The unsaturated zone at the Idaho National Laboratory is complex, comprising thick basalt flow sequences interbedded with thinner sedimentary layers. Understanding the highly nonlinear relation between water content and hydraulic conductivity within the sedimentary interbeds is one element in predicting water flow and solute transport processes in this geologically complex environment. Measurement of unsaturated hydraulic conductivity of sediments is costly and time consuming, therefore use of models that estimate this property from more easily measured bulk-physical properties is desirable.
A capillary bundle model was used to estimate unsaturated hydraulic conductivity for 40 samples from sedimentary interbeds using water-retention parameters and saturated hydraulic conductivity derived from (1) laboratory measurements on core samples, and (2) site-specific property transfer regression models developed for the sedimentary interbeds. Four regression models were previously developed using bulk-physical property measurements (bulk density, the median particle diameter, and the uniformity coefficient) as the explanatory variables. The response variables, estimated from linear combinations of the bulk physical properties, included saturated hydraulic conductivity and three parameters that define the water-retention curve.
The degree to which the unsaturated hydraulic conductivity curves estimated from property-transfer-modeled water-retention parameters and saturated hydraulic conductivity approximated the laboratory-measured data was evaluated using a goodness-of-fit indicator, the root-mean-square error. Because numerical models of variably saturated flow and transport require parameterized hydraulic properties as input, simulations were run to evaluate the effect of the various parameters on model results. Results show that the property transfer models based on easily measured bulk properties perform nearly as well as using curve fits to laboratory-measured water retention for the estimation of unsaturated hydraulic conductivity.