subsurface flow and erosion Subsurface erosion of the land can produce some spectacular features, but it has often been regarded as unusual. Until quite recently, theorists and modellers of both landscape evolution and storm-water run-off have concentrated on surface processes, following in particular the lead given by the American hydrologist Robert E. Horton in the 1940s. Hortonian theory maintained that erosion and storm run-off in humid regions are caused solely by overland flow, which occurs after rainfall rates exceed the infiltration capacity of the soil surface. According to Horton's premise, stream channels begin where the erosional power of this overland flow exceeds the erosional resistance of the soil surface.
Evidence has grown over the intervening decades to support an alternative view: that subsurface flow plays an important and sometimes even dominant role in perhaps the majority of environments. One of the most widespread and important roles that subsurface flow can play is in the formation of saturation overland flow. It is now realized that Horton's idea that overland flow is generated all over the river basin is rarely supported by observations, except perhaps after heavy storms in the desert. Surface flow normally begins in just a few areas, especially dips or hollows near the stream. If the storm lasts long enough, these areas may extend upslope away from the stream, but they rarely cover the whole basin.
Subsurface flow tends to create this localized saturation (or saturated) overland flow in two ways:(1) soils in the hollows and bank-side areas are wetter at the beginning of the storm because water has seeped downslope into these places beforehand; and(2) subsurface stormflow carries water that has infiltrated the soil upslope rapidly downhill and into these places.The longer the storm lasts, the more time there is for water to seep downslope from further and further upslope to the edge of the saturated areas, and so the saturated area expands. This saturation overland flow therefore commonly contains water from three sources:(1) ‘old’ water left over from previous storms that may be displaced by the ‘new’ water that has entered the soil during the current storm;(2) ‘new’ water that has infiltrated during the current rainstorm; and(3) rain that has fallen on the saturated areas and has not been able to infiltrate the soil.The first two are known collectively as
return flow. The proportions of each vary from place to place and storm to storm.
Subsurface flow can also deliver water directly into the stream channel either as diffuse seepage or as pipeflow emerging from the stream bank. Diffuse seepage can occur on a wide front, especially near to the channel where the water-table joins the stream. This flow may be either beneath the water-table, as ‘saturated flow’, or above it, as ‘unsaturated flow’. Flow is much faster in saturated soil, but it still tends to be rather slow, typically moving at rates of barely 1 mm per second. Flow may be more efficient in
percolines, which are localized lines of extra deep soil which concentrate drainage in some soils. Saturated flow persists for longer there, and the concentrated flow may winnow away the finer soil particles, increasing the permeability. Persistent saturation increases weathering and increased erosion may also lower the surface, capturing more surface water. Percolines can grade downslope into
seepage lines, where return flow commonly emerges or
exfiltrates. Theories of hillslope evolution still generally need to incorporate the spatial diversity in slope development that this linearly localized weathering and erosion create.
Flows are very much faster in soils that have connecting voids called
macropores. Macropores may be cracks, caused by desiccation or mass slumping, cylindrical holes (mainly biotic), vughs (between these two types), or water-sculpted soil pipes (see
piping). The larger macropores may extend for long distances, as water erodes and enlarges the segments that run downslopes. Networks of soil pipes up to 750 m long have been reported. These act as ‘soil springs’ and have been found to carry half of all the storm flow reaching the stream in a British upland basin. Rainwater can reach these pipes within an hour or so, even though they may be 50 centimetres below the surface. It seems that smaller macropores and ‘blowholes’ at the surface allow rainwater to infiltrate the soil very rapidly, causing the water-table to rise and then drain through the pipes.
While flow through the fine micropores of the soil matrix is essentially non-turbulent or laminar flow, flow through macropores can be turbulent and erosive. Typical flow rates in macropores range from 5 mm per second up to 0.5 m per second in pipe flow. The exact threshold at which erosion begins is highly variable and difficult to determine, because it depends on the nature of both the flow and the soil; individual, loose (or ‘dispersed’) clay particles may be transported through pores a mere 10 micrometres (μm) in diameter, whereas aggregated soil particles may be eroded only in the occasional, more extreme flows, even in pipes of 100 mm diameter.
All forms of subsurface flow within the soil or regolith are now commonly referred to as
through flow. The older term,
interflow, is still occasionally used synonymously, although it originated among engineering hydrologists to refer to water that arrives towards the end of stormflow in the stream after following a deep subsurface route, typically thought of as through bedrock. In some cases, subsurface flow through the soil is distinct from subsurface flow through bedrock, especially where the surface bedrock is impermeable. At other times, the two may be connected, especially near stream channels. The term
groundwater flow is generally restricted to deep seepage in the saturated zone of the bedrock and its subsequent contribution to stream flow, primarily because this is of greater economic importance as a source of public water supply. Nevertheless, in reality soil water and this groundwater
sensu stricto often interact and mix. In broad terms, all subsurface water is groundwater.
Deep groundwater flow exploits joints and other planes of weakness in rocks, just like through flow, and can develop spring networks. This reaches its most spectacular expression in the underground rivers of a karst landscape. Whereas karst is predominantly formed by solutional erosion, piping processes are essentially mechanical. Karst-like caves, sinks, and collapse features formed in non-calcareous bedrock by piping processes have been called
pseudokarst.
Spring sapping occurs where piping processes extend the head of a channel or gully or develop tributary channels. This may occur as a general washing out of the subsoil or by enlargement of a pipe until the roof collapses. Gully networks with stubby tributaries or blind valleys meeting at high angles are typical of surface channels that have been formed by subsurface erosion. Many channels are, however, indistinguishable from those formed by overland flow, whether or not surface flow has played any part in their development. The extent to which subsurface erosion is responsible for developing surface channels is still unknown, but it seems reasonable to suggest that subsurface erosion is far more important than has been supposed in humid areas with a good vegetation cover that restricts surface erosion.
As with channel initiation, subsurface erosion is often found to be inextricably linked with surface processes in many instances of severe soil erosion. Rills and gullies may stimulate subsurface erosion by creating a hydraulic gradient within the soil towards the new channel. At the same time, piping may be creating the rills.
J. A. A. Jones
