The response of pore water pressure to snow accumulation on a low-permeability clay landslide

Engineering Geology 242 (2018) 130–141

Contents lists available at ScienceDirect

Engineering Geology journal homepage: www.elsevier.com/locate/enggeo

The response of pore water pressure to snow accumulation on a lowpermeability clay landslide

T

⁎

Takashi Okamotoa, , Sumio Matsuurab, Jan Otto Larsenc, Shiho Asanoa, Kazutoki Abed a

Forestry and Forest Products Research Institute, 1 Matsunosato, Tsukuba, Ibaraki 305–8687, Japan Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto 611–0011, Japan c University Centre in Svalbard, P.O. Box 156, N-9171 Longyearbyen, Norway d College of Bioresource Sciences, Nihon University, 1866 Kameino, Fujisawa, Kanagawa 252–0880, Japan b

A R T I C LE I N FO

A B S T R A C T

Keywords: Snow load Field-based monitoring Quick clay Undrained loading Excess pore water pressure Permeability

A snowpack is known to affect landslide stability as the snow loading changes. Furthermore, the snowpack can have effects on the hydrological behavior in the landslide mass. To clarify the response of pore water pressure to snow loading, continuous field-based monitoring of hydraulic and meteorological factors in landslides was conducted for a deposit of extremely low-permeability “quick clay” in Mid-Norway. The pore water pressure increased during every snow-covered period. The pressure exhibited little response to meltwater and/or rain, but corresponded closely to changes in snow accumulation. The increase in pore water pressure during the snowcovered period was considered to be excess pore water pressure generated by undrained snow loading on the extremely low-permeability quick clay. The pore water pressure displayed a positive linear relation with the snow load, and the ratio of the increase in the pore water pressure to the snow load (rusnow) was 0.49–0.53. These values show that approximately half of the snow load contributed to the excess pore water pressure. Continuous field-based monitoring was also conducted for another landslide in relatively high-permeability deposits in Japan, where the pore water pressure showed a relatively low rusnow of about 0.15 and a different timing of the pressure peak. This comparative result indicates that the response characteristics of pore water pressure to the snow loading are strongly affected by the permeability of the landslide mass. Although the excess pore water pressure generated by the snow load theoretically had a negative effect on the slope stability, the value of excess pore water pressure at the monitored landslide was relatively too small to affect its stability.

1. Introduction Landslides located in snowy regions are strongly affected by snow conditions as well as by rainfall. Snow influences landslide stability and movement in many different ways (Fig. 1); the most common trigger related to snow is snow melting (Fig. 1 (A)). Rapid snow melting caused by sudden warming or rainfall on snowpack is a major factor in landslide occurrence because of the resulting increase in pore water pressure at the sliding surface (Coe et al., 2003; Wieczorek, 1996). Cases of snowmelt-triggered landslides have been reported in snowy regions around the world (e.g., Chigira and Chiba, 1998; Gokceoglu et al., 2005; Naudet et al., 2008). Snow loading is also a significant factor influencing slope stability (Fig. 1 (B)). When snow accumulates on a slope, the snow load generates both total stress (Fig. 1 (C)) and shear stress (Fig. 1 (D)). Nakamura and Shiraishi (1973) monitored landslide movement during snow-covered periods and observed an increase in landslide activity

⁎

during the early snow-covered period. Matsuura et al. (2003) reported that a landslide became inactive with an increase in snow accumulation, based on their field monitoring at the Busuno landslide in Japan with a maximum snow depth of 5 m. Okamoto et al. (2008) analyzed infinite slope stability with the addition of a snow load term, and suggested that the snow load had both positive and negative effects on slope stability, depending on the gradient and internal friction angle of the sliding surface. In contrast, several studies have reported the possibility that the snow load acts not only on the sliding surface, but also on the pore water pressure in the landslide mass (Fig. 1 (E)). Maruyama (1993) observed an increase in pore water pressure during mid-winter at the Sarukuyoji landslide in a heavy snow region in Japan. Matsuura (2000) observed similar fluctuations in pore water pressure during the same period at the Busuno landslide, Japan. Both studies suggested the possibility that the snow load generated excess pore water pressure. However, previous studies provided qualitative descriptions of

Corresponding author. E-mail addresses: okataka@ffpri.affrc.go.jp (T. Okamoto), [email protected] (S. Matsuura), shiho03@ffpri.affrc.go.jp (S. Asano), [email protected] (K. Abe).

https://doi.org/10.1016/j.enggeo.2018.06.002 Received 4 September 2017; Received in revised form 9 April 2018; Accepted 1 June 2018 0013-7952/ © 2018 Elsevier B.V. All rights reserved.

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Fig. 1. Diagram of snow effects on landslide stability. The focus of this study is indicated by the shaded box.

observed cases, but did not address the fluctuations and response characteristics of pore water pressure. An increase in pore water pressure caused by the snow load had a negative effect on the slope stability. Assessment of the mechanism by which increased pore water pressure causes landslides is significant for landslide risk assessment in snow-covered regions. We established a landslide research site in Norway, where there is a deposit of extremely low-permeability “quick clay”, and have been conducting field-based monitoring of hydrological factors, including pore water pressure, at 15-min intervals since 1997 (Okamoto et al., 2004). From the monitored results, we detected an increase in pore water pressure in response to snow accumulation during a three-year period. The response was more obvious than that reported in previous studies. Here, we describe the detailed process of pore water pressure response to snow accumulation and discuss its response characteristics by comparison with other landslides with different geological backgrounds.

