Study on the role of lateral unsaturated flow in triggering slope failure under varying boundary water level conditions

Highlights

• This study examined the effects of lateral unsaturated flow on shallow landslides triggering in situations of rising water level at the uphill/downhill boundary.

• Lateral unsaturated flow is important for slope stability assessment of loamy hillslopes but can be ignored for sandy hillslopes.

• The effect of lateral unsaturated flow highly depends on slope angle and the rising rate of uphill boundary water level.

• Intensive rainfall may weaken the role of lateral unsaturated flow in activating slope failure.

Abstract

Lateral unsaturated flow was found to contribute little to rainfall-induced shallow landslides, but it can be important when the landslides are triggered under conditions of varying boundary water level. By considering a hillslope with an impermeable bedrock and rising water level at the uphill/downhill boundary, this study explored how lateral unsaturated flow would affect slope stability. Three different Hillslope-Storage Boussinesq models that differ in their treatment of the unsaturated flow were coupled with the same infinite slope stability model to predict and compare the safety factor and instability time. Specific to the situations considered, this study discovered that, despite the inclusion of lateral unsaturated flow effect, the hillslope remains stable when water level at the downhill boundary rises. Under the condition of increasing uphill boundary water level, the consideration of lateral unsaturated flow leads to earlier slope failure for loamy hillslopes with a moderate slope. Also, the lateral unsaturated flow may not induce slope failure for mildly sloping loamy hillslopes and plays an ignorable role when landslides occur fast in very steep hillslopes. By contrast, the lateral unsaturated flow does not affect much the failure of sandy hillslopes. Moreover, when water level at the uphill boundary increases faster, the lateral unsaturated flow exerts a greater influence on initiating slope failure. Further examination shows that intensive rainfall may weaken the effect of lateral unsaturated flow on shallow landslides triggering under the condition of rising uphill boundary water level.

Introduction

Landslides, as one of the most widespread geohazards in the world, are an important process of landscape formation by releasing sediments from slopes to allow transportation through the fluvial systems (Iverson et al., 1997). Therefore, landslides help to maintain the long-term dynamic equilibrium between erosion and uplift (Singh et al., 2010). However, for humans, landslides are devastating disasters that cause not only enormous economic losses for destroying buildings and infrastructure (Kjekstad and Highland, 2009), but also, more severely, thousands of deaths or injuries per year worldwide (Petley, 2012). Given these consequences, researchers have conducted many studies to predict the likelihood of landslides, by assessing slope stability, to warn people living in susceptible landslide areas (Restrepo et al., 2008).

Landslides can be triggered by various factors, of which prolonged or intense rainfall is the primary one. Direct rainfall infiltration through the unsaturated soil leads to the propagation of the wetting front, which increases the pore water pressure and soil moisture content, reduces the shear strength of the soil, increases the sliding force, ultimately initiating slope failure (Baum et al., 2010; Saito et al., 2010). Early research has explored how different rainfall characteristics would affect the initiation of shallow landslides, mainly including rainfall intensity (Althuwaynee et al., 2018; Chang et al., 2017; Saito et al., 2010), rainfall duration (Aleotti, 2004; Chen et al., 2015; Godt et al., 2006; Lu and Godt, 2008), and temporal patterns of rainfall (Fan et al., 2020; Ran et al., 2018). Associated with the rainfall infiltration, flow in the unsaturated zone is predominantly vertical, and the lateral unsaturated flow is negligible for barely contributing to landslide activation (Talebi et al., 2008).

According to Iverson (2000), besides direct rainfall infiltration at the hillslope surface, lateral flow caused by saturated groundwater flow from adjacent materials may also induce slope failure. In situations of river damming and reservoir filling, the water level at the downhill boundary of a hillslope will increase, thereby deteriorating the slope stability (Kaczmarek et al., 2015; Shen et al., 2018; Wang et al., 2014). The statistics of Riemer (1992) revealed that about 85% of the landslides around reservoirs are triggered either at the stage of construction and filling or in the first few years of operation after filling. During the operation or filling of reservoirs, the boundary water level change influences the hydrostatic groundwater pressure inside the hillslope and may cause landslides near these artificial water bodies (Zhang et al., 2010). Kaczmarek et al. (2015) discovered that the filling and exploitation of the Włocławek Reservoir in northern Poland led to intensified landslide activities.

