Experimental study to estimate the criteria for shallow landslides under various geological conditions

doi:10.21203/rs.3.rs-3113503/v1

Download PDF

Research Article

Experimental study to estimate the criteria for shallow landslides under various geological conditions

https://doi.org/10.21203/rs.3.rs-3113503/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 20 Nov, 2023

Read the published version in Environmental Earth Sciences →

You are reading this latest preprint version

The purpose of this study is to experimentally estimate the criteria for shallow landslide occurrence using hydrological indicators such as matric suction and volumetric water content for representative soils with the different geological condition in which landslides frequently occur in South Korea. To investigate the detection criteria for shallow landslides, a series of landslide model tests are conducted using weathered soils obtained from regions of granite, gneiss, and mudstone where landslides occur. A landslide model test device, which includes a rainfall simulator, a slope model flume, and a measurement system with sensors, is developed to simulate shallow landslides that generally occur on natural slopes during rainfall. Based on the results of the model test, an infinite slope stability analysis considering the suction stress of unsaturated soil is applied to analyze changes in the safety factor of the slopes according to rainfall. Using the domestic standard of slope design used in South Korea, landslide detection criteria based on safety factor of slope are recommended as 1.3 for attention-level alerts and 1.0 for warning-level alerts. The matric suction corresponding to the attention and warning levels are defined as critical matric suctions, and the volumetric water contents corresponding to the critical matric suctions on the soil-water characteristic curve (SWCC) are defined as critical volumetric water contents. The proposed critical matric suctions and critical volumetric water contents can potentially be used as basic data to detect the time of shallow landslide occurrence and issue the landslide early warning.

shallow landslide

model test

geological condition

hydrological indicator

infinite slope stability analysis

safety factor of slope

Almost 70% of the Korean peninsula consists of mountainous terrain, and the upper soil layers of natural slopes above the bedrock layer tend to consist of residual soils that are produced via the weathering of bedrock and colluvial soils that have been deposited from collapses further upslope. Most upper soil layers on natural slopes in South Korea tend to be up to 2 m in depth and upper soil layers are formed through the weathering of rock found in the local area. So, soil slope behavior is greatly influenced by geological condition. Especially, most landslides in South Korea occur in granite, gneiss and sedimentary rock areas (Kim and Song, 2015).

Due to these topographical conditions and soil properties, landslides that occur on natural slopes are usually initiated along the interface between the bedrock and the soil layer. Therefore, the most common type of landslide that occurs on natural slopes in South Korea are shallow landslides with translational failure planes at depths of 2m or less (Kim and Song, 2015; Song et al., 2021). Shallow landslides are generally caused by the descent of the wetting front in the soil layer during rainfall and are less influenced by rising groundwater levels (Anderson and Sitar, 1995; Lu and Godt, 2008; Sorbino and Nicotera, 2013; Song et al., 2016; Kim et al., 2021). Wetting fronts initially form on the ground surface and descend into the soil layer of natural slopes, resulting in the saturation of the soil layer and decreasing the matric suction of the soil. The increased unit weight and decreased suction stress of the unsaturated soil affect slope stability by reducing the effective stress of the soil (Zhang et al., 2005; Lu and Likos, 2006; Matsushi et al, 2006; Godt et al., 2009; Xu et al., 2013; Zhao and Zhang, 2014).

To effectively detect shallow landslides that occur on natural slopes in South Korea, a measurement method that considers the characteristics of landslide occurrence is required along with suitable thresholds (Song et al., 2016; Zhang et al., 2019; Pecoraro and Calvello, 2021; Zhan et al., 2022). It is especially important to understand the hydrological and geotechnical properties of soils as rainfall infiltrates into the soil layer of natural slopes. For this reason, changes in hydrological factors like volumetric water content or matric suction need to be investigated to properly understand how unsaturated soil properties of natural slopes are affected by rainfall infiltration.

Some researchers have studied to understand the relationship between shallow landslide occurrence and hydrological factors such as volumetric water content and matric suction of unsaturated soils (Godt et al., 2008; Lu and Godt, 2008; Montrasio et al., 2008; Ray et al., 2010; Damiano et al., 2012; Yeh and Lee; 2013; Comegna et al., 2016; Song et al., 2016; Kim et al., 2021; Pecoraro et al., 2021). Especially, a landslide model test is one of the best way to understand and analyze the phenomenon of shallow landslide caused by rainfall (Lacerda, 2007; Olivares et al., 2009; Damiano and Olivares, 2010; Huang and Yuin, 2010; Lee et al., 2011; Wu et al., 2015; Wu et al., 2017; Sun et al, 2019; Park and Song, 2020; Zhan et al., 2022).

It is very important to estimate the volumetric water content or matric suction of soils before shallow landslides occur. But it is very difficult to measure or estimate the volumetric water content and matric suction before a shallow landslides occur in the field. Therefore, the volumetric water content and matric suction immediately before the occurrence of landslide are estimated through landslide model tests under extreme rainfall conditions, which can be estimated the threshold of shallow landslide.