2.2. Roesgrenda landslide We established a research site for landslide monitoring in Roesgrenda in Mid-Norway (latitude 63.5°N, longitude 11.5°E), 60 km northeast of Trondheim. A topographic map of the research site is provided in Fig. 2. The landslide area is on a slope along the Helgåa River at an altitude of 50–100 m above sea level [a.s.l.]. Although three relatively large-scale landslides (> 10,000 m3 in volume) occurred in 1995 and 1996, only small- to medium-scale shallow landslides (< 3000 m3 in volume and < 30 m in length) were found during our monitoring between 1997 and 2001 (Table 1). The landscape of the scarp is shown in Fig. 3. The slope of the scarp spreads out in a bowl shape with a steep inclination of 30–40°. At the bottom of the scarp is a gentle talus slope (10–20°) that was formed by deposition of collapsed soil. A geological profile of the site is shown in Fig. 4. The base rock is Paleozoic strata overlain by a 10-m-thick layer of moraine deposits, a 25- to 30-m-thick quick clay layer, and then a 10-m-thick layer of river deposits consisting of clay, silt, and gravel. The crown and upper flat area above the scarp are covered in coniferous forest. A brownish-red forest topsoil is found at the ground surface with a depth of < 0.5 m. The physical properties of the quick clay and river deposits in the site are listed in Table 2. The water content (w) of the quick clay is 25.1%, which exceeds its liquid index (wL) of approximately 16% (Kristoffersen, 1999), indicating that the quick clay is in a fragile condition. The saturated hydraulic conductivity (Ksat [m/s]) of the quick clay was estimated to be 10−9 to 10−12 m/s (Larsen, 2002; Long, 2005), which is extremely low. The Ksat of the river deposits has not been measured; however, it is assumed to be greater than that of the quick clay because a small amount of groundwater is discharged from the thin sand layer within the river deposits. The Ksat value of the topsoil at the research site is not known; however, the Ksat values of forest topsoils are similar, independent of the geological characteristics, and are widely recognized to be approximately 10−3 to 10−5 m/s from laboratory permeability tests (e.g., Hayashi et al., 2006; Noguchi et al., 1997). These values are significantly larger than that of quick clay. Usually, snow falls from October to April, and the maximum snow depth is approximately 1 m. From 1988 to 2002, the mean annual temperature in Trondheim (Risvollan, 85 m [a.s.l.]) was 4.9 °C and the mean annual precipitation was 1003.5 mm/year (Thorolfsson, 2007).

2. Research site 2.1. Quick clay Quick clay is found mainly in Scandinavia and eastern Canada. This type of sediment is highly sensitive clay that was originally deposited in marine or brackish waters. The clay was subsequently elevated above sea level during the postglacial isostatic uplift. After its emergence, quick clay lost almost all interparticle forces because of leaching of pore-water salt from marine sediment by groundwater flow (Rosenqvist, 1953). Therefore, the skeletal structure can easily collapse as a result of changes in stress caused by landform alteration or construction activity; the quick clay then liquefies, which can cause serious clay slides (Geertsema et al., 2006; L'Heureux et al., 2012). Torrence (1996) defined quick clay as soil exhibiting sensitivity > 30 and remolded shear strength of < 0.5 kPa based on the falling-cone test (Rajasekaran and Narasimha Rao, 2004). Karlsrud et al. (1984) conducted an undrained triaxial compression test, and described the extremely fragile behavior of quick clay as follows: the peak resistance was reached at an axial strain of only 0.3% and the shear resistance was reduced to 50% of the peak strength with a strain of 3%. In contrast, Karlsrud et al. (1984) also reported for quick clay that the loss in shear resistance after the peak was reached was entirely related to an increase in pore pressure.

3. Observations In November 1997, we initiated continuous monitoring at the Roesgrenda landslide research site at 15-min intervals using an automated data collection system (Okamoto et al., 2004). Our targets were 131

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Table 1 Record of landslides at the Roesgrenda research site between 1995 and 2001. No.

Date (m/d/y)

Volume [m3]

Remarks

1 2 3 4 5

03/05/1995 10/18/1995 07/05/1996 02/?/1998 03/30/1998

20000 10000 10000 – –

The largest landslide observed since 1995

6 7 8 9

05/11/1998 07/? / 1998 08/17/1998 01/20/2000

1000 150 3000 –

10 11 12 13 14

01/24/2000 02/10/2000 03/02/2000 03/04/2000 04/20/2001

– – 2000 300 –

Minor failure (< 20 m3) Minor failure (< 20 m3) Destroyed St1 Occurred at beginning of July Minor failure destroyed soil temp. sensor of St1–St4 Minor failure Minor failure

Minor failure

Fig. 3. A view of the scarp of the Roesgrenda landslide. The location at which the photograph was taken is indicated by the star in Fig. 2.

the hydrological and meteorological factors that influence landslide activity. Monitoring was conducted at the crown and the upper flat area located above the scarp, which was regarded as a potential landslide mass in the near future. An overview of the sensors is provided in Fig. 2, Fig. 4, and Table 3. Meteorological factors such as precipitation, meltwater, air temperature, and soil temperature were automatically monitored from 1997 onward. A tipping-bucket rainfall gauge (R), a meltwater gauge (Mw), and an air temperature sensor (At) were installed on an upper flat area. Thermostat heaters (20 °C) were attached to the rainfall gauge to prevent freezing. Soil temperature sensors (St1 to St6) were installed at four depths in the main scarp and at two depths in the upper flat area. Snow conditions were not automatically monitored, but were checked daily at a public meteorological station in Skjӕkerfossen (125 m a.s.l.), located 8 km from the Roesgrenda research site. In addition, snow depth and snow load were manually measured 14 and 5 times, respectively, in the upper flat area during the snow-covered periods from 1998 to 2000. Pore water pressure was automatically monitored at both the crown and the upper flat area using five stainless-steel piezometers (Oyo Corp., Model-4585) that had a measuring range of 350 kPa with an accuracy of 0.25%. The precision of the piezometer was estimated to be < 0.1 kPa from the short time-series data that were observed every 15 min. The output signal from the sensor had an electric current of 4–20 mA, which was free from line resistance of signal cables. The sensor was almost unaffected by changes in groundwater temperature because the temperature sensitivity of the transducer was as small as

Fig. 2. Location and topographical map of the Roesgrenda landslide research site (top), and arrangement of the sensors (bottom). Shading represents landslides. Landslide Nos. 6, 8, and 12 correspond to the numbers in Table 1.

132

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Fig. 4. Schematic geological cross-section of the Roesgrenda research site and locations of the hydrological sensors. Landslide Nos.1, 2, 3, and 12 correspond to the numbers in Table 1. Table 2 Physical properties of soil at the Roesgrenda site. Layer

River deposits Quick clay

Water content1)

Hydraulic conductivity1),2)

Porosity3)

Internal friction angle3)

Sensitivity1)

Wet unit weight3)

w [%]

Ksat [m/s]

n [%]

φ’ [°]

St

γt [kN/m3]

28.2 25.1

N/A 10−9–10−12

41–44 42

28 26

2–26 73–186

19.7 20.2

1) Larsen (2002), 2) Long (2005), 3) Larsen et al. (1999). Table 3 Instruments installed at the Roesgrenda research site.

