The Application of time domain reflectometry in solute transport experiments

Abstract

Contaminants can enter groundwater through the unsaturated zone as dissolved solutes. To predict the location and extent of these contaminants, transport parameters such as pore water velocity y and dispersion coefficient D are required. These parameters are often obtained through transport experiments. The goal of this study is to determine y and D using time domain reflectometry (TDR) technique. Using TDR for transport experiments under unsaturated conditions, we investigated the effects of volumetric water content θᵥ, distance of flow path, and draining-wetting history on D. TDR was used to measure θᵥ, and salt concentration in twenty-one unsaturated column experiments. The 105 cm-long column was homogeneously packed with silica sand (particle size: 53 to 425 pm). Ten TDR probes at ten depths were used to obtain in situ breakthrough curves and a chloride electrode was used to measure effluent breakthrough curves at the bottom of the column. A 35 mM NaC1 (sodium chloride) was used as the tracer with 20 mM NaC1 as background solution. We developed a three-parameter expression relating θᵥ, to measured dielectric constant Kₐ: θᵥ =aKₐᵅ + b. This calibration expression fits as closely or better than the "universal polynomial" and is also consistent with the well-known mixing model. For an isotropic soil with homogeneous water distribution, this expression is further simplified to two parameters by taking α = 0.5. The effects of temperature, porosity, soil solid and bound water can be taken into account by varying a and b of the two-parameter expression. TDR measurements have been shown to be sensitive to bound water and not particular sensitive to the other factors. To calculate y and D from breakthrough curves of step-input experiments, a new moment analysis method has been developed. The transport parameters obtained from this new method show a little difference from the parameters determined from the convection-dispersion equation using the CXTFIT model (a published computer program for estimating solute transport parameters from observed breakthrough curves). Our results demonstrated that D is dependent on measurement methods and concentrations of experimental solutions.