• soil heterogeneity;
  • vadose zone;
  • water flux;
  • recharge;
  • soil particle-size distribution;
  • spatial moments


One-dimensional flow simulations were conducted at four locations of the shallow alluvial aquifer of the upper Rhine River (at the Erstein polder) to quantify the time-dependent moisture distribution, the water flux and the water volume infiltrated in the unsaturated zone as a function of soil heterogeneities during a five-day-long flooding event. Three methods of estimating the hydraulic parameters of soil in the vadose zone were tested. They are based on the following: (1) experimental data, (2) soil particle-size distribution and (3) pedology information on soils. Water fluxes calculated from modelling approaches 2 and 3 were compared with those of the experiment-based values and the effect of these differences on the arrival time and velocity of water at the water table were analysed. Major differences in water fluxes were found among the methods of estimating the hydrodynamic parameters. At the Terrace location, the groundwater recharge predicted using soil data from methods 1 and 2 are approximately 4500 and 2400 mm, respectively. Flow simulations using soil data and the experiment-based method show the highest velocities of infiltrating water at the soil surface and largest volume of groundwater infiltration but result in the lowest centres of the moisture content mass. The results obtained using soil data based on the pedological method are similar to those calculated using soil parameters based on the particle-size distribution of extracted soil samples. Water pressure profiles calculated on Terrace and Channel location, 3 and 7 days after the inundation event agreed reasonably well with those observed when using hydrodynamic parameters from the experiment-based method. However, the flow model using the pedology-based parameters largely underestimates the time needed to achieve hydrostatic conditions of the soil water profile once water flooding at the soil surface stops. This can be mainly attributed to the low values of estimated van Genuchten parameter α. Copyright © 2012 John Wiley & Sons, Ltd.