⁎ ⁎

Symbol

Sensor type

Location

Depth [m]

Layer

Unit

Accuracy

Start of monitoring

P1 P2 P3 P4 P5 Wc1 Wc2 Wh1 Wh2 R Mw At St1–St4 St5–St6

Piezometer Piezometer Piezometer Piezometer Piezometer TDR sensor TDR sensor Tensiometer Tensiometer Rainfall gauge Meltwater gauge Air temp. sensor Soil temp. sensors Soil temp. sensors

Flat area Flat area Crown Flat area Flat area Crown Crown Crown Crown Flat area Flat area Crown Scarp Crown

22.0 18.8 10.0 9.5 9.0 0.4 0.4 1.4 1.0 – – – *1 *2

Quick clay Quick clay Bottom of river deposits Bottom of river deposits Bottom of river deposits Topsoil Topsoil Top of river deposits Top of river deposits Above the ground On the ground Above the ground Top soil, river deposits Top soil, river deposits

kPa kPa kPa kPa kPa % % mmH2O mmH2O mm mm °C °C °C

± 0.875 kPa ± 0.875 kPa ± 0.875 kPa ± 0.875 kPa ± 0.875 kPa ± 2% ± 2% ± 2% ± 2% ± 0.5 mm ± 0.5 mm

Nov. 1997 Nov. 1997 Sep. 1999 Sep. 1999 Sep. 1999 Sep. 1999 Sep. 1999 Sep. 1999 Sep. 1999 Nov. 1997 Nov. 1997 Nov. 1997 Nov. 1997 Sep. 1999

1) Four sensors were installed, one each at depths of 0.05, 0.50, 1.00 and 1.50 m. 2) Two sensors were installed, one each at depths of 0.05 and 1.00 m.

4. Results

0.015% per °C. The transducer and the porous cup of the sensor were filled with deaerated water using an acrylic cover to maintain the saturated condition until installation. When the sensor was installed, the acrylic cover was taken off in the water-saturated borehole. Two piezometers (P1 and P2) were installed almost in the middle of the quick clay layer and monitored from 1997; the other three (P3, P4, and P5) were set near the bottom of the overlying river deposits and monitored from 1999. In addition, two tensiometers (Wh1 and Wh2) and two timedomain reflectometry (TDR) soil volumetric water content sensors (Wc1 and Wc2) were installed into the surface layer, including the topsoil, and monitored from 1999.

4.1. Overview The results of monitoring of meteorological factors from 1997 to 2000 at the Roesgrenda site are plotted in Fig. 5. In this study, we regard the amount of water measured by the meltwater gauge as the rainfall and/or meltwater supplied to the ground throughout the seasons; this is referred to as Meltwater and/or Rain (MR; Matsuura, 2000). There are some missing MR data from January to April in both 1998 and 1999 as a result of freezing of the drainage conduit connected to the tipping bucket. This problem was mitigated in 2000. Annual MR including missing data in 1998, 1999, and 2000 was 798, 837, and 1003 mm, respectively.

133

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Fig. 6. Relationship between the snow depths at the Roesgrenda landslide and the Skjækerfossen meteorological station from 1998 to 2000 (Okamoto et al., 2004) Table 5 Estimates of snow conditions at the Roesgrenda research site.

Fig. 5. Observations of hydrological and snow conditions over a three-year period at the Roesgrenda landslide research site.

1997–98

1998–99

1999–2000

First snow fall (m/d/y) First continuous snow cover (m/d/y) Last snow cover (m/d/y) Duration of snow cover [days] Maximum snow depth [m]

10/20/1997 01/17/1998 04/27/1998 101 0.68

11/07/1998 12/20/1998 04/22/1999 124 0.53

11/28/1999 11/28/1999 05/02/2000 157 1.02

16, 2000. Fig. 7 shows the air temperature and the time-series depth profiles of soil temperature drawn by interpolation of soil temperatures for St1 to St6. Mean daily air temperature during the cold season (from December to March) in 1997–98, 1998–99, and 1999–2000 was −1.7, −2.8, and − 2.5 °C, respectively; the minimum temperature was −20.9 °C, recorded on January 27, 1999. The ground surface at the scarp remained frozen every winter (December to March) to a depth of > 0.5 m with fluctuations affected by the air temperature. In contrast, the ground surface in the upper flat area was variable in its freezing; it was frozen during the cold season of 1998–99, and was not frozen during the cold season of 1999–2000. Chronological changes in hydrological factors such as pressure head [mmH2O], volumetric water content [%], and pore water pressure [kPa] are also plotted in Fig. 5. The pressure heads and the volumetric

Manual snow depth measurement results for snow depth and snow load are provided in Table 4. The deepest snow depth was 0.97 m on March 23, 2000. Okamoto et al. (2004) found a high correlation (R2 = 0.937) between snow depth at the Roesgrenda research site (Sdr) and at the public meteorological station in Skjӕkerfossen (Sds), as shown in Fig. 6, and calculated the estimated daily snow depth at the research site (Sdest) using the following equation: (1)

Sd est = 0.644 Sd s + 3.96

Snow-covered year

The chronological changes in the estimated snow depth at the research site are plotted in Fig. 5 and summarized in Table 5. The first date of continuous snow cover varied from late November to midJanuary, although the snow cover disappeared at almost the same time in late April. The estimated maximum snow depth was 1.02 m on March

Table 4 Snow factors measured manually at the Roesgrenda research site and snow depth at the Skjӕkerfossen station. Roesgrenda research site Date (m/d/y)

Snow depth [m]

03/10/1998 03/19/1998 03/30/1998 12/16/1998 01/26/1999 01/31/1999 02/20/1999 03/26/1999 04/12/1999 12/16/1999 01/09/2000 01/25/2000 02/11/2000 03/23/2000

0.33 0.53 0.26 0.08 0.25 0.25 0.54 0.51 0.13 0.25 0.15 0.30 0.43 0.97

Skjӕkerfossen station Snow load [kN/m2]

0.4 0.5 0.9 1.7 3.2

134

Snow density [103kg/m3]

Snow depth [m]

0.16 0.33 0.30 0.38 0.33

0.54 0.85 0.52 0.00 0.28 0.30 0.76 0.55 0.20 0.36 0.20 0.40 0.51 1.40

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Fig. 7. Time series depth profiles of soil temperature and air temperature at the Roesgrenda research site. Profiles are based on interpolations of the soil temperature recorded by St1 to St6.