The model test can cause a landslide occurrence in a short time, and the scale and shape of slope can be selected in consideration of site conditions (Sun et al., 2019; Park and Song, 2020). Also, it is possible to measure the volumetric water content and matric suction of soils during the landslide occurrence process, and the stress state at the time of landslide occurrence can be estimated and inferred.

In this study, for representative soils with the different geological condition in which landslides frequently occur in South Korea, a series of landslide model test are carried out to experimentally estimate the threshold of shallow landslide occurrence using hydrological indicator such as matric suction and volumetric water content. In order to consider shallow landslides caused by rainfall, landslide model tests are conducted to investigate the change in unsaturated soil properties on slopes with topographical and geological conditions similar to those of natural slopes in South Korea. The model tests are performed on weathered soils with the representative geological conditions, such as granite, gneiss and mudstone, where landslides often occur. Based on the results of the model tests and the unsaturated infinite slope stability analysis, detection criteria that can be used to rationally predict initiation times for shallow landslide can be estimated.

Configuration of the model test device

The landslide model test device was designed to simulate shallow landslides that usually occur on natural slopes in South Korea. The test device is capable of simulating infinite slope failures at shallow depths that are caused by descending the wetting front during rainfall. Furthermore, the test device can be used to rationally estimate the landslide initiation time based on how unsaturated soil properties change during rainfall.

Figure 1 shows a schematic diagram of the model test device used in this study. As shown in the diagram, the landslide model test device consists of a slope model flume, a rainfall simulator, and a measurement system with sensors. The rainfall simulator, which consists of an oscillator system, water tank, and rainfall controller, is designed to control the amount of rainfall using a submarine motor pump and a flow regulator. The measurement system consists of sensors that measure volumetric water content and matric suction as well as a data logger. The rainfall simulator and the measurement system with sensors are operated through a computer, where measuring data are also stored. Figure 2 shows a panoramic view of the landslide model test device installed in the laboratory.

Figure 1. Schematic diagram of the landslide model test device

Figure 2. Panoramic view of the landslide model test device

The soil model flume is divided into three parts: a toe part, a slope part, and a crest part. The lengths of each respective part are 0.6 m (toe part), 1.5 m (slope part) and 0.3 m (crest part), and the height of each part is 0.7 m. The edges of each part of the flume are connected and secured with bearings, and a rail was installed on the bottom of the toe part to enable horizontal movement. Furthermore, a lift was installed below the crest part to adjust its height up to 1.78 m. By moving the toe part horizontally while adjusting the height of the crest part, the inclination of the slope model flume could be controlled within the range of 0–50 ^o. Specifically, the slope inclination was adjusted by using the lift to raise the crest part, then the toe part was fixed on the rail to prevent it from moving. Additionally, the inlet of the toe part was designed to open and close to enable a variety of model tests. High-strength glass windows were installed on one side of the slope model flume to facilitate direct observations of slope behavior. Additionally, the non-slip tape on the floor of the flume was attached to prevent slip between the floor and the soil. On the wall opposite to the side with the high-strength glass windows, a hole with a diameter of 5mm was drilled on the wall so that the sensor connection cables could lead out of the flume.

The rainfall simulator consists of an oscillator system, a flow regulator, a submarine motor pump, and a water tank. The oscillator system is composed of three oscillator-type nozzles and was 4 m in height and 5 m in length to ensure that the slope model flume received sufficient levels of artificial rainfall. Furthermore, the oscillator system could be moved along a rail that was attached to the ceiling to spray water on specific parts of the slope. The oscillator-type nozzles installed in the oscillator system were designed based on preliminary experimental results so that the sprinkled rainfall particles were 3–6 mm in size to resemble natural rainfall (Kim et al., 2016). Moreover, the nozzles were designed to swing in a pendulum motion with a specific angle and speed along the length of the flume to ensure that the entire model slope received even amounts of rainfall. The flow regulator was used to control the flow rate and hydraulic pressure, and the submarine motor pump was connected to the water supply pipe to ensure consistent artificial rainfall regardless of the rainfall duration. The intensity of the rainfall produced by the rainfall simulator could be adjusted within the range of 50–200 mm/hr, and a specific rainfall intensity level could be simulated using the rainfall controller operated by rotary valves of the flow regulator.

Measurement sensors

Measurement sensors were installed to analyze variations in the unsaturated soil properties of the model slope when a landslide occurs during rainfall as well as to measure the landslide initiation time. To measure changes in matric suction and volumetric water content caused by rainfall infiltration into the model slopes, tensiometers and time-domain reflectometry (TDR) sensors were installed at various depths in the crest, slope, and toe parts. Based on measured results of matric suction and volumetric water content obtained from these sensors, it is possible to detect and observe the formation and descent of a wetting front.

