Geoscience Reference
In-Depth Information
2001). Other developments of soil horizonisation, either at
the surface (Abrahams and Parsons, 1991a; Young
et al.
,
2004) or in the subsurface (Hamerlynck, McAuliffe and
Smith, 2000; Hamerlynck
et al.
, 2002), have been noted
as having significant effects on reducing infiltration rates.
As throughflow and deep percolation are rare, mech-
anisms of spatial variability in water flow are largely
surface-driven, as a result of runoff-runon processes
(Wainwright and Parsons, 2002; Rockstrom, Jansson and
Barron, 1998; and see the further discussion below) or as a
result of the pattern of vegetation, or the interaction of the
two. Vegetation draws moisture out of the soil by transpi-
ration, and there are increasing numbers of observations
of upwards vertical movement of moisture by a process
known as hydraulic lift (Richards and Caldwell, 1987;
Yoder and Nowak, 1999). This process occurs when there
are relatively saturated conditions at depth, allowing plant
roots to take up water and move it upwards internally until
transpiration stops (e.g. at night-time), at which point the
water may flow back out of root pores into surrounding
drier soils because of the hydraulic gradient.
Soil-moisture variability in space and time is important
in controlling infiltration and runoff generation by the
control it exerts on the hydraulic gradient at the start of a
rainfall event. Turnbull, Wainwright and Brazier (2010)
demonstrated that both grass and shrub vegetation in
the Chihuahuan Desert caused the more rapid drying of
the soil than on bare surfaces (Figure 11.8) - suggesting
the greater importance of transpiration compared to evap-
oration - and the evolution of related spatial structure in
the soil moisture (with the further feedbacks noted above).
As storm events may occur more rapidly than the decline
to 'background' dry conditions as seen in this example
and in Ceballos
et al
. (2002), the differential effect and
the development of connectivity of flows (see below) as
affected by vegetation in this way may be critical in under-
standing variability in the behaviour of the system during
the rainfall season.
Subsurface animal activity may also be an important
control on infiltration. Areas with termites in unvege-
tated areas in New Mexico were found to have signifi-
cantly higher infiltration rates (88
wetting to occur will produce higher infiltration rates. The
presence of burrows produces significant macropores in
the soil, and these burrows may often be co-located with
vegetation as animals attempt to avoid predation, which
is another reason why infiltration rates under plants may
be higher than elsewhere.
Macropores may also form where the soil surface be-
comes cracked. In drylands, this phenomenon most com-
monly occurs as a result of desiccation, providing a rapid
link from the surface to subsurface flows. It is commonly
found in badlands, especially in those dominated by very
fine clays, particularly montmorillonite (Bryan, Imeson
and Campbell, 1984), and where they can develop into
large-scale pipes (Torri and Bryan, 1997; Farifteh and
Soeters, 1999). Govers (1991) has suggested that the size
and spacing of desiccation cracks depends on landscape
location, with those on slopes being larger and closer
together than those on plateaus and aspect-related differ-
ences in insolation increasing the extent of cracking.
A further subsurface effect is that of air entrapment.
Baird (1997) suggested that the entrapment of air in the
profile can cause pores to be blocked to the passage of
infiltrating water, citing experimental evidence from Con-
stantz, Herkelrath and Murphy (1988) that this effect could
lead to a reduction of 80-90 % of the saturated infiltration
rate. Although rapid inputs of precipitation should mean
this effect is significant, initial assessments suggested that
it might be less important on the hillslope scale because
of the lateral disconnection of wetted areas (Baird and
Wainwright, unpublished modelling results), more recent
literature suggests that the effect is important in irrigation
settings in drylands (Hammecker
et al.
, 2003; Navarro
et al.
, 2008). It is thus also likely to be important in trans-
mission losses in rills and channels, and certainly flood
bores in the latter show the rapid mixing of air and water.
11.4
Runoff generation
11.4.1
Ponding and surface storage
±
6 mm/h) than areas
As a precipitation event proceeds, the first stage of the
runoff process is the ponding of water at the surface.
Where local irregularities produce pits in the surface,
these pits will form depression storage. Continuing rain-
fall will cause any such storage to fill, and thus produce
interconnected overland flow. The nature of this flow is
then controlled by the surface hydraulics. The time taken
for ponding is a function of initial conditions, precipita-
tion rate and soil conditions. In the simplest case it can
be estimated for constant-intensity rainfall by rearrange-
without termites (51
7 mm/h) by Elkins
et al
. (1986).
Cammeraat
et al
. (2002) also suggested that ant nests in
Spain contributed to changes in infiltration by physical
modification of the soil, but that chemical changes caused
the behaviour to vary according to moisture conditions
(and, thus, the time of year). The breakdown of plant ma-
terials by the ants produces conditions of water repellency
in the soil, so that, for initially dry conditions, areas with
ant nests will have relatively low infiltration rates, and
±
