Stochastic analysis of moisture plume dynamics of a field injection experiment

Abstract

A vadose zone field injection experiment was conducted in the summer of 2000 at the Hanford Site, Washington. The unique moisture content (θ) database is used to identify the lithology at the field site and to interpret, visualize, and quantify the spatiotemporal evolution of the three‐dimensional (3‐D) moisture plume created by the injection experiment. We conducted a hierarchical geostatistical analysis to examine the large‐scale geologic structure for the entire field site and then investigated small‐scale features within different layers. Afterward, variogram analysis was applied to the θ field measured for seven different days during the injection experiment. Temporal variations of sills and ranges are related to the observed moisture plume dynamics. A visualization of the 3‐D moisture plume evolution illustrates effects of media heterogeneity. Statistics of changes in moisture content as a function of distance reveal large variance near the wetting front; the coefficient of variation increases with decreasing mean. These findings support the gradient‐ and mean‐dependent variability in the moisture content distribution as reported by existing stochastic theories. Spatial moment analysis is also conducted to quantify the rate and direction of movement of the plume mass center and its spatial spreading. The ratio of horizontal to vertical spreading at varying moisture contents suggests moisture‐dependent anisotropy in effective unsaturated hydraulic conductivity, confirming existing stochastic theories. However, the principal directions of the spatial moments are found to vary as the moisture plume evolves through local heterogeneity, a feature that has not been recognized in the theories.