For the TDR that was used to measure the volumetric water content in the soil layer, the 5TE model produced by Decagon Devices was used. This sensor measures dielectric permittivity in the soil layer and converts it into the volumetric water content. Additionally, TDR can be used to measure electrical conductivity and temperature in the soil layer. The tensiometers were used to measure matric suction in the soil layer, and the MPS-6 model produced by Decagon Devices was used. The small sensors were selected to reduce the disturbance influence caused by the sensor size during the model test. Also, these sensors selected in this test have been used in previous studies (Wu et al., 2017; Sun et al., 2019; Park and Song, 2020). The data logger was the CR1000 model produced by Campbell Scientific, which provides 13 channels. The data logger collects data from each measurement sensor using analog methods. Additionally, it is able to simultaneously call the measuring data of each sensor, and the time interval can be set as short as one second. In this test, the time interval was set as 10 seconds.

The engineering and unsaturated properties of the soils used in the model tests were investigated. In the landslide model test, weathered granite soil, weathered gneiss soil and weathered mudstone soil were used as slope materials. A series of soil laboratory tests were conducted to estimate the engineering properties of each weathered soil, and the soil-water characteristic curve (SWCC) of each weathered soil was also estimated to analyze the unsaturated soil properties.

To estimate the engineering properties of soils used in the model tests, unit weight tests, specific gravity tests, sieve analysis tests, permeability tests, and direct shear tests were carried out. As a result of the soil laboratory tests, the particle size distribution of the weathered granite soil was classified as well-graded sand (SW) according to the Unified Soil Classification System (USCS). Also, the particle size distributions of the weathered gneiss and mudstone soils were classified as poor-graded sand (SP). The total unit weights of the weathered granite, gneiss, and mudstone soils under field conditions were 1.55 t/m³, 1.54 t/m³, and 1.62 t/m³, respectively. Table 1 presents the engineering properties of the weathered granite, gneiss, and mudstone soils through the soil laboratory tests.

Table 1

Engineering properties of the weathered granite, gneiss, and mudstone soils
Properties	Symbol	Unit	Measured values
Properties	Symbol	Unit	Granite soil	Gneiss soil	Mudstone soil
Specific gravity	${G}_{s}$	-	2.68	2.56	2.54
Natural water content	${\omega }$	%	15.90	19.38	27.56
Total unit weight	${\gamma }_{t}$	t/m³	1.55	1.54	1.62
Dry unit weight	${\gamma }_{d}$	t/m³	1.34	1.29	1.27
Effective particle size	${D}_{10}$	mm	0.11	0.097	0.007
Uniformity coefficient	${C}_{u}$	-	6.1	3.9	35.7
Coefficient of curvature	${C}_{g}$	-	1.5	0.8	8.2
USCS	-	-	SW	SP	SP
Coefficient of permeability	${k}_{s}$	cm/sec	1.89×10^− 3	8.03×10^− 4	3.40×10^− 5
Cohesion	$\text{c}{\prime }$	kPa	15.4	11.1	19.8
Internal frictional angle	${\phi }{\prime }$	^o	27.7	32.7	26.1

Table 1. Engineering properties of the weathered granite, gneiss, and mudstone soils

The SWCCs of the weathered granite, gneiss, and mudstone soils were estimated using an automated SWCC apparatus (Song et al., 2010; Song et al., 2012). This apparatus has a simple measurement principle and is capable of continuous measurements. Therefore, the apparatus can accurately calculate the SWCC of an unsaturated soil by minimizing human error that may be generated by the test experimenter. The automated SWCC apparatus consists of a pressure panel, a flow cell, a water reservoir, an air bubble trap, sample preparation accessories, and a measurement system. Furthermore, the apparatus can apply air pressures in the range of 0-300 kPa to the soil. The axis-translation technique (Hilf, 1956) was adapted in this apparatus.

The SWCC of each weathered soil was estimated using the van Genuchten model based on the matric suction of the soil and the amount of pore water measured from the automated SWCC apparatus. van Genuchten (1980) proposed an equation (Eq. (1)) using the relationship between effective degree of saturation (${S}_{e}$) and pressure head ($\text{h}$) based on the formula proposed by Mualem (1976) to estimate the SWCC of an unsaturated soil. Eq. (1) shows the formula for estimating SWCCs proposed by van Genuchten (1980).

$${S}_{e}=\frac{\theta -{\theta }_{r}}{{\theta }_{s}-{\theta }_{r}}={\left[\frac{1}{1+\left\{\alpha {\left({u}_{a}-{u}_{w}\right)}^{n}\right\}}\right]}^{m}$$

Where, ${{\theta }}_{\text{r}}$ is the residual value of the soil water content, ${{\theta }}_{\text{s}}$ is the saturated value of soil water content, ${\alpha }$ is a parameter related to the air entry value (AEV), $\text{n}$ is a parameter related to the slope of the inflection point of the SWCC, and $\text{m}$ is a parameter related to residual water content.