Fig. 8. Fluctuations in pore water pressure during snow-covered periods. a: snow depth and pore water pressure increase; b: snow depth and pore water pressure at maximum; c: pore water pressure and snow depth decrease (snowmelt period); d: snow cover disappears and pore water pressure stabilizes.

135

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

showed a peak (“b”). In the meltwater period, the pore water pressure rapidly dropped as the snow depth decreased (“c”). Finally, when the snow cover disappeared, the pore water pressure stopped rapidly dropping (“d”). Only P3 in 1999–2000 showed a different pattern in that the peak appeared during a different period. 4.2. Generation of excess pore water pressure by snow load In general, a major factor influencing fluctuations in pore water pressure is vertical infiltration of rain and/or meltwater from the surface into the ground. At the Roesgrenda research site, the topsoil in the upper flat area was frozen to a depth of approximately 0.2 m during the cold season of 1998–99 (from December to March), as illustrated in Fig. 7. Deeply frozen soil reduces water infiltration into the subsurface layer, resulting in increased surface runoff and a decrease in groundwater recharge (Bayard et al., 2005; Stadler et al., 1996). The critical frost depth for impedance of infiltration is estimated to be between 0.2 and 0.4 m based on field experiments and numerical experiments in the cold region of Japan (Iwata et al., 2010). According to these previous studies, fluctuations in pore water pressure at the site should be small during the cold season of 1998–99; however, in fact the pore water pressures tended to increase. In landslides with low-permeability layers, slow groundwater flow can cause a long-term response of the pore water pressure to supplied rainfall and/or meltwater. Iverson (2000) measured pore water pressures of the shear zone of a landslide with low saturated hydraulic conductivity (Ksat = 5 × 10−8 [m/s]), and demonstrated that pore water pressures could respond significantly to long-term seasonal rainfall but negligibly to rainfall of less than a few months. Schulz et al. (2009) demonstrated that a gradual increase in pore water pressure continued into July, although snow melt occurred mainly in March and April. Asano et al., (2008) also obtained field monitoring results for a deep-seated landslide that indicated that the peak in pore water pressure was delayed by about one month after the peak of snow melting. At the Roesgrenda research site, however, the rainfall pattern is irregular during the non-snow season (approximately May to November), which means that there is no seasonal antecedent rainfall corresponding to the pore water pressure increase during the snow-covered period.

Fig. 9. Graphic explanation of the equations used to calculate pore water pressure increase during the snow-covered period.

water content in the topsoil fluctuated widely in response to MR. In contrast, the pore water pressures at the quick clay layer and the bottom of the river deposits were monotonous, with fluctuations of < 5 kPa. Fluctuations in pore water pressure in the snow-covered period of 1997–98, 1998–99, and 1999–2000 are plotted individually in Fig. 8. The pore water pressures showed consistent long-term variations that increased in January to March and decreased in April to May in both the quick clay and the bottom of the river deposits. The variations of pore water pressures in the snow-covered period did not exhibit any correspondence with MR. For example, in the snow-covered period of 1999–2000, the pore water pressure kept increasing from December to March, although the mean daily MR in the same period was lower by 1.9 mm/d than the value of 3.2 mm/d in the period of the previous August to November. In the snowmelt period (the following April), pore water pressure rapidly decreased despite a large mean daily MR of 8.0 mm/d. The fluctuation patterns of pore water pressure more closely resembled the estimated snow depth patterns. The clearly harmonized periods of both patterns are color-coded in Fig. 8. When the snow depth increased, the pore water pressure also increased (“a” in Fig. 8). When the snow depth remained at the maximum, the pore water pressure

Fig. 10. Relationships between snow load and pore water pressure increase (Δut) at the five piezometers during the snow-covered period of 1999-2000. 136

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

snow accumulation in the Roesgrenda research site, we attributed almost all of the increase in pore water pressure to undrained snow loading on the low-permeability quick clay layer with river sediments, and consequent generation of excess pore water pressure. In addition, the deeply frozen topsoil at the scarp prevents groundwater discharge, which contribute to the undrained conditions. This explanation is strongly supported by two facts: the saturated hydraulic conductivity of the quick clay was extremely low, and the response of the pore water pressure to rain and meltwater was insensitive. We also need to consider shearing of the landslide mass because further excess pore water pressure would be generated followed by negative dilatancy during undrained shearing because quick clay is a normally consolidated clay. The shear stress is generated by the snow loading acting on the ground. Its value is dependent on geometry, such as the ground surface inclination and the surrounding terrain. At the Roesgrenda landslide, however, the value of shear stress would actually be negligibly small because the observed upper area was almost flat and at a sufficient distance from the surrounding crests. Observational results indicating a similar increase in pore water pressure during the snow-covered period were obtained in previous studies (Maruyama, 1993; Matsuura, 2000). 4.3. Relationship between snow accumulation and pore water pressure To understand the effects of the snow load on pore water pressure more precisely, we examined the relationship between snow accumulation and pore water pressure. Here, snow accumulation is explained by two indexes: snow load and estimated snow depth. The snow load, which was manually measured five times each winter, indicates the true value of snow load, and can be used for quantitative analyses. The estimated snow depth indicates continuous changes in the approximate snow load, which is useful to understand the long-term trends of snow loading. In this section, both indexes for snow accumulation are used for relational analyses. Firstly, we considered the relationship between the snow load and pore water pressure increase. Here, pore water pressure increase (Δut) denotes the difference in the pore water pressure value (ut − u0) measured between any date (t) and one day before the snow-cover period (t0) (Fig. 9). Fig. 10 shows the relationship between snow load (σt) and pore water pressure increase (Δut) for the quick clay and the river deposit. The relationship is linear in the quick clay layer (P1 and P2). The regression coefficient of the correlation yields the ratio of pore water pressure increase in response to the snow load, which is defined as rusnow. Theoretically, rusnow has a minimum value of zero in completely drained conditions and a maximum value of one under fully undrained conditions, ignoring non-uniformity of the topology and snow distribution. The value of rusnow is 0.49–0.53 in the quick clay layer with very high correlation coefficients (R2 = 0.995–0.996), which means that approximately half of the snow load contributes to the excess pore water pressure. The small intercept value (0.18–0.25 kPa) obtained in the regression expression probably represents pore water increase caused by other factors such as groundwater flow. These relationships strongly support our assumption that excess pore water pressure is generated in the quick clay layer with extremely low permeability when the clay is compressed by the snow load. In contrast, the relationship is logarithmic in the overlying river deposits (R2 = 0.840–0.984), and rusnow cannot be determined. The logarithmic relationship results from dwindling of the pore water pressure increase under the large snow load. The snow load was relatively small from December to January, and increased from February to March (Table 4). Considering the time axis, part of the excess pore water pressure dissipated with time in the relatively high-permeability river deposits. Next, we considered the long-term relationship between estimated snow depth and pore water pressure for three winter seasons (Fig. 11). Each relationship was color-coded, with the snow accumulation period in white and the snow melting period in black, to understand the trajectory path easily. The boundary date was conveniently determined by

