Geostatistical characterization of unsaturated hydraulic conductivity using field infiltrometer data

Estimation of water infiltration and retention in surficial soils is a critical aspect of many geotechnical and environmental site evaluations. The recent development of field-usable tension infiltrometers now allows insitu measurements of unsaturated hydraulic conductivity (K-U), thus avoiding some uncertainties associated with remolded soil samples tested in the laboratory. Several different geostatistical ''mapping'' methods can be used to spatially characterize K-U, including ordinary and indicator kriging, as well as spatial simulations that provide realizations (stochastic images) of K-U that exhibit more natural variability than do the smoothed spatial estimations of kriging. Multivariate procedures, such as cokriging and Markov-Bayes simulation, can incorporate information from a secondary attribute (e.g., particle size information) to enhance the spatial characterization of an undersampled K-U field. These geostatistical procedures are demonstrated and compared for a case study at a 700 sq. meter site comprised of coarse soil material. Results indicate that percent-by-weight fractions can be used effectively to enhance insitu spatial characterization of K-U.