Parameter Estimation Analysis of the Evaporation Method for Determining Soil Hydraulic Properties

oped (Wind, 1968; Becher, 1971; Boels et al., 1978; Schindler, 1980; Tamari et al., 1993; Wendroth et al., Soil hydraulic properties are important parameters affecting water 1993; Halbertsma and Veerman, 1994). flow in variably saturated soils. We estimated the hydraulic properties from a laboratory evaporation experiment using both a parameter An important modification was suggested by Wind estimation method and the modified Wind method. The parameter (1968), who introduced an iterative graphical procedure. estimation method combined a one-dimensional numerical solution He first estimated the water retention characteristic of the Richards equation with the Marquardt–Levenberg optimization from average water content and pressure head readings scheme. In our study we used both numerically generated data and at several locations in a homogeneous soil sample, and data measured in the laboratory. Two experiments were carried out subsequently determined hydraulic conductivities from on 10-cm-high soil cores containing two different soils. Pressure heads the measured pressure head profile and changes in the inside the cores were measured with five tensiometers, while evaporawater content distribution. The water content profile tive water loss from the top was determined by weighing the soil was obtained from results of the first step. The Wind samples. The objective function for the parameter estimation analysis iterative method was later automated by several rewas defined in terms of the final total water volume in the core and pressure head readings by one or several tensiometers. An analysis searchers (e.g., Boels et al., 1978; Halbertsma and Veerof numerically generated data showed that the optimization method man, 1994). Wendroth et al. (1993) developed a method was most sensitive to the shape factor (n ) and the saturated water that required measurement of the pressure head at only content (us) and least to the residual water content (ur). Pressure two locations. In the traditional setup, only unsaturated heads measured close to the soil surface were found to be more hydraulic conductivities in the pressure head range from valuable for the parameter estimation technique than those measured approximately 250 to 2700 cm could be determined. at lower locations. The optimized hydraulic parameters corresponded Good estimates of the conductivity close to saturation closely with those obtained using Wind’s analysis. All optimizations could not be obtained because the hydraulic gradients gave similar results for the soil hydraulic properties within the range were too small. To overcome these problems near satuof measured pressure heads (0 to 2700 cm). Extrapolation beyond this ration, Wendroth et al. (1993) imposed two different range involved a high level of uncertainty because of high correlation between parameters ur and n. evaporation rates: initially a relatively high evaporation rate to obtain large pressure head gradients near saturation and, after reestablishing hydraulic equilibrium, much lower evaporation rates as controlled by the preM laboratory and field methods exist to devailing laboratory conditions. Using numerically genertermine soil hydraulic properties, especially for ated data, several researchers (Tamari et al., 1993; Wenthe unsaturated hydraulic conductivity (Klute and Dirkdroth et al., 1993; Mohrath et al., 1997) evaluated the sen, 1986; Green et al., 1986). Most methods remain assumptions of the evaporation method or the effects relatively time consuming and costly, and are often limof experimental errors. ited to relatively narrow ranges of water content. One An alternative method of analyzing transient flow fairly simple laboratory method for simultaneous estiduring an evaporation experiment is to use parameter mation of both retention and unsaturated hydraulic conestimation techniques (Kool et al., 1987). Methods of ductivity data for the past 30 yr has been the evaporation this type typically involve the coupling of a numerical method. This method was first introduced by Gardner model for variably saturated water flow with a parameand Miklich (1962), who imposed a series of constant ter optimization algorithm. The Levenberg–Marquardt fluxes on one side of an initially equilibrated sample, method (Marquardt, 1963) has been especially popular and measured the pressure head response of two tensifor this purpose. Starting with the studies of Zachmann ometers. The flux needed to be sufficiently small to et al. (1981) and Dane and Hruska (1983), the parameter assume a constant hydraulic conductivity and diffusivity estimation method is increasingly being used for estiin the sample (Halbertsma and Veerman, 1994). Becher mating unsaturated soil hydraulic functions. Computer (1971) simplified the evaporation method by using a models applicable to laboratory column outflow meacontinuous evaporation rate. Several other modificasurements have been given by Kool et al. (1985a,b) and tions of the evaporation method with simultaneous meaParker et al. (1985) for one-step outflow procedures, surements of evaporation rate and pressure heads at and by van Dam et al. (1992, 1994) and Eching and different heights in the sample have since been develHopmans (1993) for multistep approaches. A more general parameter estimation model applicable to transient J. Šimůnek and M.Th. van Genuchten, U.S. Salinity Lab., USDAwater flow subject to less restrictive initial and boundary ARS, 450 West Big Springs Road, Riverside, CA 92507-4617; O. Wendroth, Inst. for Soil Landscape Research, ZALF, Eberswalder conditions was developed by Kool and Parker (1987). Str. 84, D-15374 Müncheberg, Germany. Received 15 Aug. 1997. Applications to evaporation experiments have been pre*Corresponding author ([email protected]). sented by Feddes et al. (1988), Ciollaro and Romano (1995), and Santini et al. (1995). In a review, Feddes et Published in Soil Sci. Soc. Am. J. 62:894–905 (1998).