Fig. 11. Relationships between snow depths and pore water pressures in the quick clay layer during the snow-covered period for three successive years.

Considering these soil and meteorological conditions, the increase in pore water pressures in the Roesgrenda research site is caused neither by vertical infiltration of meltwater nor by slow groundwater lateral flow. Hutchinson and Bhandari (1971) observed generation of excess pore water pressures by pushing piezometers into mudslides to simulate undrained loading on the sliding mass. Generation of excess pore water pressure in a saturated layer by undrained loading were experimentally reproduced by triaxial compression tests (Sassa et al., 1996). Considering the harmonic fluctuation between pore water pressure and 137

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

Fig. 12. Location and topographical map (top) and longitudinal section (bottom) of the Busuno landslide research site.

topsoil is 5 × 10−5 m/s, from a laboratory permeability test (Osawa et al., 2017). These values are significantly higher than that of the quick clay layer (Ksat = 10−9 to 10−12 [m/s]). Fig. 13 shows the response of pore water pressure in the Busuno landslide during the snow-covered period of 1993–94 (Matsuura, 2000). The pore water pressure in the Busuno landslide increased as snow load increased in a similar manner to the Roesgrenda landslide. However, there were two significant differences in the response characteristics of pore water pressure to snow load. One difference was the ratio of pore water pressure increase in response to the snow load (rusnow). In the Busuno landslide, pore water pressures increased by approximately 2 kPa for 13.1 kN/m2 of snow load. The value of rusnow in the Busuno landslide was thought to be approximately 0.15, significantly smaller than rusnow (0.49–0.53) in the quick clay landslide. The other difference occurred during the snowmelt period: as the snow depth decreased in the snowmelt period, the pore water pressure further increased and reached its maximum value in the Busuno landslide, whereas it decreased in the quick clay landslide. The differences are illustrated diagrammatically in Fig. 14. The sensitivity of the pore water pressure response to snow load is thought to depend on the permeability and compressibility of the landslide mass. The quick clay layer, which is composed of leached, normally consolidated clay, has high compressibility. Bjerrum (1967) conducted a leaching test on marine clay, and showed an increase in compressibility after leaching. Torrance (1974) also described how leaching of normally consolidated marine clay, as for quick clay, induces spontaneous consolidation. When a maximum of about 1 m of snow accumulates on the Roesgrenda landslide site in winter, a major compressive force acts on the landslide mass as a result of the snow load. However, because the permeability of the quick clay layer is extremely low, the pore water cannot easily drain. As a result, the pore

the last date of increase in snow depth, and was March 26 (1997–98), March 24 (1998–99), and April 14 (1999–2000). The pore water pressures were positively correlated with the snow depth, and almost all showed small differences in value (< 0.5 kPa) between the beginning and the end of the snow-covered period. This result suggested that the increase in the pore water pressure was affected only by the snow load during the snow-covered period at the Roesgrenda landslide.

5. Discussion 5.1. Characteristics of pore water pressure response to snow load We detected a characteristic response of pore water pressure increase to undrained snow loading on a low-permeability layer at the Roesgrenda landslide. As mentioned earlier, there have been several other observations of pore water pressure increase response to snow accumulation (Maruyama, 1993; Matsuura, 2000). We compared the response characteristics between the Roesgrenda landslide and the Busuno landslide, Japan, which has relatively higher permeability, and discussed the response characteristics of pore water pressure to snow load. The Busuno landslide is a reactivated landslide in a Neogene deposit in Japan (latitude 37.0°N, longitude 138.5°E; Fig. 12), and is located in a heavy snow area with a snow depth of about 3–5 m and a maximum snow load of about 11–16 kN/m2 (Matsuura et al., 2005). Landslide movement is active from late fall to early winter, and the cumulative displacement is > 1 m every year (Matsuura et al., 2003). The landslide mass consists of clay with rubble and highly weathered mudstone, and the specific sliding surface occurs at a depth of around 5 m. The Ksat of the whole landslide mass is 2.7–3.9 × 10−6 m/s, measured by a simplified pumping test (Kanto Regional Forest Office, 2003); that of the 138

