocation of clays and other long-term processes. The topographic index will be limited as a predictor wherever flow lines in the subsurface depart radically from the surface topography, as in fractured systems or deep groundwater systems, or where the soil is relatively dry such that the effective upslope contributing areas for subsurface flow are very small (e.g., Barling et al., 1994).
It has also been noted that variability of soil characteristics may have an important control on runoff production for all the suggested mechanisms. This will be true for the variability of soil series at the catchment scale and also for the variability in characteristics found within a soil series that may not show clear spatial patterns. Variability in soil characteristics associated with the occurrence of vegetation may also be important. Dunne et al. (1991), for example, suggest that the infiltration capacities of the soil beneath the plants of a sparse vegetation canopy can be much higher than between the plants, to the extent that depths of overland flow generated during an event may be greatly affected by local infiltration beneath the plants. Similarly in the Tiger Bush area of the HAPEX-SAHEL experiment in Niger, it is thought that surface runoff from bare soil areas may serve to increase the water available to the adjacent stripes of Tiger Bush (Peugeot et al., 1997).
In all these cases, it is clear that the knowledge of mean soil hydrological characteristics may not be sufficient to understand surface runoff production. Rather the distribution of characteristics within a catchment area may be important. Even with detailed information about soil properties, there is some doubt that the nature of runoff production is truly predictable in detail, especially for infiltration excess type mechanisms. The plot studies of Hjelmfelt and Burwell (1984), for example, suggested that measured surface runoff from adjacent plots may be unpredictable while the study of Loague (1990) and Loague and Kyriakidis (1997) on the R-5 catchment at Chickasha, Ohio, revealed the difficulty of modeling spatially heterogeneous infiltration excess runoff production even on a small catchment with detailed soil measurements.
The effects of vegetation on the effective rainfall intensities at ground level and on infiltration rates have already been noted. There are also other effects associated with the heterogeneity of vegetation. If surface runoff is generated, then the effective roughness of the surface may be controlled primarily by the surface vegetation. On surfaces covered by grasses, for example, this may lead to very low velocities for surface runoff (velocities as low as 20m/h have been measured in the field). The vegetation may also protect the surface from erosion by both rainsplash and flowing water.
Whatever the nature of the surface or subsurface runoff generation processes, the amount of runoff generation will predominantly be determined by the forcing of the rainstorm inputs and the state of antecedent wetness of the catchment. Incident precipitation intensities can vary greatly in space, particularly during convective rainstorm events when patterns of intensities depend strongly on the growth and decay of storm cells and the overall movement of the storm (e.g., Smith et al., 1996).
Runoff generation may depend strongly on patterns of rainfall intensity as suggested, for example, by simulations of the Walnut Gulch experimental catchment in Arizona by Goodrich et al. (1994).
Was this article helpful?