Using the measured matric suction and volumetric water content, nonlinear least squares analysis was applied to calculate the fitting parameters of ${\alpha }$, $\text{n}$, and $\text{m}$ for the van Genuchten model and estimate the SWCC. Figure 3 shows the SWCCs of the weathered granite, gneiss, and mudstone soils, and Table 2 presents the calculated fitting parameters for each weathered soil.

Table 2

Fitting parameters of SWCCs for the weathered granite, gneiss, mudstone soils
Soils	Path	A (kPa^− 1)	n	m	kPa
Granite soil	Drying	0.145	1.990	0.497	2.7
Granite soil	Wetting	0.245	1.712	0.416	1.8
Gneiss soil	Drying	0.299	2.018	0.504	1.7
Gneiss soil	Wetting	0.846	1.601	0.375	0.5
Mudstone soil	Drying	0.050	1.873	0.466	8.0
Mudstone soil	Wetting	0.512	1.506	0.366	1.0

Figure 3. SWCCs of the soils used in model tests

Table 2. Fitting parameters of SWCCs for the weathered granite, gneiss, mudstone soils

Table 3 shows the experimental conditions of the model slopes consisting of weathered granite, gneiss, or mudstone soil that were tested. The total unit weight and initial water content of each soil were derived from the in-situ conditions of natural slopes. As shown in the table, the model slope soils had total unit weights in the range of 1.54–1.62 t/m³ and initial water contents in the range of 10–27%. The inclination of the model slopes was set as 35 ^o, and the rainfall intensity was set as 200 mm/hour to artificially induce landslides. The reason for applying extreme rainfall intensity like 200 mm/hour is to focus on the data at the point of landslide occurrence.

Table 3

Experimental conditions of the model slopes for the landslide model tests
Soils	Total unit weight (t/m³)	Initial water content (%)	Slope angle (^o)	Rainfall intensity (mm/hr)
Granite soil	1.55	10–20	35	200
Gneiss soil	1.54	10–20	35	200
Mudstone soil	1.62	20–27	35	200

Table 3. Experimental conditions of the model slopes for the landslide model tests

Figure 4. Configuration of the model slope including locations of measurement sensors

The model slope was formed as an infinite slope with a long length and shallow soil layer. The infinite slope is a typical natural slope where shallow landslides mainly occur in South Korea. In order to build the model slope, the slope shape and angle were determined first, and soils obtained from natural slopes were set considering the field conditions. A No. 5 sieve (4.75 mm) was used to remove gravels and pebbles from the obtained soils. The prepared soils were laid 5cm at a time in the flume to form the model slope with a depth of 60 cm, and compaction tools such as rubber mallets and spatulas were used to ensure that the unit weight of the soil placed in the model slope coincided with the unit weight of natural soil.

The measurement sensors were also installed to measure variations in the soil layer during the setting process of the model slope. Figure 4 shows the locations and types of sensors that were installed in the model slope. As shown in the figure, three rows of sensors (lines A, B, and C) were installed in each part (toe, slope, and crest) of the model slope. In each row, tensiometers and TDRs were installed at depths of 20 cm, 40 cm, and 60 cm below the soil surface to measure matric suction and volumetric water content.

The rainfall simulator should be stabilized to spray the target rainfall intensity during the test. The slope model flume was initially covered with a waterproof sheet like tarpaulin. Once the rainfall simulator stabilized after reaching the target rainfall intensity, the waterproof sheet on the flume was removed to expose the model slope to the uniform rainfall. As the model slope was exposed to the rainfall, the measurement sensors were used to measure changes in volumetric water content and matric suction in the model slope caused by rainfall infiltration in real time. Based on the measurement results, the property changes of the unsaturated soil in the model slope due to rainfall infiltration and the failure mode of the model slope was investigated.

Failure modes of the model slopes

Figure 5 shows the failure modes of each model slope as time progressed. As shown in the figures, erosion was initially observed on the soil surface during the early stages of rainfall, and then horizontal cracks and local failure began to form at the toe part. As rainfall continued, the local failure at the toe part continued to progress and propagate toward the crest part, ultimately leading to the failure of the entire model slope. Therefore, it was confirmed that the model slope underwent a progressive failure mode where the failure first initiated at the toe part then propagated to the crest part.

Meanwhile, it was shown that the overall slope failure progressed fastest in the slope composed of weathered granite soil and slowest in the slope composed of weathered gneiss soil. Moreover, it was found that all model slopes exhibited similar failure modes regardless of geological characteristics. Therefore, it can be said that the landslide progress rate and the initial landslide occurrence time can vary depending on the soil characteristics related to geological conditions.