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

quick clay layer even if it has extremely low permeability. These factors could inhibit the generation of high excess pore water pressure, and could result in the excess pore water pressure with rusnow of 0.49–0.53. Additionally, the remaining stress was transmitted to the effective stress. When the snowmelt period begins, the snow load decreases, and the pore water pressure declines accordingly. When the meltwater reaches the low-permeability clay layer, it does not infiltrate but instead flows down near the surface, so it has no effect on fluctuations in pore water pressure. After all the snow has melted, the pore water pressure stops rapidly dropping. In contrast, the moving body of the Busuno landslide has many cracks as hydraulic paths as a result of its large displacement, and its permeability is relatively high. Thus, more pore water is discharged from the Busuno landslide mass than from the quick clay landslide mass, causing lower excess pore water pressure with rusnow of 0.15. When the snowmelt period begins, the decrease in snow cover should cause a slight decrease in the excess pore water pressure. However, during the snowmelt period, large amounts of meltwater can infiltrate the ground from the water paths, causing a large increase in pore water pressure. This is the reason why the pore water pressure reaches its maximum value during the snowmelt period. Comparison of these results demonstrates that the permeability of the landslide mass controls the fluctuations in pore water pressure during the snow-covered period, including generation of excess pore water pressure. 5.2. Possible influence of snow load and excess pore pressure on the landslide In a snowy region, some landslide activities are greatly influenced by both pore water pressure and by snow load (e.g., Maruyama, 1993; Matsuura et al., 2003). A snow load on a landslide mass provides both a total stress and a shear stress on the sliding surface. Matsuura et al. (2017) investigated slope stability under snow loading using limit equilibrium analysis without considering pore water pressure, and indicated that the direction of the snow load effect on the slope stability was determined by the balance between the mean inclination of the sliding surface (α) and the internal friction angle (φ’). Specifically, the snow load stabilized the slope when φ’ > α, and antithetically destabilized the slope when α > φ’. In the Roesgrenda landslide, the angle of the potential sliding surface (α) within the quick clay was estimated to be 27–37° based on the inclinations of the past scarps from 1995 to 1998 (Larsen et al., 1999), which is larger than the internal friction angle (φ’) of 26° obtained from the peak shear strength of samples by a laboratory test (Table 2). This relation of the angle (α > φ’) means that the snow load has a negative effect on slope stability at the Roesgrenda landslide. Furthermore, the excess pore water pressure in response to snow loading causes a drop in shear strength as a result of the decrease in the effective normal stress acting on the sliding surface. The Roesgrenda landslide, which has a high ratio of rusnow, is considered to be influenced by the snow load, which acts as a destabilizing factor in theory.

Fig. 13. Pore water pressure fluctuations during the snow-covered period of the Busuno landslide in a Neogene deposit, Japan (Matsuura, 2000).

water is subjected to much of the snow load, which leads to generation of high excess pore water pressure with rusnow of 0.49–0.53. Here, if all layers in the Roesgrenda landslide mass were fully saturated and undrained, rusnow would be theoretically 1.0 and the effective stress would not be changed because the additional total stress applied to the soil by the snow load would be entirely transmitted to the pore water pressure. However, the observed pore water pressure indicated rusnow of 0.49–0.53. The reasons for the generation of approximately half rusnow are considered to be as follows. Five piezometers (P1 to P5) and two tensiometers (Wh1 and Wh2) constantly observed their piezometric head below the ground surface (see Fig. 5), which indicated the existence of unsaturated zones in the landslide mass and surface layer. In those conditions, the additional total stress is not entirely transmitted to the pore water pressure. In addition, long-term, slow snow loading over a few months allows the groundwater to be partly discharged from the

Fig. 14. Patterns of the fluctuations in pore water pressure response to snow cover of landslides with different permeability values. 139

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

differences in pore water pressure (< 0.5 kPa) between the beginning and the end of the snow-covered period, which suggested that the pore water pressure during the snow-covered period was affected only by the snow load. (4) The response characteristics of pore water pressure to snow accumulation were compared with the Busuno landslide in Japan, which contains relatively more permeable deposits (Ksat = 2.7–3.9 × 10−6 [m/s]). Estimated rusnow of the Busuno landslide was approximately 0.15, significantly smaller than the value of 0.49–0.53 in the Roesgrenda landslide. In the snowmelt period, the pore water pressure in the Busuno landslide further increased and reached a maximum because of the snowmelt infiltration, whereas it decreased in the quick clay landslide. These differences in the response characteristics were considered to reflect differences in the permeability and compressibility of the landslide mass. (5) In theory, the generated excess pore water pressure based on the high rusnow could also decrease the effective normal stress on the sliding surface and decrease the slope stability. However, in fact it could have little effect because the value of monitored excess pore water pressure was as small as one two-hundredth of the original overburden pressure on the potential sliding surface. Quantitative assessment of the combined effect on slope stability of the snow load and the generated excess pore water pressure remains a key issue to be clarified.

We monitored the surface displacement processes of a landslide (No. 12 in Table 1) for 93 days from its initial displacement on Nov. 30, 1999, to the final stage of slope failure on Mar 2, 2000 (Okamoto et al., 2004). Our monitoring showed long-term increases in snow load and pore water pressures on/in the quick clay layer before the slope failure (No. 12) occurred (see Fig. 5). However, there was no obvious connection of timing between the pore water pressure increase and the slope failure. Here, considering the stress, the increase in the excess pore water pressure from the snow load was < 2 kPa during each snowcovered period. In contrast, the value of overburden pressure on the quick clay layer at 20 m depth is estimated to be approximately 400 kPa from the wet unit weight of the landslide mass (see Table 2), which is 200 times the value of the excess pore water pressure. Because of the great difference in stress, we regard the excess pore water pressure during the snow-covered period as having little contribution to slope stability at the Roesgrenda landslide. However, the contribution of the snow load and related pore water pressure increase to landslide activity is open to dispute, because the estimated value of the snow load on the Roesgrenda landslide is < 5 kPa, which is much smaller than the value of overburden pressure on the quick clay layer that is approximately 20 m deep. Quantitative assessment of the effects of snow load and excess pore water pressure requires detailed information on the distribution of the potential sliding surface and of the snow load, for which further research is needed. 6. Conclusions

The unexpected increase in pore water pressure caused by snow loading suggests new insights regarding landslide activity of low-permeability deposits in snow-covered regions.

The aim of this study was to examine the response characteristics of pore water pressure to snow loading on a landslide mass. Field-based monitoring of hydraulic and meteorological factors was conducted at the Roesgrenda landslide research site in Mid-Norway, where an extremely low-permeability “quick clay” (saturated hydraulic conductivity: Ksat = 10−9 to 10−12 [m/s]) was deposited. The observational results for three years demonstrated an obvious increase in pore water pressure in response to snow loading. The following conclusions can be drawn:

Acknowledgements We thank Professor Emeritus Lars Olav Grande of the Department of Geotechnical Engineering, Norwegian University of Science and Technology, and the members of the Norwegian Water Resources and Energy Administration for their support in installing the sensors at the research site. We also thank Dr. Hiromu Daimaru and Dr. Yasuhiko Okada of the Forestry and Forest Products Research Institute for their helpful suggestions. This study was supported by the Special Coordination Funds for Promoting of Science and Technology program [grant number 63], the Science and Technology Agency of Japan. We thank Lucy Muir, PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript.

