SMALL-SCALE VARIABILITY IN THE FLOW OF WATER AND SOLUTES, AND IMPLICATIONS FOR LYSIMETER STUDIES OF SOLUTE LEACHING

This paper discusses results from an experiment in which the fluxes of non-reactive solutes and water were monitored in an 8 x 8 array of ad jacent collectors (each of 36 cm(2) area, covering a total area of 0.2 3 m(2)) located at Im depth in a poorly-structured sandy loam soil pro file. Water was applied uniformly to the soil surface at constant rate s of either 4.3 or 19 mm/h, and a pulse of non-reactive solute (chlori de) was added once the flows of water had become steady. Water continu ed to be applied at steady rate until all of the applied solute had be en leached. The breakthrough curves for individual cells were analysed to determine the mean travel velocity and dispersivity. The water flu xes in individual collectors were very stable, but varied by over an o rder of magnitude, with collectors showing particularly rapid flow ten ding to be clustered. About 80% of the total flow was collected from 4 0% of the overall area of the array of collectors. However, there was only two-fold variation between cells in mean travel time velocity. Th is, coupled with a large 'mobile' water content (equivalent to about 7 0% of the porosity) implies that rapid flow through a relatively small volume of macropores was not responsible for transporting a large pro portion of the water and solutes, and was not a major factor in the sp atially variable discharge. We conclude that only a small increase in water-filled pore space was required to conduct the extra water applie d at the faster application rate. The small amount of extra water-fill ed porosity brought into play at the higher flow rate served to increa se the flow velocities through the matrix of water pathways that were conducting water at the slower application rate, rather than acting as a 'bypass' giving rise to very rapid flow velocities. Analysis of the breakthrough curves suggests that small scale hydrodynamic dispersion was the dominant contributor to dispersion at the 'lysimeter' scale. The results have implications for the design of and interpretation of lysimeter experiments and the interpretation of measurements of contam inant fluxes made using drainage samplers. We conclude that in the cas e of structureless sandy soils, lysimeters of the order of Im deep and 1 m diameter are sufficiently large to be considered representative o f a field soil at the 1 m scale at least in situations where macropore flow is not an important mechanisms of solute transport. Comparison o f these results with other lysimeter studies on the same soil conclude d that the nature of the lower boundary of lysimeters has substantial influence on the flow pathways and the consequent breakthrough curves. Drainage samplers which have collection areas much less than 0.1 m(2) are likely to collect water and any dissolved contaminents at rates v ery different from the average flux densities measured at much larger scales, and so require careful interpretation. Finally, various hypoth eses are considered to explain the lateral redistribution of water tha t occurred in the light of the experimental results.