Application of geophysics to predict rainfallinduced dynamic changes in the hydraulic properties of an unsaturated soil slope

A large number of geotechnical, agricultural and environmen-tal site investigations need knowledge of the distribution of hydraulic properties of the unsaturated soils and their temporal changes. Unsaturated hydraulic properties are also important in climate modelling and in the study of biological and chemical processes in the vadose zone. The ability of the soil to retain or transmit water and its constituents is reflected by these properties. In hillslopes, the flow of water, controlled by the hydraulic properties of the soil, is crucial for water resource management and for predicting slope failure caused by heavy rainfall. One of the most important hydraulic properties is the water retention curve or the soil-water characteristic curve (SWCC), which relates volumetric water content or degree of water saturation to matric suction. Water content and matric suction are related to other important soil properties including: hydraulic conductivity function, cohesion and internal friction (e.g., Guo, 1985; Lu et al., 2012; Robinson et al., 2017). The stability of an unsaturated soil slope or a river embankment subjected to rainfall is determined to a great extent by SWCC. Recently, it has been shown that changes in seismic shear-wave velocity (VS) in unsaturated soils can be used to monitor in situ the distribution of water and the dynamic water transportation in the vadose zone (Suzaki et al., 2017). In this research, among various models for SWCC, the van Genuchten (1980) model was assumed: (1) where s is matric suction, SW is degree of water saturation, a and n are van Genuchten parameters that depend on the soil-type. These describe, respectively, the air-entry suction and the slope of SWCC. A Bishop-type model for effective stress was coupled with a well-tested form of the stress-suction-saturation-depend-ence of the small-strain shear modulus G0 (Sawangsuriya et al., 2009; Han and Vanapalli, 2016), which gave (2) where x is a sample point representing a certain spatial location in the subsurface, and A and B are fitting parameters that depend on the soil-type. More recently, it has been possible to invert the SWCC parameters a and n from G0, SW and bulk density rb obtained from geophysical measurements (Suzaki et al., 2019a, 2019b). In this article, we first illustrate the accuracy in the result of SWCC inversion that one might expect when the input data is realistically noisy. Once the in-situ SWCC is obtained by inver-sion, in the next stage, considering a projected time series for the future rainfall, the distribution of hydraulic properties, viz SW and s, within the soil embankment at different times in the future is calculated through seepage analysis. This provides an important piece of information to assess the vulnerability of a dyke in the future, given an anticipated rainfall. We analyse the prediction error and its cause. Considering Coulomb failure for a sand embankment, we examine the predicted factor of safety (FOS) and its temporal evolution in comparison with the true values.