Figure 5. Failure modes of the model slopes over time

Changes in matric suction and volumetric water content of soil layer

Figure 6 describes variations in the matric suction of the soil layer in the model slope over time as measured by the tensiometers, which were installed in the model slope at depths of 20 cm, 40 cm, and 60 cm below the soil surface. In order to compare matric suction according to test running time for soil layers with different geological condition, the horizontal and vertical axes of figure are shown on same scale. Through the measurement results, we attempted to understand the saturation rate in the slope during rainfall and analyze the change in matric suction of the soil layer according to slope failure. As shown in the figures, the saturation in the model slope started at the soil surface and steadily descended into the soil layer, whereas the matric suction measured in the slope sharply decreased as saturation progressed. Furthermore, it can be seen that slope failure occurred immediately after saturation, as shown by the measured results that described changes in matric suction in the soil layer.

Figure 6. Changes in matric suction in the soil layers of the model slopes over time

Figure 7 describes variations in the volumetric water content of the soil layers of the model slopes over time as measured by the TDRs, which were also installed in the model slopes at depths of 20 cm, 40 cm, and 60 cm below the soil surface, same as the tensiometers. The TDRs were installed to measure changes in the volumetric water content of the soil layer of the slope during rainfall. The horizontal and vertical axes of figure are shown on same scale to compare volumetric water content according to test running time for soil layers with different geological condition. As shown in the figures, the saturation in the slope began at the soil surface and steadily progressed downward. It was found that slope failure occurred at the location where saturation took place in the soil layer.

Figure 7. Changes in the volumetric water content of the soil layers of the model slopes over time

The matric suction and volumetric water content results measured by the tensiometers and TDRs that were installed at the same positions and depths were also compared. Figure 8 shows both measuring data obtained from the model test with weathered mudstone soil. As shown in the figure, the matric suction sharply decreased to almost zero as the model slope was saturated by rainfall infiltration. In contrast, the volumetric water content increased at a relatively slower rate up to the saturated volumetric water content.

Upon comparing the variations in the volumetric water content and matric suction measurement results that were obtained from the same positions in the model slope as saturation progressed, it was found that the volumetric water content at a given depth increases before the matric suction begins to decrease at the same depth. Whereas the matric suction tends to decrease sharply as saturation progresses, the volumetric water content increases at a relatively slower rate. This is due to differences in the measurement principles of the tensiometers and TDRs in the soil layer. It is judged that tensiometers are more suitable for use in early warning systems for landslide detection. Therefore, when performing landslide monitoring in the field, the features of different sensors should be identified and utilized. In this study, detection criteria for landslide early warnings are proposed based on the matric suction measurements obtained from the tensiometers in the soil layers of model test slopes.

Figure 8. Comparison of matric suction and volumetric water content in the soil layer of the model test slope with weathered mudstone soil

Rainfall infiltration changes the stress state of soils by increasing the volumetric water content and reducing matric suction, eventually leading to slope failure even if the soils is not fully saturated. It is very important to estimate the volumetric water content or matric suction of soils just before shallow landslides occur. But it is very difficult to measure or estimate the volumetric water content and matric suction before a shallow landslides occur in the field. Therefore, the infinite slope stability analysis based on unsaturated soil mechanics was applied to estimate the criteria for shallow landslide using the data of landslide model test.

An infinite slope stability analysis considering suction stress(Lu and Godt, 2008; Song et al., 2016) was employed to calculate the safety factor of unsaturated infinite slopes based on the volumetric water content or the matric suction of the soil layer as obtained from model tests during rainfall. The safety factor of an unsaturated infinite slope considering suction stress can be calculated using Eq. (2). Figure 9 illustrates the analysis section of an infinite slope that considers the shape of the model slope and the positions of the sensors that measured the matric suction and volumetric water content of the soil layer.

$${F}_{s}=\frac{\text{tan}{\phi }^{{\prime }}}{tan\beta }+\frac{2{c}^{{\prime }}}{\gamma zsin2\beta }-\frac{{\sigma }^{s}}{\gamma z}\left(tan\beta +cot\beta \right)tan{\phi }^{{\prime }}$$

The safety factor of slope is recommended as detection criteria for landslides based on the design standard of slope construction works (Ministry of Land, Infrastructure and Transport, 2016) in South Korea. The safety factor of slope criteria for landslide detection were estimated using the results of the landslide model tests and the design standard of slope construction works. As a result of determining the safety factor of an unsaturated infinite slope, the recommended criteria include 1.3 for attention-level alerts and 1.0 for warning-level alerts.

Figure 9. Cross-sectional view of the model test for slope stability analysis

As shown in Eq. (2), suction stress was taken into consideration to analyze changes in the stability of an infinite slope during rainfall. Lu and Likos (2006) proposed a formula for suction stress in unsaturated soils as shown in Eq. (3).

$${\sigma }^{s}=-({u}_{a}-{u}_{w}){S}_{e}$$

Equation (3) can be further expanded by using the relationship between matric suction and volumetric water content or the effective degree of saturation. By substituting Eq. (1) regarding SWCC proposed by van Genuchten (1980) into Eq. (3) and excluding the effective degree of saturation, it is possible to obtain a formula to calculate suction stress for a whole range of matric suction values, as shown in Eq. (4).

