Predicting the effects of landuse on water quality – Stage I

Add groundwater component to SPARROW (NIWA, Lincoln Ventures, Objective 2)

Physical process descriptions
Review of SPARROW model concepts
SPARROW description of groundwater processes
Recommended parameter values
Initial results

Physical process descriptions

Nature of the source of N

This section is concerned with nitrogen as a diffuse surface source that is transported through groundwater, ultimately to surface waters. Almost all the nitrogen transported by this pathway is in the form of nitrate ions, because this is the end product of a series of biochemical processes that convert organic nitrogen to the oxidised mineral state. Nitrate is highly mobile in subsurface waters (unsaturated and saturated) because its negative charge is repelled by the generally negative charge of most soils and other subsurface materials.

The biochemical processes for nitrate production occur almost entirely within the organic soil layer and plant root zone near the land surface. The amount of nitrate available for leaching downwards from the soil depends on the type of land use, which in turn determines the productivity of the soil ecology. Tracer experiments have demonstrated that only a small proportion of applied nitrogen fertiliser is ever leached directly to groundwater. The correlation between fertiliser use and leached nitrate arises from the increased productivity of the whole agricultural system, as quantified by the amount of nitrogen per land area being cycled through soil, plant, and animal.

Increasing productivity of a soil is usually associated with increasing concentration of nitrate in soil water available for plant uptake. When rainfall or irrigation exceeds the water holding capacity of the soil some of the resident soil water, containing nitrate, is displaced below the active soil layer so that the nitrate is no longer accessible to plants. Observations of nitrate leached from highly-productive dairy farming in New Zealand (Figure 6-1) show that the concentration of nitrate in leachate, from this particular management regime, is relatively constant between regions with different hydrology but the mass leached per area (kg/ha/y) depends on the amount of soil-water drainage.

Leached nitrate mass (kg/ha/y) for dairy farming in different regions can be described by a simple leaching model that demonstrates the small variation of nitrate concentration in the leachate

Figure 6-1: Leached nitrate mass (kg/ha/y) for dairy farming in different regions can be described by a simple leaching model that demonstrates the small variation of nitrate concentration in the leachate (Bidwell, 2002).

It is now accepted practice for environmental management in New Zealand to characterise the environmental effects of nitrate leaching from rural land use in terms of a value of nitrate concentration (g/m³ or mg/L) and a value of mean annual drainage (mm/y). The nitrate mass leached (kg/ha/y), which may be more familiar to the agricultural community, is derived as:

(1)

Source split

Most of the nitrogen on rural land, in the nitrate form as a diffuse area source, is resident within the soil profile. Therefore, the relative amounts that are transported to surface water directly, or via groundwater are determined by the interaction of flow processes with soil water. Preliminary results from recent experiments on a steep hillslope in Waikato indicate that as much as 80% of the total drainage from this kind of catchment has passed through the soil profile to groundwater. Overland surface runoff is infrequent and is a small proportion of total drainage. The remaining portion of the 20% non-groundwater flow is considered to be saturated flow above less permeable subsoil, which has passed through the soil profile but may emerge further down the hillslope. The proportion of direct flow to surface waters would be expected to decrease for less steep catchments and those with more permeable subsoil. One exception to this guideline is for pasture with low permeability subsoil (poorly drained soils) in which artificial subsurface drainage is installed. These drainage systems can rapidly deliver soil water, surplus to the soil capacity, to surface streams. The discharges are point sources to the stream, but are more conveniently treated as diffuse sources at catchment scale.

Installed drainage is more likely on lowland catchments with mild surface slopes, where the cost is justified by more productive pasture and improved animal health. Experimental results in Southland (Monaghan et al., 2000) showed that surface runoff from cattle-grazed pasture contained very little (< 5 kg N/ha/y) nitrate whereas the installed mole-tile drainage system discharged up to 56 kg N/ha/y (for the most intensive treatment). Measurements of water flow from the installed mole-tile system, for one year, (Monaghan et al., 2002) accounted for about 25% – 75% of the total drainage volume (from water balance) of 366 mm, depending on soil type at each of the six experimental plots. The average nitrate concentration (6.9 g/m³) is consistent with the land use (no applied fertiliser, 2.3 cows/ha). Results of this kind suggest that installed subsurface drainage delivers water of the same nitrate quality as is leached to groundwater, but without the opportunity for further nitrogen transformation.

Delivery to groundwater

Flow of soil water drainage through the vadose zone (the unsaturated soil below the plant roots) to groundwater is predominantly vertical because of the constant influence of gravity in relation to any incipient horizontal pressure gradients that are rapidly equalised by small changes in unsaturated water content. Regions of saturated flow can occur in the vadose zone where there are layers with insufficient vertical hydraulic conductivity, or at the boundaries of layers with certain kinds of contrast in water retention characteristics. At these saturated regions water can move with a horizontal component and appear on the land surface as a seep or spring. However, these horizontal flow paths usually constitute only a very small part of the drainage. Even subsoil layers with hydraulic conductivities of only a few millimetres per day, often deemed to be impermeable, are sufficiently conductive to transmit the mean annual drainage in a catchment that, in Waikato, is typically less than 1000 mm/y (~ 3 mm/d). As water moves deeper into the vadose zone, time variations in drainage from the soil layer become more attenuated and flow rates tend more towards time-averaged values.

