Numerical and Field Evaluation of Soil Water Sampled by Suction Lysimeters

Porous cup suction-lysimeters are widely used for extracting soil water in solute transport monitoring, bat it is not clear how the sampled concentration can change with the ambient solute concentration, or with applied suction inside the lysimeter. This research was designed to numerically evaluate soil water sampled by suction lysimeters and to test simulation results with field observations. An axisymmetric three-dimensional finite element model (SWMS_2D) was used to simulate soil water sampled by suction lysimeters in a Zimmerman fine sand (mixed, frigid Argic Udipsamments). To test the simulated results, KBr was uniformly spread on the soil surface. After 183 mm rainfall, suction was applied to the lysimeters (with ceramic cups at a depth of 0.8 m) to withdraw soil water; soil column samples were also removed from the surrounding soil and sectioned to determine solute concentration in situ using a bromide electrode. Simulation showed that when there is a concentrated solute band near a porous cup, as can occur with banded fertilizers, the concentration of sampled soil water in the lysimeter is considerably higher than the resident concentration in the soil at the same depth. The influence of a concentrated solute band on the concentration of sampled soil water decreased as the distance to the cup increased. Variations in applied suction (25, 35, and 45 kPa) appear to have little influence on the amount and concentration of field lysimeter extracted soil water. The measured amount of water in the lysimeter (78.4 mL) was comparatively more than the simulated amount (53.9 mL) when the mean water content of nine soil columns was used as an initial condition for the simulation, but was within the range of simulated amounts (32.5–91.5 mL) when the lower and upper 95% limits of the measured soil-water content profiles were used as initial conditions for the simulations. Measured Br⁻ concentration of the sampled water (3.13 × 10⁻⁴ M) was lower than the simulated concentration (4.66 × 10⁻⁴ M) when the geometric means of each layer from the nine soil column Br⁻ concentrations were used as the initial conditions. The difference between measured and simulated concentration may have been due to variability in soil hydraulic properties.