$${\sigma }^{s}=-({u}_{a}-{u}_{w}) {u}_{a}-{u}_{w}\le 0 \left(4\text{a}\right)$$

$${{\sigma }}^{\text{s}}=-\frac{{\text{u}}_{\text{a}}-{\text{u}}_{\text{w}}}{{\left(1+{\left[{\alpha }\left({\text{u}}_{\text{a}}-{\text{u}}_{\text{w}}\right)\right]}^{\text{n}}\right)}^{\frac{\text{n}-1}{\text{n}}}} {u}_{a}-{u}_{w}>0 \left(4\text{b}\right)$$

To estimate the landslide detection criteria, suction stress was calculated using the measured matric suction values from the landslide model tests. Figure 10 shows the variation in suction stress in the model slope over time calculated using the matric suction measurements and fitting parameters of unsaturated soil used in the model tests. As shown in the figure, suction stress was greatest in weathered mudstone soil and lowest in weathered granite soil. Additionally, the suction stress converges to almost zero upon saturation.

The general behavior of suction stress in unsaturated soils is associated with the dependency on soil type. The behavior of suction stress is controlled primarily by pore size distribution and pore geometry of soils pores (Lu and Likos, 2004). For coarse soils, the surface adsorption is generally very limited because the specific surface and surface charge properties are relatively small, and an air-entry pressure is lower as controlled by the relative large pore throats formed among soil particles. For fine soils, the surface adsorption is significantly greater because the specific surface is much larger than coarse soil, and the air-entry pressure is larger as controlled by the relatively small.

Figure 10. Variations in suction stress in the soil layer of model slopes over time

The safety factor of slope for attention-level and warning-level alerts for landslides were suggested based on the landslide model test results and the design standard of slope construction works in South Korea. The matric suctions corresponding to the attention level and warning level are defined as critical matric suctions. From the SWCCs, the volumetric water content values corresponding to the critical matric suctions can be calculated, and these values were defined as critical volumetric water contents. Both critical matric suction and critical volumetric water content were proposed as detection criteria for landslides.

The suction stress according to rainfall was calculated based on the results of the landslide model tests, and in turn, the safety factor of an unsaturated infinite slope was calculated using suction stress. Figure 11 shows the slope stability analysis results considering the suction stress of unsaturated soils that were used estimate the detection criteria for landslides. As shown in the figure, the points at which the safety factor of slope reached 1.3 and 1.0 were selected as the detection criteria for landslides. That is, an attention-level alert is issued when the safety factor of slope decreases to 1.3, and a warning-level alert is issued when the safety factor of slope decreases to 1.0.

Table 4 presents the landslide detection criteria that were estimated based on the results of the landslide model tests and the slope stability analysis considering suction stress. The proposed landslide detection criteria can be used for the purpose of landslide risk management in the field. More specifically, it is judged that landslide early warning systems could be implemented at actual monitoring stations constructed on natural slopes by applying the proposed detection criteria based on the volumetric water content or matric suction of unsaturated soils.

Table 4

Proposed detection criteria for shallow landslides
Monitoring item	Matric suction (kPa)		Volumetric water content (%)
Safety factor of slope	1.3 for attention-level	1.0 for warning-level	1.3 for attention-level	1.0 for warning-level
Granite soil	4.05	1.32	36.8	43.0
Gneiss soil	4.21	0.39	29.1	41.2
Mudstone soil	8.48	2.60	29.7	35.4

Figure 11. Slope stability analysis results used to estimate the landslide detection criteria

Table 4. Proposed detection criteria for shallow landslides

In this study, a series of landslide model tests were performed to estimate detection criteria for shallow landslides that occur on natural slopes in South Korea. For this purpose, a model test device capable of simulating shallow landslides caused by rainfall was developed, and model tests were performed using weathered soils obtained from natural slopes with different geological conditions. To analyze changes in the unsaturated soil properties of the slopes during rainfall, the matric suction and volumetric water content of the soil layers were measured using tensiometers and TDR sensors. Based on the results of the model tests, an infinite slope stability analysis was conducted by considering the suction stress of unsaturated soils during rainfall to derive detection criteria for shallow landslides. The findings of the study can be summarized as follows:

In order to detect shallow landslides that occur on natural slopes in South Korea, a landslide model test device consisting of a rainfall simulator, a slope model flume, and a measurement system with sensors was developed.

The failure mode of the model slopes can be described as a progressive collapse where the initial local failure started at the toe part of the slope and then extended and propagated to the crest part. Additionally, the slope failure progressed rapidly through the model slopes in the order of weathered granite soil, weathered mudstone soil, and weathered gneiss soil.

An infinite slope stability analysis considering the suction stress of unsaturated soils was used to estimate changes in the safety factor of slopes during rainfall based on the model test results. Using the landslide model test results and the design standard of slope construction works, detection criteria for shallow landslides based on safety factor of slope were recommended as 1.3 for attention-level alerts and 1.0 for warning-level alerts.

