Keywords

1 Introduction

Large seismic lateral ground movements that can cause severe damage to lifeline facilities, including roads, railways, waterworks, sewers, transmission lines, etc., are most often associated with soil liquefaction. However, there is more to liquefaction-induced impact than just the immediate damage. Liquefied soil can remain soft, thus causing some problems for post-quake rehabilitation works. The question of how long the liquefied soil can remain soft does not seem to have received as much attention as those on the direct impact of liquefaction. Aguirre and Irikura (1997) examined the main shock and after-shocks of the 1995 Kobe Earthquake, Japan. This earthquake caused severe liquefaction of Port Island, an artificial island of Kobe, the largest in the world at the time of construction, with various facilities including container ship berths, a convention center, universities, etc. Aguirre and Irikura (1997) conducted a back analysis to obtain the chronological change in the shear wave velocities of the island’s fill before, during, and after the mainshock. Their study showed that the liquefied state remained at least three hours after the mainshock but no more than 24 h. The fill’s stiffness decreased close to zero when liquefaction occurred and later increased gradually, following a trend that resembles a consolidation curve.

Konagai et al. (2003) reported that sand ejecta lined a dried river trace in a dry, barren valley about 100 km west of Tehran, Iran, after the Changureh Earthquake of June 22, 2002, that hit this semidesert area. A trench excavated across the sand ejecta revealed liquefaction escape structures, i.e., sand-filled vertical fissures through a silty surface layer. When a loader, a heavy wheeled tractor used to excavate the trench, reached the wet and soft sandy layer 2 m below the ground surface, the loader had stuck in the liquefied sand. It took a while for the loader to get out of the liquefied sand layer. The trench was excavated on July 29, 2002, about a month after the earthquake. This fact suggested that liquefied sands can remain soft even a month after an earthquake, making rehabilitation works with heavy equipment extremely difficult.

This article reviews case histories showing that liquefied soils can remain soft for months.

2 Quantitative Evidence on How Long Liquefied Sand Remains Soft

The 2011 Tohoku Earthquake, also known as the Great East Japan Earthquake, has shown that a long stretch of landfills along the northern shorelines of Tokyo Bay had very high liquefaction susceptibility. These landfills are almost flat, and there was little to see regarding lateral spreading. However, some field surveys have shown quantitative evidence on how long liquefied sand remains soft. Before getting to the main point, we had better have an overall picture of the liquefied Tokyo Bay shore area. Konagai et al. (2013) attempted to detect soil subsidence from raster images converted from airborne LiDAR (Light Detection and Ranging) data before and after the earthquake. Konagai et al. (2013) used the template matching technique for clusters of pile-supported buildings and bridge piers chosen as templates in source images of the target areas to eliminate deep-seated tectonic displacements and thus extract ground subsidence caused solely by soil liquefaction. Figure 1a shows the obtained soil subsidence map of Urayasu. Since the LiDAR survey was conducted for Urayasu on April 20 after almost all sand ejecta were cleared up from streets, the amount of ground elevation loss is considered to be primarily due to the removal of sand ejecta. The aerial photograph of the same area (Fig. 1b) was taken in 1948 by the US Army [19]. This photo clearly shows that it was post-World War II when most of the land reclamation in Urayasu was undertaken and that today, most of the city spreads over the reclaimed land of sand dredged from Tokyo Bay.

Fig. 1
figure 1

a Soil subsidence map of Urayasu (based on Konagai et al. 2013), and b aerial photo of Urayasu on Nov. 8, 1947 [USA M636-A-No2, altitude: 6096 m, photo scale by flight: 39974, Geospatial Information Authority of Japan (2011)]

