Ž . Geoderma 86 1998 61–81 Using an inverse method to estimate the hydraulic properties of crusted soils from tension-disc infiltrometer data ˇ a, ) b a Jirı Simunek , Rafael Angulo-Jaramillo , Marcel G. Schaap , ˇ´ ˚ Jean-Pierre Vandervaere b , Martinus Th. van Genuchten a a U.S. Salinity Laboratory, USDA, ARS, 450 West Big Springs Road, RiÕerside, CA 92507-4617, USA b ( Laboratoire d’etude des Transferts en Hydrologie et EnÕironment LTHE, CNRS URA 1512, ´ ) INPG, UJF BP 53, 38041 Grenoble Cedex 9, France Received 15 July 1997; accepted 24 March 1998 Abstract An inverse procedure was used to estimate the soil hydraulic characteristics of a two-layered soil system—soil surface crust and subsoil—from data obtained during a tension-disc infiltration experiment. The inverse procedure combined the Levenberg–Marquardt nonlinear parameter optimization method with a numerical solution of the axisymmetric variably-saturated flow equation. The objective function was defined in terms of the cumulative infiltration curve and the final water content measured directly below the tension-disc infiltrometer at the end of the experiment; this final water content was assumed to correspond to the final supply pressure head. We analyzed two infiltration experiments carried out with a 25-cm diameter tension-disc infiltrom- eter. One experiment was carried out on a two-layered system, and a second after removal of the surface crust covering the sandy subsoil. Both experiments were performed with six consecutive supply tensions. We first analyzed the infiltration experiment for the subsoil only, thus yielding its hydraulic characteristics. Subsequent analysis of the infiltration experiment for the two-layered system with known hydraulic properties of the subsoil provided estimates of the hydraulic properties of the surface crust. We further compared the estimated hydraulic parameters of the w subsoil with those obtained using Wooding’s analytical method Wooding, R.A., 1968. Steady x infiltration from a shallow circular pond. Water Resour. Res. 4, 1259–1273 and predictions based on a neural network model requiring textural input information. All three methods generated roughly the same results. The numerical inversion technique proved to be a convenient tool for ) Corresponding author. Tel: q1-909-369-4865; Fax: q1-909-342-4964; E-mail: jsimunek@ussl.ars.usda.gov 0016-7061r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved. Ž . PII: S0016-7061 98 00035-4