Estimation of Soil Hydraulic and Solute Transport Parameters from Transient Field Experiments using Inverse Modeling

Estimation of unsaturated soil hydraulic and solute transport properties by Inverse modeling has thus far been limited mostly to analyses of one-dimensional experiments in the laboratory, often assuming steady-state conditions. This is partly because of the high cost and difficulties in accurately measuring and collecting adequate field-scale data sets, and partly because of difficulties in describing spatial and temporal variability in the soil hydraulic properties. In this study we estimated soil hydraulic and solute transport parameters from several two-dimensional furrow irrigation experiments under transient conditions. Three blocked-end furrow irrigation experiments were carried out, each of the same duration but with different amounts of infiltrating water and solutes resulting from water depths of 6, 10, and 14 cm in the furrows. Two more experiments were carried out with the same amounts of applied water and solute, and hence for different durations, on furrows with water depths of 6 and 10 cm. The saturated hydraulic conductivity (Ks) and solute transport parameters in the physical equilibrium convection-dispersion (CDE) and physical nonequilibrium mobile/ immobile (MIM) transport models were inversely estimated using the Levenberg-Marquardt optimization algorithm in combination with the HYDRUS-2D numerical code. Estimated Ks-values ranged from 0.0389 to 0.0996 cm min-1, with a coefficient of variation of 48%. Estimated immobile water contents (θim) were more or less constant at a relatively low average value of 0.025 cm3 cm-3, whereas the first-order exchange coefficient (ω) varied between 0.10 and 19.52 min-1. The longitudinal dispersivity (DL) ranged from 2.6 to 32.8 cm, and the transverse dispersivity (DT) from 0.03 to 2.20 cm. DL showed some dependency on water level and irrigation/solute application time in the furrows, but no obvious effect was found on Ks and other transport parameters. Agreement between measured and predicted infiltration rates was satisfactory, whereas soil water contents were somewhat overestimated and solute concentrations underestimated. Differences between predicted solute distributions obtained with the CDE and MIM transport models were relatively small. This finding and the value of optimized parameters indicate that observed data were sufficiently well described using the simpler CDE model, and that immobile water did not play a major role in the transport process.