The matric suction values corresponding to the attention level and warning level are defined as critical matric suctions. Also, the volumetric water content values corresponding to the critical matric suctions on the SWCC are defined as critical volumetric water contents.

The critical matric suction and volumetric water content can be used as basic data to detect the time of shallow landslide occurrence, and verification and correction of these values will be carried out through comparison with measurement data in landslide monitoring sites of South Korea in the future.

Acknowledgements

This research was supported by the Basic Research Project (Grant No. 23-3412-1) of the Korea Institute of Geoscience and Mineral Resources (KIGAM) funded by the Ministry of Science and ICT of Korea.

Anderson, S.A. and Sitar, N. (1995) Analysis of rainfall-induced debris flow, Journal of Geotechnical Engineering, ASCE, 121(7), 544-552.
Comegna, L., Damiano, E., Greco, R., Guida, A., Olivares, L. and Picarelli, L. (2016) Field hydrological monitoring of a sloping shallow pyroclastic deposit, Canadian Geotechnical Journal, 53, 1–13.
Damiano, E. and Olivares, L. (2010) The role of infiltration process in steep slope stability of pyroclastic granular soils: laboratory and numerical investigation, Natural Hazards, 52(2), 329-350
Damiano, E., Olivares, L. and Picarelli, L. (2012) Steep-slope monitoring in unsaturated pyroclastic soils, Engineering Geology, 137–138, 1–12.
Godt, J.W. Baum, R.L. Savage, W.Z. (2008) Transient deterministic shallow landslide modeling: requirements for susceptinility and hazard assessments in a GIS framework, Engineering Geology, 102(3-4), 214-226.
Godt, J.W., Baum, R.L. and Lu, N. (2009) Landsliding in partially saturated material, Geophysical Research Letter, 36: L02403.
Hilf, J.W. (1956) An investigation of pore water pressure in compacted cohesive soils, Technical Memorandum No.654, U.S. Department of the interior, Bureau of Reclamation, Design and Construction Division, Denver, CO.
Huang, C.C. and Yuin, S.C. (2010) Experimental investigation of rainfall criteria for shallow slope failure, Geomorphology, 120, 326-338.
Kim, J.G., Kang, J.W., Jeong, G.S., Choi, H. and Kim, M.S. (2016) Suggestion for the optimum operating methods of rainfall simulator through the analysis of raindrop’s characteristics, Journal of the Association of Korean Geographers, 5(1), 51-60 (in Korean with English abstract)
Kim, K.S. and Song, Y.S. (2015) Geometrical and geotechnical characteristics of landslides in Korea under various geological conditions, Journal of Mountain Science, 12, 1267-1280.
Kim, K.S., Jeong, S.W., Song, Y.S., Kim, M. and Park, J.Y. (2021) Four-year monitoring study of shallow landslide hazards based on hydrological measurement in a weathered granite soil slope in South Korea, Water, 2021(13): 2330
Lee, L.M., Kassim, A. and Gofar, N. (2011) Performances of two instrumented laboratory models for the study of rainfall infiltration into unsaturated soils, Engineering Geology, 117(1-2), 78-89.
Lu, N and Likos, W.J. (2004) Unsaturated soil mechanics, John Wiley and Sons Inc., 556p.
Lu, N. and Godt, J.W. (2008) Infinite-Slope Stability under Steady Unsaturated Seepage Conditions, Water Resources Research, 44: W11404.
Lu, N. and Likos, W.J. (2006) Suction stress characteristic curve for unsaturated soil, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 132(2):131-142.
Matsushi, Y., Hattanji, T. and Masukura, Y. (2006) Mechanisms of shallow landslides on soil-mantled hillslopes with permeable and impermeable bedrocks in the Boso Peninsula, Japan, Geomorphology, 76, 92-108.
Ministry of Land, Infrastructure and Transport of Korea (2016) Design standard of slope for construction work, 122p. (in Korean)
Montrasio, L. and Valentino, R. (2008) A model for triggering mechanisms of shallow landslides, Natural Hazards, 8(5), 1149-1159.
Mualem, Y. (1976) A new model for predicting the hydraulic conductivity of unsaturated porous media, Water Resources Research, 12(3):513-522.
Olivares, L., Damiano, E., Greco, R., Zeni, L., Picarelli, L., Minardo, A., Guida, A. and Bernini, R. (2009) An instrumented flume to investigate the mechanics of rainfall-induced landslides in unsaturated granular soils, Geotechnical Testing Journal, 32(2), 108-118
Park, J.Y. and Song, Y.S. (2020) Laboratory experiment and numerical analysis on the precursory hydraulic process of rainfall-induced slope failure, Advances in Civil Engineering, 2020, 2717356.
Pecoraro, G. and Calvello, M. (2021) Integrating local pore water pressure monitoring in territorial early warning systems for weather-induced landslides, Landslides, 18(4), 1191-1207.
