An important requirement for accurate modeling of field-scale unsaturated transport using local-scale parameters is a method to account for multiscale heterogeneity and the connectivity of the facies that control flow. A series of infiltration tests were conducted at the Hanford Site along a 60-m-long transect, instrumented at 1-m intervals to measure water content (θ) and capillary pressure head (h) to quantify the spatial correlation structure of hydraulic properties and assess connectivity of the facies. Hydraulic parameters including the pore connectivity–tortuosity tensor (Li) were inversely estimated using the Subsurface Transport Over Multiple Phases (STOMP) numerical simulator coupled with the parameter estimation code, UCODE. Results show that six of eight parameters required for a modified van Genuchten–Mualem model could be inversely estimated using θ measured during transient infiltration from a surface line source and approximated prior information. Soils show evidence of saturation-dependent anisotropy that was well described with the connectivity tensor. Variability of the vertical saturated hydraulic conductivity, Ksv, was larger than the horizontal, Ksh. The autocorrelation ranges for Ksh, Ksv, the inverse of the air-entry value, α, and the horizontal connectivity, Lh, were between 2.4 and 4.6 m, whereas the van Genuchten shape parameter, n, and saturated water content, θs, showed no autocorrelation. Accurate upscaling of hydraulic properties requires the correct assessment of the connectivity of facies.