Full size image

At the location shown by a placemark in Fig. 1, where significant liquefaction occurred during the earthquake, Kiyota et al. (2016) conducted a series of routine Swedish Weight Sounding (SWS) tests once every one or a few months. A pre-quake SWS test was also conducted in 2002, 9 years before the earthquake. SWS is a simple manually operated penetration test under a dead-load of 100 kg in which the number of half-rotations (\({N}_{SW}\)) required for a 25 cm penetration of a rod (screw point) is recorded (Japanese Industrial Standards (JIS) A1221). The obtained \({N}_{SW}\) values for the fill layer were then converted to equivalent N values in the standard penetration test (SPT) (see Fig. 2). A significant reduction in the SPT-N value was seen about a week after the earthquake. Then, the SPT-N values increased with time later and returned to the original values about two months after the quake. Once the shaking is over, the excess pore water pressures that developed in water-saturated sand during the earthquake gradually dissipate. However, this figure suggests that pore water pressure dissipation can take a long time, and liquefied sand can remain soft for months. Even after this long period, the sand barely regains its original strength and hardly exceeds it. The sand is again loosely consolidated with high susceptibility to re-liquefaction.

Fig. 2
figure 2

Chronological change in estimated SPT-N values of fill layer in Urayasu City (Kiyota et al. 2016)

Full size image

To examine if the loosely deposited granular fabric of sand was reproduced, Yamamoto (2015) conducted a Multi-channel Analysis of Surface Waves (MASW) in the Shin-Kiba area, another reclaimed land immediately west of Urayasu. MASW measures the dispersive nature of surface waves (usually the fundamental-mode Rayleigh waves) using an array of geophones and thereby can be used to back-analyze the shallow underground shear-wave velocity distribution. Figure 3 shows a bird’s-eye view of the Shin-Kiba area overlain by the liquefaction-induced soil-subsidence map (Kajihara et al. 2017; Konagai et al. 2013). The MASW survey was conducted along a street through two severely subsided blocks (depicted by two broken-line ellipses in Fig. 3). Figure 3 also shows the shear wave velocity (\({v}_{s}\)) profile in 2D (surface distance and depth) format. The profile shows low \({v}_{s}\) zones 2–5 m underground immediately beneath the two severely subsided blocks. This finding suggests that sands within the low \({v}_{s}\) zones are as weak as before the earthquake and thus remain vulnerable to re-liquefaction.

Fig. 3
figure 3

Bird’s-eye view of Shin-Kiba area (Google Earth) overlain by the liquefaction-induced soil-subsidence map. Subsided areas are shown in blue. Shear wave velocity (\({v}_{s}\)) profile in 2D (surface distance and depth) format along a 760-m stretch of a road is also overlain

Full size image

3 Lateral Spreading and Post-liquefaction Movements

Lateral spreading is one of the most damaging forms of liquefaction. Typically, it occurs on sloping ground. When the underlying soils liquefy, then even a mild slope drives the liquefied soils towards open channels, waterways, etc. The lateral movement due to spreading can be significant depending on the geometries of the moving soil mass confined by the topographic surfaces and the underlying liquefied layer. The travel distance can vary from tens of centimeters to tens of meters or even more. Regardless of whether the movement is large or small, the liquefied sand can remain soft, causing post-liquefaction settlements, lateral deformations, etc. This chapter introduces two case histories of lateral spreading. Particularly, the former is extreme in terms of runout distance.

3.1 Widespread Ground Deformation Caused by the 2018 Sulawesi, Indonesia Earthquake

A devastating Mw 7.5 earthquake struck Central Sulawesi, Indonesia, on September 28, 2018, resulting in over 4000 fatalities and severe damage to several areas in and around Palu City. Sulawesi is a vast island of Indonesia with four spindly arms spinning outward. Palu city is the innermost part of Palu Bay that cuts into the northernmost arm joint. The south-north trending Palu Basin that extends straight to Palu Bay features the unique terrain of the epicentral area. The epicentral area is a typical pull-apart tectonic basin with a thick alluvial deposit filling the valley. This earthquake’s most spectacular and devastating aspect was lateral slides of the almost flat alluvial soil deposit along the west and east bounds of the basin. Lateral spreads were the most serious in Balaroa, Petobo, Jono Oge, and Sibalaya (Rohit et al. 2021; Kiyota et al. 2020). The maximum flow distance at each site was reportedly several hundred meters or more. Thus, this is a case history of extremely large lateral spreading.

