Geoscience Reference
In-Depth Information
(2003; Kirchner
et al.
2010) considers this to be a “paradox”: how do many catchments store water for
such long periods of time but so easily mobilise so much “old” water during storms.
In Section 5.5.3, the kinematic wave model for subsurface flow was used to explain how the displace-
ment of subsurface water was a result of the difference between the velocity of flow and the wave speed
or celerity of the response. Since the effective storage coefficient controlling the celerity is always less
(and sometimes two orders of magnitude less) than the saturated porosity controlling the mean pore water
velocity, the effect of an input propagates downslope faster than the mean pore water velocity, resulting
in the displacement of stored water. An extreme case is that of a saturated pipe filled with soil. The
effective storage coefficient is close to zero, so the celerity is close to infinite. Pouring water in at one
end immediately causes a displacement at the other end, clearly of “old” water rather than “new” water,
in this case.
This explanation is useful conceptually but the situation on real hillslopes is somewhat more complex
because of the space and time variability of both unsaturated zone and saturated zone flows and the
interaction between them, complicated by the possibility of a dynamic capillary fringe above the water
table. In addition, both the unsaturated and saturated flows may involve a very wide range of pore water
velocities, from preferential flows in (often discontinuous) macropores to relatively immobile storage
in the soil matrix. Account also needs to be taken of the distributed nature of the inputs of new water
to the system. Inputs near the base of the slope might contribute to the hydrograph at the same time as
displacements of old water are taking place.
There is some evidence of this type of complex response from small-scale artificial tracer experiments
carried out on undisturbed soil cores. Reeves
et al.
(1996), for example, looked at the case where a “return
flow” (applied to the base of a core and labelled by a fluorobenzoate tracer) mixed with water stored in the
core and water added from a sprinkler system above (labelled by a different fluorobenzoate tracer). The
core had been taken from a valley bottom hollow and brought to saturation prior to the experiments and,
at the end of the tracer runs, the return flow was switched to a methylene blue dye solution to stain the
flow pathways of the return flow. The results showed that the pathways of the return flow were strongly
preferential in the lower part of the core, while the mix of tracers in water overflowing from the surface
of the saturated core suggested that some of the rainfall added was infiltrating into the soil and displacing
stored water, despite the core being saturated. This suggests some rather local displacement circulations
during such an event, even on a saturated surface. Local displacement in generating overland flow was
also inferred from field evidence by Iorgulescu
et al.
(2007).
There is also the possibility of deep subsurface pathways contributing to the hydrograph (Loague
et al.
,
2005; Ebel
et al.
, 2008). Clearly, the inference of hydrograph sources, flow pathways and residence times
of water in a catchment remains complex (McDonnell
et al.
, 2010).
11.7
Case Study: End Member Mixing with Routing
The Haute-Mentue catchment data described in Section 11.6 have been used in a slightly more sophisti-
cated modelling study of source identification by Iorgulescu
et al.
(2005, 2007) that has also tried to allow
for routing times in the catchment for the different sources. The catchment is in Switzerland, on the plateau
area north of Lausanne, with a strong differentiation of the chemistry of the soil and bedrock (Figure 11.2).
Six subcatchments were monitored; the results in this case study come from the Bois-Vuacoz (24 ha).
Isotope and chemical concentrations were sampled in rainfall and streamflow during a sequence of events
in a wetting-up period at the end of the summer. Thus it was expected that the storages in the system
would be dynamic and that the effective concentrations for the different sources would change over time.
Thus, a model was proposed that included the partitioning of inputs into a direct runoff component (DP),
a soil water component (AS) and a groundwater component (GW). The partitioning was made a nonlinear
function of the storage in the soil. Stationary transfer functions with fast and slow components were used
to route all three components to the stream. The concentrations in each component are assumed to evolve