The water level at the uphill boundary of a hillslope may also increase. When there are cracks or ponds at the uphill boundary and the soil layer is covered by a very thin layer of fine-grained soil, vertical infiltration into the soil matrix is ignorable and rainfall flows into the cracks or ponds, leading to a rising uphill boundary water level and lateral flow in the soil matrix (Galeandro et al., 2014; Leach and Moore, 2015; Xu et al., 2016). Xu et al. (2016) showed that the water level increase in the subvertical crack at the uphill boundary had contributed considerably to the initiation of Kualiangzi landslide in the Sichuan Basin, China. Subsequently, they proposed a significant negative correlation between uphill boundary water level and slope stability.

Although lateral unsaturated flow can be ignored in rainfall-induced shallow landslides, it may be critical when landslides are activated by the above-mentioned water level variations at the uphill or downhill boundary of a hillslope. Under these conditions, lateral rather than vertical unsaturated flow controls the drainage flux in the unsaturated zone, subsequently the change of groundwater pressure heads and moisture content in hillslopes. Early research showed that the consideration of lateral unsaturated flow improved the prediction accuracy of hillslope drainage process (Kong et al., 2016). A common way to evaluate slope stability is to use the groundwater pressure heads and moisture content derived by hillslope hydrological models as the input parameters of a slope stability model. Therefore, lateral unsaturated flow is likely to be important for slope failure induced by varying boundary water levels. However, previous work did not investigate this problem specifically. The hydrological trigger is a major type of shallow landslides triggers and the lateral unsaturated flow is a nonignorable factor influencing the hillslope hydrological process. In this regard, it is worth examining the impact of lateral unsaturated flow on triggering slope failure under varying boundary water level conditions.

Although the role of lateral unsaturated flow in initiating slope failure remains unexplored, it has been found to be important for hillslope hydrology in many ways. Lateral unsaturated flow commonly exists in hillslopes, it dramatically affects the redistribution of soil moisture and contributes to the nonnegligible drainage flux in the unsaturated zone. Gannon et al. (2017) proposed that lateral unsaturated water flux led to the formation of spatial soil heterogeneity in the headwater catchment at the Hubbard Brook Experimental Forest in New Hampshire. By integrating lateral unsaturated flow into the Hillslope-storage Boussinesq (HSB) model, Kong et al. (2016) achieved more accurate prediction of the hillslope drainage process (e.g., groundwater outflow rates, hydrographs). Lateral unsaturated flow is also critical for recession flow analysis. Luo et al. (2018) revealed that neglecting lateral unsaturated flow would result in an inaccurate estimate of the slopes of recession curves in certain situations. More importantly, with lateral unsaturated flow considered, there could be three power-law regimes (two transition points) of the recession curves, instead of two power-law regimes (one transition point) that were reported in most previous studies.

Due to the strong link to hillslope hydrology, it is anticipated that lateral unsaturated flow may play a role when slope failure is activated under the conditions of varying boundary water level, but to what extent remains unclear. Therefore, in this study, we hypothesized that the triggering of shallow landslides may be attributed to lateral unsaturated flow in situations of varying boundary water level at the uphill or downhill boundary, and the extent depends on several factors, mainly including slope angle, soil type, and hillslope length.

To test the hypothesis, we coupled three HSB models, which differ in their treatment of the unsaturated flow, with the same infinite slope stability model and ran a series of simulations. The HSB models we selected were proposed by Troch et al. (2003) (Troch-HSB model, which neglects unsaturated flow), Hilberts et al. (2005) (Hilberts-HSB model, which considers vertical unsaturated flow) and Kong et al. (2016) (Kong-HSB model, which considers both vertical and lateral unsaturated flow), respectively.

Note that in some studies, the slope stability model was coupled with 2-D/3-D models that account for variably saturated flow based on the Richards’ equation (Cai and Ugai, 2004; Collins and Znidarcic, 2004; Hussain et al., 2007). However, these models are computationally expensive even for small-scale problems. Meanwhile, uncertainties of subsurface properties make the parametrization of these models rather difficult (Bresciani et al., 2014). In comparison, Boussinesq-type equations are 1-D and can be analytically solved for various problems, so their computational efficiency is higher (Rocha et al., 2007; Troch et al., 2002). Therefore, we focused on the HSB models instead of those based on the Richards’ equation.

Following previous research, the three HSB models were used to simulate the groundwater pressure heads and soil moisture content of a hillslope with landslide potential. Simulation results were then used by the same infinite slope stability model to calculate the time-varying slope safety factor (FS), which is the ratio of shear stress available to downslope shear stress, and the instability time (if landslides occurred), which corresponds to the time when FS falls below one. We set up numerous cases to include the factors that would affect the role of lateral unsaturated flow, e.g., soil type, slope angle, slope length, boundary conditions and rainfall intensity. The research outcomes would promote a deeper understanding of slope failure mechanisms and assist in more accurate evaluation of shallow landslides.