Various theories for lateral-spread initiations and motions are discussed by many researchers (Kiyota et al. 2020; Rohit et al. 2021). They have a common opinion that liquefiable deposits and multiple capping layers might have triggered the flow slides. Bradley et al. (2019) reported that aqueduct-supported cultivation, primarily for wet rice, raised the water table near ground level, saturating sandy alluvial soils. Thus, the devastating lateral spreading in the Palu Valley could have been a direct consequence of irrigation. However, the cause of the flow slide cannot be determined solely by the presence of the irrigation channel because there was no primary irrigation channel in Balaroa, the west side of Palu Valley (Kiyota et al. 2020).

To discuss (1) immediate and (2) post-quake ground deformations, Konagai et al. (2022) used two sets of topographic data; (1) Pre-and Post-quake Digital Terrain Models of the Palu Basin and (2) Line-of-sight (LOS) deformations over the Palu Basin obtained from ALOS 2/PALSAR 2 on October 12, 2018, and January 4, 2019. The Digital Terrain Models [AW3D DTMs, 2 m resolution, NTT Data, RESTEC] before and after the earthquake are random gatherings of various source data sets at different times. The DTM before the earthquake covers the period from January 1, 2010, to September 27, 2018, while the post-quake DTM covers the period from September 28, 2018, to November 20, 2018. Konagai et al. (2022) subtracted the pre-quake DTM from the post-quake DTM to evaluate the vertical components of the quake-induced Eulerian ground displacement. The difference of the DTMs was first resampled for a blurred image of 10 m resolution to avoid rapid scatters in the image intensity. Then, the Two-Dimensional (2D) Kolmogorov–Zurbenko (KZ) filter (Zurbenko 1986, Yang and Zurbenko 2010) was applied to the image to smear the rapid intensity changes further. The 2D-KZ filter is a series of iterations of a moving average square filter \(\left(2L+1\right)\times \left(2L+1\right)\), where L is a positive integer. It has thus two parameters, the half-length L of the square window and the number of iterations k of the moving average operation. Since the resampled DTM has a 10 m resolution and L was set at 5, the moving average window is 110-m square.

Figure 4 shows the 2D-KZ filtered image of the elevation change over the 8600 m × 8600 m flat-land area south of Palu City. Note that the 4th iteration of the moving average with the half-length of the window set at 50 m conservatively truncates the original target area of 9000 m × 9000 m by 400 m and 400 m in north–south and east–west directions. The area includes two extensive earth flows at Petobo and Jono Oge. The ground deformation was not exclusive to these large flow slide sites. Surprisingly, almost the entire analyzed area, particularly west of the irrigation canal, shows a checkerboard elevation change pattern. The pattern shows alternating positive and negative values of elevation change ranging from − 1.0 m or below to + 1.0 m or more prominent. Though the calibrations of the DTMs are needed, these peak values are substantially large enough to acknowledge the statistical significance of the 2D-KZ filtered image of the elevation change. The positive and negative peaks appear periodically at an average interval of about 1 km. This pattern can reflect the subsurface soil profile that can include liquefied layers.

Fig. 4
figure 4

© Maxar Technologies, NTT DATA Corporation) before and after the earthquake: the two-dimensional (2D) Kolmogorov–Zurbenko (KZ) filter was applied to the image to smear the rapid intensity changes and draw smoother contour lines. The contour lines are drawn with the unit and the interval set at the “meter” and “0.2 m,” respectively. Yellow polygons show locations of extensive flow slides (Konagai et al. 2022)

Difference in the digital terrain models (incl.