(1) The pore water pressure showed cyclic variation of < 5 kPa, increasing during the snow-covered period in the quick clay and the overlying river deposits. Time-series changes in the pore water pressure corresponded closely to changes in snow accumulation rather than MR (Meltwater and/or Rain). During mid-winter, the topsoil froze and prevented water infiltration and discharge, and no seasonal heavy rainfall occurred before the snow-covered period. Considering the surface conditions and the observed results, increases in pore water pressure during the snow-covered period should be excess pore water pressure generated by undrained snow loading on the low-permeability quick clay and the layer of river deposits. (2) The increase in pore water pressure in the quick clay exhibited a positive linear relation with high correlation to the measured snow load, which strongly supported the explanation of generation of excess pore water pressure by undrained snow loading. The ratio of the increase in the pore water pressure to the snow load (rusnow) was obtained from the regression coefficient of the linear relation. The value was 0.49–0.53 in quick clay, meaning that about half of the snow load contributed to the excess pore water pressure. Meanwhile, the amount of increase in the pore water pressure in the overlying river deposits exhibited a logarithmic relation with the snow load. The reason for the logarithmic relation was considered to be that part of the excess pore water pressure gradually dissipated with time from the relatively high-permeability river deposits. (3) The time-series change in pore water pressure also exhibited a longterm correlation with the change in estimated daily snow depth, which approximately indicated snow load. There were small

References Asano, S., Matsuura, S., Okamoto, T., Kurokawa, U., 2008. Estimate of groundwater distribution from drainage and deformation of a slip zone in a large-scale landslide area. In: 12th International Conference of International Association for Computer Methods and Advances in Geomechanics (IACMAG), pp. 4525–4532. Bayard, D., Stähli, M., Parriaux, A., Flühler, H., 2005. The influence of seasonally frozen soil on the snowmelt runoff at two alpine sites in southern Switzerland. J. Hydrol. 309, 66–84. http://dx.doi.org/10.1016/j.jhydrol.2004.11.012. Bjerrum, L., 1967. Engineering geology of Norwegian normally-consolidated marine clays as related to settlements of buildings. Géotechnique 17, 83–118. http://dx.doi.org/ 10.1680/geot.1967.17.2.83. Chigira, M., Chiba, T., 1998. The landslide, steam explosions, and debris avalanche sequence of May 1997 at Sumikawa spa, Akita prefecture. Japan. J. Nat. disaster Sci. 20 (1), 21–31. Coe, J.A., Ellis, W.L., Godt, J.W., Savage, W.Z., Savage, J.E., Michael, J.A., Kibler, J.D., Powers, P.S., Lidke, D.J., Debray, S., 2003. Seasonal movement of the Slumgullion landslide determined from global positioning system surveys and field instrumentation, July 1998-March 2002. Eng. Geol. 68, 67–101. http://dx.doi.org/10.1016/ S0013-7952(02)00199-0. Geertsema, M., Cruden, D.M., Schwab, J.W., 2006. A large rapid landslide in sensitive glaciomarine sediments at Mink Creek, northwestern British Columbia. Canada. Eng. Geol. 83, 36–63. http://dx.doi.org/10.1016/j.enggeo.2005.06.036. Gokceoglu, C., Sonmez, H., Nefeslioglu, H.A., Duman, T.Y., Can, T., 2005. The 17 March 2005 Kuzulu landslide (Sivas, Turkey) and landslide-susceptibility map of its near vicinity. Eng. Geol. 81, 65–83. http://dx.doi.org/10.1016/j.enggeo.2005.07.011. Hayashi, Y., Kosugi, K., Mizuyama, T., 2006. Changes in pore size distribution and

140

Engineering Geology 242 (2018) 130–141

T. Okamoto et al.

s12665-017-6693-7. Nakamura, H., Shiraishi, Y., 1973. On influence of snow upon landslides. J. Japan Landslide Soc. 10 (1), 6–16. (in Japanese with English abstract). https://doi.org/10. 3313/jls1964.10.6. Naudet, V., Lazzari, M., Perrone, A., Loperte, A., Piscitelli, S., Lapenna, V., 2008. Integrated geophysical and geomorphological approach to investigate the snowmelttriggered landslide of Bosco Piccolo village (Basilicata, southern Italy). Eng. Geol. 98, 156–167. http://dx.doi.org/10.1016/j.enggeo.2008.02.008. Noguchi, S., Rahim Nik, A., Kasran, B., Tani, M., Sammori, T., Morisada, K., 1997. Soil physical properties and preferential flow pathways in tropical rain forest, Bukit Tarek, Peninsular Malaysia. J. For. Res. 2, 115–120. http://dx.doi.org/10.1007/ BF02348479. Okamoto, T., Larsen, J.O., Matsuura, S., Asano, S., Takeuchi, Y., Grande, L., 2004. Displacement properties of landslide masses at the initiation of failure in quick clay deposits and the effects of meteorological and hydrological factors. Eng. Geol. 72, 233–251. http://dx.doi.org/10.1016/j.enggeo.2003.09.004. Okamoto, T., Matsuura, S., Asano, S., 2008. Deformation mechanism of a shallow landslide in a snow-covered area. J. Japan Landslide Soc. 44, 358–368. (in Japanese with English abstract). https://doi.org/10.3313/jls.44.358. Osawa, H., Matsuura, S., Matsushi, Y., Okamoto, T., 2017. Seasonal change in permeability of surface soils on a slow-moving landslide in a heavy snow region. Eng. Geol. 221, 1–9. http://dx.doi.org/10.1016/j.enggeo.2017.02.019. Rajasekaran, G., Narasimha Rao, S., 2004. Falling cone method to measure the strength of marine clays. Ocean Eng. 31, 1915–1927. http://dx.doi.org/10.1016/j.oceaneng. 2000.12.001. Rosenqvist, I.T., 1953. Considerations on the sensitivity of Norwegian quick-clays. Géotechnique 3, 195–200. http://dx.doi.org/10.1680/geot.1953.3.5.195. Sassa, K., Fukuoka, H., Scarascia-Mugnozza, G., Evans, S., 1996. Earthquake-inducedlandslides : distribution, motion and mechanisms. Soils Found. 36, 53–64. http://dx. doi.org/10.3208/sandf.36.special_53. Schulz, W.H., McKenna, J.P., Kibler, J.D., Biavati, G., 2009. Relations between hydrology and velocity of a continuously moving landslide-evidence of pore-pressure feedback regulating landslide motion? Landslides 6, 181–190. http://dx.doi.org/10.1007/ s10346-009-0157-4. Stadler, D., Wunderli, H., Auckenthaler, A., Flühler, H., Bründl, M., 1996. Measurement of frost-induced snowmelt runoff in a Forest soil. Hydrol. Process. 10, 1293–1304. http://dx.doi.org/10.1002/(SICI)1099-1085(199610)10:10<1293::AIDHYP461>3.0.CO;2-I. Thorolfsson, S.T., 2007. Urban hydrological data collection in cold climate. Experiences at Risvollan, Trondheim, Norway. In: Proceedings of International Symposium on New Directions in Urban Water Management, pp. 1–8 (Paris). Torrance, J.K., 1974. A laboratory investigation of the effect of leaching on the compressibility and shear strength of Norwegian marine clays. Géotechnique 24, 155–173. http://dx.doi.org/10.1680/geot.1974.24.2.155. Torrence, J.K., 1996. On the development of high sensitivity: Mineralogical requirements and constraints. In: Proceedings of the 7th International Symposium on Landslides. Trondheim, Norway, pp. 491–496. Wieczorek, G.F., 1996. Landslides: investigation and mitigation. Chapter 4-Landslide triggering mechanisms. Transp. Res. Board Spec. Rep. 247, 76–90.