Ray, R., Jacobs, J., and de Alba, P. (2010) Impacts of unsaturated zone soil moisture and groundwater table on slope instability, Journal of Geotechnical and Geoenvironmental Engineering, 136(10), 1448-1458.
Smith, J.B., Godt, J.W., Baum, R.L., Coe, J.A., Burns, W.J., Lu, N., Morse, M.M., Sener-Kaya, B., Kaya, M. (2014) Hydrologic monitoring of a landslide-prone hillslope in the Elliott State Forest, Southern Coast Range, Oregon, 2009–2012: U.S. Geological Survey Open-File Report 2013–1283, 61p.
Song, Y.S., Chae, B.G. and Lee, J.T. (2016) A method for evaluating the stability of an unsaturated slope in natural terrain during rainfall, Engineering Geology, 210:84-92.
Song, Y.S., Chae, B.G., Kim, K.S., Park, J.Y., Oh, H.J. and Jeong, S.W. (2021) A landslide monitoring system for natural terrain in Korea: Development and application in hazard evaluations, Sensors, 2021(21):3040
Song, Y.S., Cho, Y.C. and Hong, S. (2016) Analyses on variations in the unsaturated characteristics of a mine waste-dump slope during rainfall, Environmental Earth Science, 75, 1106.
Song, Y.S., Hwang, W.K., Jung, S.J. and Kim, T.H. (2012) A comparative study of suction stress between sand and silt under unsaturated conditions, Engineering Geology, 124:90-97.
Song, Y.S., Lee, N.W., Hwang, W.K. and Kim, T.H. (2010) Construction and application of an automated apparatus for calculating the Soil-Water Characteristic Curve, Journal of Engineering Geology, 20(3):281-295. (in Korean with English abstract)
Sorbino, G. and Nicotera, M.V. (2013) Unsaturated soil mechanics in rainfall-induced flow landslides, Engineering Geology, 165, 105-132.
Sun, P., Wang, G., Wu, L.Z., Ogwe, O. and Zhu, E. (2019) Physical model experiments for shallow failure in rainfall-triggered loess slope, Northwest China, Bulletin of Engineering Geology and the Environment, 78, 4363-4382.
van Genuchten, M.T. (1980) A closed-form equation for predicting the hydraulic conductivity of unsaturated soils, Soil Science Society of America Journal, 44:892-898.
Wu, L.Z., Huang, R.Q., Xu, Q., Zhang, L.M. and Li, H.L. (2015) Analysis of physical testing of rainfall-induced soil slope failure, Environmental Earth Science, 73, 8519-8531.
Wu, L.Z., Zhou, Y., Sun, P., Shi, J.S., Liu, G.G. and Bai, L.Y. (2017) Laboratory characterization of rainfall-induced loess slope failure, Catena, 150, 1-8.
Xu, L., Qiao, X.J., Wu, C.X., Iqbal, j. and Dai, F.C. (2012) Causes of landslide recurrence in a loess platform with respect to hydrogeological process, Natural Hazards, 64,1654-1670
Yeh, H.F. and Lee, C.H. (2013) Soil water balance model for precipitation-induced shallow landslides, Environmental Earth Science, 70(6), 2691-2701.
Zhan, Q., Wang, S., Guo, F., Chen, Y., Wang, L. and Zhao, D. (2022) Early warning model and model test verification of rainfall-induced shallow landslide, Bulletin of Engineering Geology and the Environment, 81, 318
Zhang, L.L., Zhang, L.M. and Tang, W.H. (2005) Rainfall-induced slope failure considering variability of soil properties, Geotechnique, 55(2), 183-188.
Zhang, S., Zhang, X., Pei, X. and Wang, S. (2019) Model test study on the hydrological mechanisms and early warning thresholds for loess fill slope failure induced by rainfall, Engineering Geology, 258, 105135.
Zhao, H.F. and Zhang, L.M. (2014) Instability of saturated and unsaturated coarse granular soils, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 140(1), 25-35.

No competing interests reported.

Download PDF

Journal Publication

published 20 Nov, 2023

Read the published version in Environmental Earth Sciences →

Editorial decision: Major revision
21 Sep, 2023
Reviews received at journal
20 Jul, 2023
Reviewers agreed at journal
18 Jul, 2023
Reviewers invited by journal
18 Jul, 2023
Editor assigned by journal
14 Jul, 2023
Submission checks completed at journal
30 Jun, 2023
First submitted to journal
27 Jun, 2023

You are reading this latest preprint version

Experimental study to estimate the criteria for shallow landslides under various geological conditions

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Landslide Model Test Equipment

Configuration of the model test device

Measurement sensors

Geotechnical Characteristics of Slope Model

Experimental Conditions and Method of the Model Tests

Results and Analysis of the Model Tests

Failure modes of the model slopes

Changes in matric suction and volumetric water content of soil layer

Analysis and Discussion of Detection Criteria for Shallow Landslides

Conclusions

Declarations

Acknowledgements

References

Additional Declarations

Status:

Journal Publication

Version 1