Sampling the unsaturated zone can serve as an early warning system for groundwater contamination. However, one of the greatest uncertainties in monitoring groundwater contamination is the possibility that solutes flowing in preferred paths bypass samplers. Thus scaling point-source measurements to field-scale estimates may be unreliable and misleading even in the case when geostatistical methods are employed.

Methods for continuously sampling solutes in the unsaturated zone involve the collection of soil water drained by either the force of gravity (e.g., gravity pan samplers, agricultural tile lines, and shallow wells) or by applying a "capillary" suction (e.g., porous cup samplers, wick lysimeters). Whilst all these sampling techniques result in collection of solutes and water in the unsaturated zone, only tile lines and pan samplers measure the amount moving to the aquifer. Gravity pan samplers collect the percolate from a saturated portion of the soil immediately in contact with the sampler, and thus may lead to "bypassing" (i.e., solutes and water flowing around the sampling device).

The figures below (Figures 1a and 1b) illustrate how dramatic differences in sampled concentration of solutes can occur within a very small volume. Four suction lysimeters were installed at a depth of 60 cm, and two each at 90 cm, 120 cm, 150 cm and 180 cm depths in a sandy loam near Freeville, New York. The soil was noted as having worm holes. The depth to groundwater was 2 m. A 4 cm pulse of bromide, with a concentration of 8000 mg/l, was applied on first day of the experiment and this was followed by 4 cm irrigation each day. The suction lysimeters located at 60 cm were affected by preferential flow paths directly connected to the surface (some induced along the sampling tubes) giving rise to the widely varying breakthrough curves (Figure 1a). The remaining sampling devices did not appear to have this problem. Based on the Convection-Dispersion model, one would expect to see the solute peaks decrease in concentration with depth. However, as shown in Figure 1b, the peak concentration observed in the capillary fringe at 180 cm, for instance, occurred prior to the peak seen at 150 cm, which, in turn, arrived prior to the maximum concentration seen at the 120 cm depth. This pattern was caused by preferential flow of water and solutes from the finer soils at the 90 cm depth through the coarse soil directly to the groundwater table, bypassing the soil suction lysimeter.

Figure 1a. Breakthrough curves for four porous cup samplers located at the 60 cm depth

 

Figure 1b. Breakthrough curves for porous cups at 90, 120, 150, and 180 cm. Each line is the average of two replicates

There are currently two general approaches, claimed to be independent of the sampling device, that can be used to sample solute flow in the field. The first is Bouma's (1990) morphological approach in which a Representative Elementary Volume (REV) is used. Soil volumes can be considered "representative" if they are large enough so that individual flow path differences can be averaged. According to Lauren, et al. (1988), sampling volumes of 30 peds or more result in "representative" samples. Thus, REV's characterise "an average flow path" in which the soil heterogeneities can be studied as a stochastic or statistical phenomena. However, in situations where concentrations on the order of parts per billion are considered (e.g. pesticide transport), then this approach will be inadequate. Rather, individual preferential pathways that may be responsible for the transport of chemicals must be sampled. In this case, the approach by Barcelona and Morrison (1988) to locate the sampling devices in the likely pathways of water and contaminants is better. When the groundwater is at shallow depth, then the use of artificial tile lines for sampling are useful even though the tile lines integrate samples over a field scale.

A modification of the gravity pan sampler is the Alundum tension plate sampler, in which the percolate is extracted from the unsaturated soil by suction applied across a alundum filter disc. Fibreglass wick lysimeters operate by the same principle. The wicks are self-priming and act as a hanging water column, thus providing a suction to the unsaturated soil. The wicks are also non-reactive and therefore can be used to sample solutes. Original wick sampling units consisted of a single fibreglass wick spread over a 30 cm x 30 cm area. The disadvantage of this design is that solutes entering the sides of the sampling unit had to travel a considerable distance to the center, whilst solutes near the center, flowed without delay, thus giving rise to large instrument dispersion. This design was improved by Boll, et al. (1991) and others giving rise to a multi-segment percolation system shown by Figure 2. These units can be installed in situ in field sites or alternatively undisturbed soil cores from selected sites can be extracted and transported to the laboratory for detailed study. Twenty-five individual fibreglass wicks are placed on a 5 x 5 grid on the basal surface area of the sampling unit (see Figures 3a, 3b, 3c).

Figure 3b. Three dimensional view of the alloy-cast base-plate installed with spring-loaded wick lysimeters prior to mounting on soil column

 

Figure 3c. 12 V variable velocity, X-Y scanning irrigation unit. The unit is mounted on top of the soil column and delivers uniform rainfall. Note that the flux is controlled by peristaltic pump.

The base-plate is then firmly pressed against the soil surface by springs. The length of the wick provides a capillarity equivalent to that found in the soil and thus can be used to sample unsaturated flow. Being a porous medium, the wicks have been shown to provide boundary conditions which mimic those found in the undisturbed soil. For instance, the capillary force in the wick decreases with increasing flux, thus eliminating flow - field distortions created by suction-cup lysimeters.

Barcelona, M. J., & Morrison, R. D. (1988). Sample collection, handling and storage: Water, soils and aquifer solids. In D. W. Nelson & R. H. Dowdy (Ed.), Methods for Groundwater Quality Studies, Proceedings of the National Workshop, Agricultural Research Division, University of Nebraska, Lincoln, Nebraska. 49 - 62.

Boll, J., Selker, J. S., Nijssen, B. M., Steenhuis, T. S., Van Winkle, J., & Jolles, E. (1991). Water quality sampling under preferential flow conditions. In R. G. Allen, T. A. Howell, W. O. Pruitt, I. A. Walter, & M. E. Jensens (Ed.), Lysimeters for Evapotranspiration and Environmental Measures, Proceedings American Society of Civil Engineers , International Symposium on Lysimetry. New York City. American Society of Civil Engineers. 290 - 298.

Bouma, J. (1990). Using morphometric expressions for macropores to improve soil physical analysis of field soil. Geoderma, 46, 3 - 11.

Lauren, J. G., Wagenet, R. J., & Wisten, J. H. M. (1988). Variability of saturated hydraulic conductivity in a Glossaquic Hapludalf with macropores. Soil Science, 145, 20 - 28.