Full size image

For confirmation of the post-liquefaction displacement buildup, Konagai et al. (2022) obtained Line-of-sight (LOS) deformations over the same area from ALOS 2/PALSAR 2 on two different days after the earthquake (October 12, 2018, and January 4, 2019). The 2D-KZ filter was applied to the image of LOS deformations to smear the rapid intensity changes and extract smooth contour lines. Figure 5 shows during the three months that the widespread area west of the irrigation canal had been deforming with ridges developing about 15 mm high and troughs sunken about 20–30 mm deep. The troughs drawn with blue contour lines seem to have developed on the depositional areas of the extensive flow slides and along the meandering trace of Palu River. This pattern of LOS displacements contains errors arising from the uncertainty of estimated orbit or topography and the delay caused by the disturbance of water vapor in the atmosphere. However, the LOS displacements are remarkable, mainly on the western side of the irrigation canal. Thus, this pattern triggers us to acknowledge the statistical significance of the 2D-KZ filtered image of the LOS displacements.

Fig. 5
figure 5

Line-of-sight (LOS) deformations over the same area as Fig. 1 from ALOS 2/PALSAR 2 on two different days after the earthquake (October 12, 2018, and January 4, 2019): DInSAR processed by ©RESTEC, Included ©JAXA. The two-dimensional (2D) Kolmogorov–Zurbenko (KZ) filter was applied to the image to smear the rapid intensity changes and draw smoother contour lines. The contour lines are drawn with unit and interval set at the “millimeter” and “5 mm,” respectively. Yellow polygons show locations of extensive flow slides (Konagai et al. 2022)

Full size image

We have heard little about the severe problems confronting rehabilitation of the liquefied Palu Basin, probably because the liquefied flat land was primarily agricultural. However, the findings mentioned above teach us essential lessons to enter the immediate and following stages of rehabilitating liquefaction-devastated areas. We must take necessary measures considering the liquefied ground can remain soft for an extended period.

3.2 Ground Fissures that Diagonally Traversed a Highway at Kausaltar, Kathmandu, Nepal

The 2015 Nepal earthquake (Mw = 7.8), also known as the Gorkha earthquake, struck central Nepal on April 25, 2015, at 11:56 a.m. local time (6:11 a.m. UTC), one of the worst natural disasters to hit central Nepal since 1934 Nepal-Bihar Earthquake. An about 500 m embankment section of the Kathmandu-Bhaktapur Road, a part of Nepal’s arterial Araniko Highway, diagonally crosses a small shallow swampy valley in Kathmandu Basin (Fig. 6). The location is about 2 km southeast of the Tribhuvan International Airport, Kathmandu. Several lines of vertical ground offsets traversed diagonally across this road, making up a swath of ground offset lines (red lines in Fig. 6).

Fig. 6
figure 6

Kathmandu-Bhaktapur road crossing a small swampy valley: red lines show ground fissures diagonally traversing the road. Blue and red place marks are boreholes drilled by JICA and the authors’ team (Konagai et al. 2016; Shiga et al. 2022)

Full size image

The Japan International Cooperation Agency (JICA 2022a) provided grant aid to widen the existing two-lane highway to an arterial, high-standard four-lane road with two service lanes (Construction period: November 2008 to March 2011). The government of Nepal later constructed service roads on both sides and other facilities like overhead pedestrian bridges to secure the efficient utilization of the road. Therefore, the damage to the highway and its quick rehabilitation was a matter of serious worry for concerned parties on both the Japanese and Nepali sides.

Geological and fossil evidence indicates that the Kathmandu Valley was covered by a large lake called Paleo Kathmandu Lake between approximately 2.8 million and 10,000 years ago (Sakai 2001). The small, shallow swampy valley traversed by the highway may have been one of the marks of disappearing lake water, which dried up almost 10,000 years ago.

The offset lines on the ground are about parallel, trending in ENE to WSW direction. These offset lines disappeared beyond their eastern and western ends and were about 300–400 m long at the most, indicating that the failure was just localized within this short extent of the swath. On the eastern end of the fissures, there was a two-span continuous pedestrian overpass. Konagai et al. (2015, 2016) measured elevations of the middle lane of the highway from the eastern pedestrian bridge using a laser rangefinder (Fig. 7). The maximum vertical offset of about 2 m reached near the boundaries between the terrace and the valley. On the other hand, the sagging part of the highway embankment has been slightly pushed up, as shown in Fig. 7.

