Comparison of three stream tube models predicting field-scale solute transport

Abstract. In this paper the relation between local- and field-scale solute transport parameters in an unsaturated soil profile is investigated. At two experimental sites, local-scale steady-state solute transport was measured in-situ using 120 horizontally installed TDR probes at 5 depths. Local-scale solute transport parameters determined from BTCs were used to predict field-scale solute transport using stochastic stream tube models (STM). Local-scale solute transport was described by two transport models: (1) the convection-dispersion transport model (CDE), and (2) the stochastic convective lognormat transfer model (CLT). The parameters of the CDE-model were found to be lognormally distributed, whereas the parameters of the CLT model were normally distributed. Local-scale solute transport heterogeneity within the measurement volume of a TDR-probe was an important factor causing field-scale solute dispersion. The study of the horizontal scale-dependency revealed that the variability in the solute transport parameters contributes more to the field-scale dispersion at deeper depths than at depths near the surface. Three STMs were used to upscale the local transport parameters: (i) the stochastic piston flow STM-I assuming local piston flow transport, (ii) the convective-dispersive STM-II assuming local CDE transport, and (iii) the stochastic convective lognormal STM-III assuming local CLT. The STM-I considerably underpredicted the field-scale solute dispersion indicating that local-scale dispersion processes, which are captured within the measurement volume of the TDR-probe, are important to predict field-scale solute transport. STM-II and STM-III both described the field-scale breakthrough curves (BTC) accurately if depth dependent parameters were used. In addition, a reasonable description of the horizontal variance of the local BTCs was found. STM-III was (more) superior to STM-II if only one set of parameters from one depth is used to predict the field-scale solute BTCs at several depths. This indicates that the local-scale solute transport process, as measured with TDR in this study, is in agreement with the CLT-hypothesis.