Freeze and Witherspoon (1967) studied regional flow systems using numerical simulation for a wide upland plateau draining to a valley. This is a very similar topographic configuration to that of a floodplain draining to a river channel and for that reason their results provide a good basis for considering subsurface flow within floodplain sediments. Figure 5a shows the flow net for a simple system of homogenous permeability. The water table configuration, which closely follows the smooth topography, has a gentle slope. The hinge line (which divides the recharge from the discharge area) lies on the edge of the channel and the entire floodplain is a recharge area. Water-table slope and hydraulic gradient are greatest near the channel in the discharge zone. Throughout the rest of the floodplain, the flow pattern is relatively uniform. Figure 5b shows the water table and flow pattern which results in hummocky terrain, as might arise where the floodplain comprises infilled oxbows. There are numerous sub-basins and water may be discharged locally to the nearest topographic low or flow towards the main channel. The effect of a layer of high hydraulic conductivity is shown in figure 5c. The permeable layer forms a conduit in which flow is concentrated with recharge from the overlying alluvium and a steep upward discharge to the channel.
Figure 5. Flow nets for a floodplain-channel system (After Freeze and Witherspoon, 1967): (a) for a homogenous sedimentary layer and flat terrain; (b) for a homogenous sedimentary layer and hummocky terrain; (c) for a floodplain with a more permeable layer at depth showing bypass flow beneath the upper layer.
In many cases, the floodplain is coupled to its hillslope and the combined drainage system must be considered. Figure 6 summarises the ways in which water can flow across a floodplain to reach the stream channel. Overland flow moving on to the floodplain will infiltrate provided that sufficient storage capacity exists i.e. the water table is not at or close to the ground surface. Infiltration will be maximised where the infiltration capacity of the floodplain soil is high, andvice versa. If the floodplain is completely waterlogged, overland flow will move rapidly across the floodplain to the river (Waddingtonet al., 1993). Subsurface slope drainage may follow one of three main pathways:
1. Where the floodplain sediments are impermeable, but there is a permeable layer at depth, either a gravel deposit below the alluvium or a permeable bedrock, subsurface flow can move quickly under the floodplain alluvium to the stream (cf. Fig. 5c). This is thought to be a very common situation in British floodplains where late Pleistocene gravels are overlain by less permeable Holocene alluvium. There is relatively little flow through the upper alluvium which, though remaining largely saturated, remains somewhat stagnant. In such cases, the potential of the floodplain to act as a nitrate buffer zone may be greatly reduced. A similar example is given in Plenet and Gibert (1992). Examples of deep groundwater upwelling to recharge a valley-bottom wetland are provided by Lloyd and Tellam (1995) and Gilvearet al. (in press).
2. In the absence of a permeable substrate, slope drainage is forced to the surface, moving as overland flow to the channel. Waddington et al. (1993) studied a riparian wetland near Toronto, Canada. Springs emerging at the wetland margin produce rivulets which flow across the riparian zone to the spring. This is a major discharge pathway with the result that much of the wetland’s subsurface buffer potential is effectively bypassed. Where the floodplain terrain consists of hummocks and pools (cf. Fig. 5b), the pattern of surface flow across the floodplain may be complex (cf. Burt and Gardiner, 1984).
3. Only where floodplain sediments are permeable and homogenous will slope water move through the alluvium to the stream in a uniform, unconcentrated manner (cf. Fig. 5a). In terms of buffer zone
27 The hydrological role of floodplains
T.P. Burt
functions which require interaction between water and soil, only this last situation will provide the optimal hydrological condition. Correll (this volume) shows that the depth of the permeable layer is critical: if subsurface flow is too deep, the saturated zone lies well below the soil and the buffer zone is unable to function effectively. Optimally, the depth of permeable substrate above an aquiclude should be sufficient to fully saturate the soil horizons, but without generating overland flow.
Figure 6. Summary of main flow paths by which hillslope discharge moves through a floodplain to reach the channel: (a) for surface water inputs from upslope; (b) for subsurface water inputs from upslope.
Many field studies have used only a single line of piezometers to monitor flow across the floodplain (e.g. Grieveet al., 1995). In most cases the hydraulic gradient does not lie orthogonal to the channel but is aligned downvalley in response to the floodplain gradient (Anderson and Kneale, 1982). In order to sample a given body of water as it moves across the floodplain (for example, to observe progressive loss of nitrate), it is necessary to adopt a three-dimensional approach using a grid of piezometers (see for example, Haycock and Burt, 1993). A single row of piezometers may involve waters of different origin and false conclusions could be made concerning the evolution of the water quality as a result. Sedimentary structures such as infilled oxbows may also complicate flow patterns on a floodplain, again requiring a three-dimensional survey of water table height. Vegetation patch mosaics found on floodplains imply that sedimentary structures and resulting flow paths may be very complex in some instances (Large and Petts, 1994). It is important to emphasise therefore that water does not flow across the floodplain in a simple and uniform manner. Flow paths may well be tortuous and will very often tend in a downvalley direction.
An important outcome is that the pattern of groundwater discharge along a stream will be spatially variable, with certain locations forming important inflow sites (Anderson and Burt, 1978). In headwater basins, diffuse hillslope runoff will tend to concentrate in hillslope hollows (or swales); these are the variable source areas discussed earlier. In such cases non-point (or diffuse) pollution may
flow across the floodplain as concentrated (overland) flow and enter the river channel at a single point source. These sources must clearly form the focus of attention in any buffer zone scheme. A further cause of spatial variation in inflows relates to the thickness and permeability of the floodplain sediments. Where these are thin or impermeable, surface runoff is likely. Thicker, permeable sediments tend to favour subsurface flow and where there is an aquifer below, saturated conditions may occur only at great depth, with important consequences for buffer zone functions (see Correll, this volume).