Fig. 7
figure 7

Change in elevation along the deformed section of Kathmandu-Bhaktapur road (Konagai et al. 2015)

Full size image

There has been little visible evidence of soil liquefaction on the ground surface at Kausaltar. No clear sand ejecta was found. However, some reports indicated the lateral spreading was due to underground liquefaction. The reconnaissance team of the Geotechnical Extreme Event Reconnaissance (GEER) Association (Hashash et al. 2015) excavated a 2.5 m deep trench across the easternmost large fissure, and it revealed liquefaction escape structures that were capped by the clay/fill layers. There were five borehole logs available along the deformed section of the highway. The authors’ team drilled four more boreholes to obtain the soil profiles and the groundwater levels in the target area. Shiga et al. (2022) then obtained the height difference between the upper surface of the aquifer and the lower surface of the silty sand layer at each borehole. They deduced spatial variation of the height difference as shown in Fig. 8. It stands out that the red area where the silty sand lies beneath the groundwater level overlaps the area of fissures associated with vertical ground offsets (white lines). Though the groundwater level may occasionally fluctuate, this suggests that the deeper silty sand layer within the aquifer could have been the primary cause of the lateral spreading.

Fig. 8
figure 8

Difference between the upper surfaces of the aquifer and the silty-sand layer in Kausaltar (based on Shiga et al. 2022)

Full size image

This ground deformation was responsible for the damage to the two-span continuous pedestrian overpass (Fig. 9), from which the authors measured the highway elevations (Konagai et al. 2015, 2016). Its north pier rested immediately on the northeastern-most line of ground offset, while the other two were on the relatively intact hill terrace. Consequently, the northern pier was on an outward tilt, causing the joint between the pier and the deck to open up. The opening expanded even more during the one month between the two authors’ surveys (May 2 to May 31, 2015). There could have been a good chance for any single supported overpass deck to fall upon the highway with its spans expanded. Though it was a two-span continuous overpass, the northern half of the deck was demolished, given the authors’ report. This sequence of events indicates that the ground had moved a little even after the earthquake.

Fig. 9
figure 9

Pedestrian overpass over Kathmandu-Bhaktapur Road at Kausaltar (27°40.475′ N 85°21.865′ E) (based on Konagai et al. 2015): the northern pier was on an outward tilt, causing the joint between the pier and the deck to open up

Full size image

4 Summary

A strong earthquake motion can create excess pore water pressure in water-saturated sandy soils and temporarily decrease the effective stress and shear strength, leading to excessive ground settlements and lateral spreading. The excess pore water pressures often dissipate too slowly, leaving the soil soft even for months, thus causing some problems for post-quake rehabilitation works. The question of how long the liquefied soil can remain soft did not seem to have received due attention compared with those on the direct impact of liquefaction. Two case histories introduced in this article were:

  1. (1)

    Widespread ground deformation of Palu Basin caused by the 2018 Sulawesi, Indonesia Earthquake, and

  2. (2)

    Ground offsets traversed a highway diagonally at Kausaltar, Kathmandu, Nepal, in the 2015 Gorkha earthquake.

In Case history (1), the post-quake InSAR imageries from ALOS 2/PALSAR 2 showed that the widespread area west of the irrigation canal of Palu Basin had been deforming, with ridges developing about 15 mm high and troughs sunken about 20–30 mm deep over the three months (October 12, 2018, to January 4, 2019).

In Case history (2), several lines of vertical ground offsets reaching 2 m traversed diagonally across a 500 m embankment section of the Kathmandu-Bhaktapur Road, a part of Nepal’s arterial Araniko Highway, thus slowing the traffic for rehabilitation. Though the lateral spreading was confined within a narrow and shallow swampy valley at the bottom of Kathmandu Basin, there were some pieces of evidence that the ground was deforming slowly and steadily. A pedestrian overpass span increased gradually over a month (May 2 to May 31, 2015). Given this observation, the deck of the overpass was demolished.

Those involved in post-quake rehabilitation works should remember that liquefied soils can remain soft for months. For this, more case histories like those shown in this article are essential and expected to be collected for further quantitative discussion about post-quake soil deformation buildups.