The vertical delivery of soil-water drainage to the groundwater surface provides recharge that can be spatially variable as it depends on the drainage rates and nitrate concentrations of the land use directly overhead. There is probably very little lateral (horizontal) dispersion in the vadose zone of vertical recharge between areas of different nitrate production, such as a catchment with areas of forestry (low nitrate, lower recharge) and pasture (higher nitrate, higher recharge). This spatial variation of nitrate delivery is relevant to groundwater interaction with streams (see Section 9.1.2).

Attenuation

Attenuation of nitrate, in the present context, refers to processes that convert nitrate to other forms of nitrogen that are not relevant to in-stream processes. Dilution of nitrate, without transformation, is treated separately by SPARROW. For nitrate that is leached below the root zone, denitrification is the only transformation process that provides a permanent sink for nitrogen (Korom, 1992), in the form of conversion to gaseous nitrous oxide and nitrogen. Denitrification is a microbially mediated process, requiring anaerobic conditions, which can be classified into two types according to whether the electron donor is organic (heterotrophic bacteria) or inorganic (autotrophic bacteria). Suitable inorganic reducing agents commonly found in groundwater are manganese (Mn²⁺), iron (Fe²⁺) and sulfides. It is commonly reported that groundwaters containing Fe²⁺ are low in nitrate, and there is some debate about whether this can also occur as an abiotic chemical process (Korom, 1992). Iron and manganese are widespread in New Zealand groundwaters (Daughney, 2003)

The required anaerobic conditions can occur in the saturated regions of the vadose zone, such as in poorly drained soils, in the groundwater itself, or in riparian areas at the aquifer-stream interface. Although these denitrification processes are well recognised and are the subject of experimental investigation there are almost no data that can be applied at catchment scale. However, the possibility of these processes occurring should be recognised in the formulation of the SPARROW model.

Groundwater – stream exchange

Most of the water flowing in streams is from groundwater, but the groundwater “catchment” does not necessarily coincide with the surface topography that is usually considered to delineate a catchment. There are practical difficulties in defining groundwater catchments because (1) they are not observable from the land surface and (2) groundwater flow systems of different magnitudes can be superimposed vertically on one another (Winter et al., 2003). The boundaries of these unconfined groundwater flow systems are defined primarily by the distribution of recharge inflow and the location of outflow surface water bodies rather than geological structures. The boundary of a groundwater system can vary as recharge varies. This disparity between topographical catchment and groundwater extent is especially relevant to lakes and wetlands because they may interact with larger and deeper groundwater systems if the location of the surface water body is further down slope in a topographical catchment. Figure 6-2 illustrates how surface water bodies may have groundwater catchments that differ from the topographical catchment.

Hydrologic section of a hypothetical catchment, showing how surface water bodies can receive groundwater from various parts of the topographical catchment that do not coincide with the upslope areas

Figure 6-2.: Hydrologic section of a hypothetical catchment, showing how surface water bodies can receive groundwater from various parts of the topographical catchment that do not coincide with the upslope areas.

Streams can interact with groundwater by only gain or loss of water within a particular reach, or by gain and loss at different times within the same reach. For some streams on alluvial outwash fans near hills, for example, there is loss of streamflow to groundwater without any interaction between the state of groundwater and the magnitude of this loss.

There is little useful data for quantifying interactions between surface water and groundwater, other than for the more “conventional” catchment for which a workable guideline is that most (> 80% ?) of the recharge through the land surface flows, via groundwater, to the nearest reach of the stream. However, the “unconventional” situations need to be (and can be) recognised within the SPARROW formulation so that simulations can be run for investigating possible causes of anomalous observations of nitrate concentrations in streams. In order to incorporate this kind of modelling capability, cross connections between the groundwater network and the streamflow network do allow for a proportion (not yet quantified) of the nitrate load from an area element to contribute to any of the downstream stream reaches before being subject to instream attenuation.

Review of SPARROW model concepts

The existing SPARROW model (Alexander et al, 2002) has 3 conceptual components that are used to model stream water quality:

1. Source calculation (can be several sources for each stream)

2. Delivery from Source to Stream (or other water body)

3. Attenuation within Stream/Reservoir (or other water body)

The existing SPARROW model can be visualised as in Figure 6-3.

Simple sketch of the SPARROW model

Figure 6-3: Simple sketch of the SPARROW model

The SPARROW model has now been extended to include a simple groundwater component as shown in red in Figure 6.4.

Extension to SPARROW to include groundwater

Figure 6-4: Extension to SPARROW to include groundwater (the red lines represent new components associated with groundwater)

Several points should be noted about the above extension:

1. Sources (S and S) are split between surface and groundwater (Lincoln Ventures are estimating the surface-vs-g/w proportions). Probably expect point sources to be all 100% surface source

2. Need to parameterise the delivery function (D) for percolation of each contaminant to g/w

3. Need to estimate attenuation (A) of each contaminant in g/w

4. In some settings we may need to estimate the fraction of river water which is diverted to g/w (E: red downward link)

5. May need to estimate fraction of g/w flow which is diverted to river water (E: red upward link)

The extended model has been coded as an enhancement to SPARROW, and will be called from the Desktop Tool using the same approach as shown in Figure 5-4. The parameterisation of the groundwater component is described in the following two sub- sections.