Estimating Hydraulic Properties of a Fine-textured Soil Using a Disc Infiltrometer

Inverse optimization of parameters offers an economical means to infer soil hydraulic properties from in situ measurements of infiltration. We evaluated optimization strategies to inversely estimate soil hydraulic parameters using field measured tension disc infiltrometer data. We estimated the parameters n, , and Ks of the van Genuchten-Mualem (VGM) model, and a piecewise representation of conductivity near saturation using a numerical inversion of Richards' equation. In addition to cumulative infiltration, optimizations included in the objective function water retention data, water contents from cores extracted after termination of infiltration, or transient measurements of water contents using time domain reflectometry (TDR) probes. Three-parameter fits to field data were non-unique because of a positive correlation between and Ks. In contrast, fits of n and Ks with estimated from separate fits to retention data improved parameter identifiability while not compromising the fit to measured infiltration. Inverse optimizations that included in the objective function both water retention and cumulative infiltration, led to excellent fits of this data when initial volumetric water contents were >0.23 cm cm . Close fits to cumulative infiltration were also obtained at lower water contents, however, water retention data was underestimated likely because of hysteresis. Optimizations of cumulative infiltration with final soil core water content or TDR data led to estimates of final water contents that closely approximated measured water contents. However, measured TDR water contents were poorly matched by simulations at early times. A piecewise loglinear interpolation of hydraulic conductivity near saturation improved fits to measured cumulative infiltration and water retention data as compared with using the VGM model at all pressure heads. P DERIVED from in situ measurements of soil hydraulic properties are crucial to understanding and describing the dynamic processes of water flow in the field. The tension disc infiltrometer (Perroux and White, 1988) has become a valuable tool to investigate the hydraulic properties of soils at or near the surface. R.C. Schwartz and S.R. Evett, USDA-ARS, Conservation and Production Research Lab., P.O. Drawer 10, Bushland, TX 79012-0010. Received 16 Aug. 2001.*Corresponding author ([email protected]). Preprint from Soil Sci. Soc. Am. J. 66:1409-1423 (2002). This infiltration-based method is particularly suitable for quantifying changes in near surface hydrology resulting from soil management activities such as tillage (Sauer et al., 1990; Logsdon et al., 1993). Although unconfined flow below the infiltrometer disc complicates the analyses of infiltration measurements, various methods have been devised to infer hydraulic properties from disc infiltrometer measurements. These techniques are based on quasi-analytical solutions of transient flow at early times (e.g., Smettem et al., 1994), Wooding's (1968) analysis of steady state infiltration from a disc source (Ankeny et al., 1991; Logsdon and Jaynes, 1993; Hussen and Warrick, 1993), or by inverse parameter optimization of the axisymmetric form of Richards' equation (Šim nek and van Genuchten, 1996). Angulo-Jaramillo et al. (2000) discuss many of the difficulties associated with both the transient and steady state analysis of unconfined infiltration. The focus of this paper is the estimation of soil hydraulic properties through inverse parameter optimization of the governing equations that describe water flow from a disc source. Inverse procedures tend to be less restrictive than direct analysis using quasi-analytical solutions and have the potential to yield information about conductivity and water retention over a wide range in pressure heads from a single infiltration experiment. Šim nek and van Genuchten (1996, 1997) proposed an inverse method to estimate hydraulic properties using cumulative infiltration data from a disc infiltrometer. Based on the results of inverse fits to numerically generated data, they concluded that identifiability of parameters is improved when other information is included in the objective function. The most promising scenario was an objective function that included initial and final water contents as well as cumulative infiltration data. Final water contents were assumed to be in equilibrium with the supply pressure head and taken at the soil surface upon the termination of infiltration experiments. Šim nek et al. (1998b) later used this method in conjunction Abbreviations: TDR, time domain reflectometry; VGM, van GenuchtenMualem. 1410 SOIL SCI. SOC. AM. J., VOL. 66, SEPTEMBER-OCTOBER 2002 t z K( ) h( ) z 1 1 r r r K( ) h( ) r [1] (h) r s r