Predicting contaminant migration within the vadose zone, for performance or risk assessment, requires estimates of unsaturated hydraulic conductivity for field soils. Hydraulic conductivities, K, were experimentally determined as a function of volumetric moisture content, 0, for Hanford sediments. The steady state head control method and an ultracentrifuge method were used to measure K(0) in the laboratory for 22 soil samples. The van Genuchten model was used to fit mathematical functions to the laboratorymeasured moisture retention data. Unsaturated conductivities estimated by the van Genuchten-Mualem predictive model, using the fitted moisture retention curve and measured saturated hydraulic conductivity, K s, were compared to those obtained by a scaled-predictive method that uses a single K(0) measurement as a match point near the dry regime. In general, the measured K values and those predicted from van Genuchten-Mualem relationships showed considerable disagreement. This suggests that the use of laboratory-measured K s results in an inadequate characterization of K(0) for the desired range of moisture content. Deviations between the measured and predicted K were particularly severe at relatively low moisture contents; for some samples, there were differences in excess of 2 orders of magnitude at low 0. However, use of the same moisture retention curve-fitting parameters and a single steady state head control-based K(0 ) measurement near the dry regime resulted in considerable improvement. In fact, for the coarse-textured soils considered in this study, results indicate that a K(0) measurement near the dry regime must be used to obtain reliable estimates of unsaturated K at low 0. The study provided important insight on application of two different experimental techniques of measuring unsaturated conductivities.

Stochastic models typically assume that the saturated hydraulic conductivity K s is lognormally distributed, whereas a is normally distributed [Yeh et al., 1985a[Yeh et al., , 1985b Mantoglou and Gelhat, 1987]. However, unlike for K s, there are little data to justify a particular distribution for a. Nonetheless, White and Sully [1992] noted that since Ks and a are related to an internal length scale of the soil's pore geometry, we can expect a priori on physical grounds that they will be correlated and that a 1567

The ideas and suggestions of Dr. Norman Hubbard (ONWI) are greatly appreciated. The experimental efforts of many researchers have been reviewed and incorporated into this paper. Appreciation is expressed to the principle investigators below and to the many others involved in their projects:

Abstract. Vadose zone soils in desert environments often contain a high gravel fraction (>2 mm size). Laboratory measurements of moisture retention are typically made on the fine fraction (<2 mm size); the measurements are then corrected for field conditions by accounting for the gravel fraction in the sample. A frequently used correction procedure for gravelly soils is tested. A total of 23 samples, containing gravel fractions up to 80% (by weight), were analyzed in the laboratory. Retention data on <2 mm size fraction were obtained and compared with measurements on bulk (soil and gravel) field samples. Results indicate that for the relatively wet moisture regime considered in this study, retention data for gravelly soils can be estimated reasonably well by correcting the pressure cell measurements for the field bulk density and the fraction of fines present in the field samples. FUrther work is needed on testing the correctio n procedure for the dry moisture regime.

The views presented in this paper and conclusions drawn are solely those of the authors at Pacific Northwest Laboratory and should not be interpreted as representing those of the numerous technical contributors from the above institutions. Several colleagues supplied detailed and useful comments on the draft version of this document. We wish to acknowledge the time expended and quality of the comments from Mr.