hydraulic properties of forest soil resulting from structural development. J. Hydrol. 331, 85–102. http://dx.doi.org/10.1016/j.jhydrol.2006.05.003. Hutchinson, J.N., Bhandari, R.K., 1971. Undrained loading, a fundamental mechanism of mudflows and other mass movements. Géotechnique 21, 353–358. http://dx.doi.org/ 10.1680/geot.1971.21.4.353. Iverson, R.M., 2000. Landslide triggering by rain infiltration. Water Resour. Res. 36, 1897–1910. http://dx.doi.org/10.1029/2000WR900090. Iwata, Y., Hayashi, M., Suzuki, S., Hirota, T., Hasegawa, S., 2010. Effects of snow cover on soil freezing, water movement, and snowmelt infiltration: a paired plot experiment. Water Resour. Res. 46, 1–11. http://dx.doi.org/10.1029/2009WR008070. Kanto Regional Forest Office, 2003. Simplified pumping test. In: Research Report on the Landslide Mechanism and Planning for the Countermeasure Works at the Busuno (Touge) District, pp. 34–42 (in Japanese). Karlsrud, K., Aas, G., Gregersen, O., 1984. Can we predict landslide hazards in soft sensitive clays? Summary of Norwegian practice and experiences. In: Proceedings of the 4th International Symposium on Landslides, pp. 107–130 (Toronto). Kristoffersen, R., 1999. An investigation of the stability of a slope in quick clay at Roesgrenda. Verdal, Norway 1–23 (in Norwegian). Larsen, J.O., 2002. Some Aspects of Physical Weather Related Slope Processes. Ph. D thesis. Norwegian University of Science and Technology. Larsen, J.O., Grande, L., Matsuura, S., Okamoto, T., Asano, S., Park, S.G., 1999. Slide activity in quick clay related to pore water pressure and weather parameters. In: Griffiths, J.S., Stokes, M.R., Thomas, R.G. (Eds.), Landslides: Proceedings of the 9th International Conference and Field Trip on Landslides. A. A. Balkema, Rotterdam, Bristol, pp. 81–88. L'Heureux, J.-S., Eilertsen, R.S., Glimsdal, S., Issler, D., Solberg, I.-L., Harbitz, C.B., 2012. The 1978 Quick Clay Landslide at Rissa, Mid Norway: Subaqueous Morphology and Tsunami Simulations, in: Proceeding of the 5th International Symposium on Submarine Mass Movements and their Consequences. Springer Netherlands, Dordrecht, pp. 507–516. http://dx.doi.org/10.1007/978-94-007-2162-3_45. Long, M., 2005. Some analyses of the Roesgrenda quick clay slides. In: Senneset, K., Flaate, K., Larsen, J.O. (Eds.), Landslides and Avalanches: Proceedings of 11th International Conference and Field Trip on Landslides. Taylor & Francis, Norway, pp. 233–239. Maruyama, K., 1993. Change type of pore water pressure in landslide area. J. Japan Landslide Soc. 30, 27–35. (in Japanese with English abstract). https://doi.org/10. 3313/jls1964.30.27. Matsuura, S., 2000. Fluctuations of pore-water pressure in a landslide of heavy snow districts. J. Japan Landslide Soc. 37 (2), 10–19. http://dx.doi.org/10.3313/jls1964. 37.2_10. Matsuura, S., Asano, S., Okamoto, T., Takeuchi, Y., 2003. Characteristics of the displacement of a landslide with shallow sliding surface in a heavy snow district of Japan. Eng. Geol. 69, 15–35. http://dx.doi.org/10.1016/S0013-7952(02)00245-4. Matsuura, S., Matsuyama, K., Asano, S., Okamoto, T., Takeuchi, Y., 2005. Fluctuation of the seasonal snowpack in a mountainous area of the heavy-snow district in the warmtemperate zone of Japan. J. Glaciol. 51, 547–554. http://dx.doi.org/10.3189/ 172756505781829052. Matsuura, S., Okamoto, T., Asano, S., Osawa, H., Shibasaki, T., 2017. Influences of the snow cover on landslide displacement in winter period: a case study in a heavy snowfall area of Japan. Environ. Earth Sci. 76, 362. http://dx.doi.org/10.1007/

141