Hysteresis and Uncertainty in Soil Water-Retention Curve Parameters

Accurate estimates of soil hydraulic parameters representing wetting and drying paths are required for predicting hydraulic and mechanical responses in a large number of applications. A comprehensive suite of laboratory experiments was conducted to measure hysteretic soil-water characteristic curves (SWCCs) representing a wide range of soil types. Results were used to quantitatively assess differences and uncertainty in three simplifications frequently adopted to estimate wetting-path SWCC parameters from more easily measured drying curves. They are the following: (1) ${\alpha }^w=2{\alpha }^d$, (2) $n^w=n^d$, and (3) ${\theta }_s^w={\theta }_s^d$, where $\alpha$, $n$, and ${\theta }_s$ are fitting parameters entering van Genuchten’s commonly adopted SWCC model, and the superscripts $w$ and $d$ indicate wetting and drying paths, respectively. The average ratio ${\alpha }^w/{\alpha }^d$ for the data set was $2.24\pm 1.25$. Nominally cohesive soils had a lower ${\alpha }^w/{\alpha }^d$ ratio ($1.73\pm 0.94$) than nominally cohesionless soils ($3.14\pm 1.27$). The average $n^w/n^d$ ratio was $1.01\pm 0.11$ with no significant dependency on soil type, thus confirming the $n^w=n^d$ simplification for a wider range of soil types than previously available. Water content at zero suction during wetting (${\theta }_s^w$) was consistently less than during drying (${\theta }_s^d$) owing to air entrapment. The ${\theta }_s^w/{\theta }_s^d$ ratio averaged $0.85\pm 0.10$ and was comparable for nominally cohesive ($0.87\pm 0.11$) and cohesionless ($0.81\pm 0.08$) soils. Regression statistics are provided to quantitatively account for uncertainty in estimating hysteretic retention curves. Practical consequences are demonstrated for two case studies.