Keywords

1 Introduction

Erosion and sedimentation by various geomorphic processes have caused many disasters in Japan, which is characterized by a large population of 125 million in an area of 380,000 km2. Both structural and non-structural measures have managed such geomorphic hazards for over a few hundred years (e.g., Kanbara and Imamori 2020). Western Japan is affected by warm currents such as the Kuroshio Current and the Tsushima Current. Especially in summer, the seawater becomes warmer, and the updrafts supply energy to typhoons and often form cumulonimbus clouds continuously. They lead to heavy rainfall in Western Japan. The Japanese archipelago is situated parallel to and near the plate boundaries, so earthquakes have also been a significant threat. This paper introduces several cases of catastrophic sediment disasters caused by landslides and the countermeasures adopted in Western Japan. The word “sediment disaster” (Toki 1993) in this paper describes the disaster caused by the movement of rock, debris, soil or their combination under the influence of gravity and/or water due to erosion, transportation, and sedimentation processes.

2 Geological background of Western Japan

Geological Features in Japan

As mentioned above, the Japanese archipelago, with 111 active volcanoes (Japan Meteorological Agency 2017), lies in the tectonic orogenic belt. The Japanese Islands belong, based on tectonic regionalization, to four plates, namely: North American Plate, Pacific Plate, Eurasia Plate, and Philippine Sea Plate. It can also be divided into Northeastern Japan and Southwestern Japan by the geological structure. Their geological boundary is the Fossa Magna, which is part of the western boundary of the North American Plate.

Southwestern Japan is on the Eurasia Plate, on the subduction zone of the Philippine Sea Plate. Furthermore, the Median Tectonic Line (MTL) is divided into the inner zone of Southwest Japan (on the Sea of Japan side) and the outer zone of Southwest Japan (on the Pacific side) (The Japan landslide Society 2012). The outer zone of Southwestern (SW) Japan consists of the accretionary complex (AC) as sediments carried by the Philippine Sea Plate.

Figure 1 shows the geological structure of Western Japan and the locations and geology of the three sites, namely: Unzen, Aso, and Hiroshima introduced in this paper.

Fig. 1
figure 1

Geotectonic subdivision of SW Japan (added three local geological sites to Isozaki et al. 2010: map view modified from Isozaki and Itaya (1991). Symbols for geotectonic units (belts) and major boundaries (tectonic lines) are from Isozaki et al. 2010. See text for details of Izozaki et al. 2010. The coordinates of the latitude and longitude of the center of the circle with a diameter of 3 km in Fig. 1a–c are as follows: Fig. 1a Unzen site: 32° 45′ 32.99″ N, 130° 21′ 26.61″ E, Fig. 1b Aso site: 34° 52′ 59.78″ N, 130° 59′ 23.18″ E, Fig. 1c Hiroshima site: 34° 28′ 53.32″ N, 132° 29′ 18.40″ E. [Geology] 1 (H_sad): valley floor, intermountain basin, river and coastal plain deposits, Cenozoic Quaternary Holocene, 2 (H_v_ad): volcanic rocks debris avalanche deposits, Cenozoic Quaternary Holocene, 3 (Q1-H_v_af): volcanic rocks volcanic fan deposits, Cenozoic Quaternary Pleistocene Gelasian - Holocene, 4 (Q3_vas_al): dacite and rhyolite lava and pyroclastic rocks, Cenozoic Quaternary Late Pleistocene, 5 (Q2_vas_al): dacite and rhyolite lava and pyroclastic rocks, Cenozoic Quaternary Middle Pleistocene, 6 (H_vas_al): dacite and rhyolite lava and pyroclastic rocks, Cenozoic Quaternary Holocene, 7 (Q3_sn): non-marine sediments, Cenozoic Quaternary Late Pleistocene, 8 (Q2_vis_al): andesite and basaltic andesite lava and pyroclastic rocks, Cenozoic Quaternary Middle Pleistocene, 9 (Q3_vbs_al): basalt lava and pyroclastic rocks, Cenozoic Quaternary Late Pleistocene, 10 (C13-J1_soh_J1): chert Early to Middle Jurassic accretionary complex, 11 (J1_22_sx_J1): mixed rock Early to Middle Jurassic accretionary complex, 12 (K21_pam_a): Igneous rocks/ massive granite island arc and continental, Late Cretaceous Cenomanian–Santonian, (see Geological Survey of Japan 2022).

Local Geology Around Three Sites of Unzen, Aso, and Hiroshima

Figure 1 shows the geological structure classification of SW Japan. The three target sites are shown in Fig. 1a (Unzen site), Fig. 1b (Aso site), and Fig. 1c (Hiroshima site), which are partially quoted from “Seamless Digital Geological Map of Japan (1:200,000) V21” (Geological Survey of Japan 2022).

All three survey sites are located in the inner zone of SW Japan and belong to the Mino-Tamba belt of the Hiroshima site, the Ryoke belt of the Aso site, and the Suo metamorphic belt of the Unzen site (Fig. 1). The distribution area of granite is widespread near the Hiroshima site, and the granite (K21_pam_a) near the site is markedly weathered/decomposed, and erosion due to rainfall and outflow of decomposed granite soil are likely to occur. In the vicinity of the Aso site, tephra from the eruption of Aso Volcano covers a wide area and is susceptible to rainfall erosion. At the Unzen site, pyroclastic flows caused by the eruption of Mt. Unzen and rain-induced debris flows/mud flows occurred. The area is easily eroded/denudated in its present state.

3 Disaster of the Unzen-Mayuyama Mega Slide and Subsequent Tsunami in 1792

Outline of the 1792 Unzen-Mayuyama Landslide

The most famous and worst landslide and its resulting landslide-induced tsunami disaster in Japan is the 1792 Unzen-Mayuyama mega slide. This disaster in the history of volcanic hazards in Japan is called “the Shimabara Catastrophe.” Mayuyama is one of the Unzen compound volcanoes in the eastern part of Shimabara Peninsula, Nagasaki Prefecture, Kyushu (Fig. 2).

Fig. 2
figure 2

(Modified from source: Sassa et al. 2016)

Overview of the 1792 Unzen-Mayuyama landslide and sampling locations

At about 8 PM on 21 May 1792, the landslide was induced by the Shimabara-Shigatusaku earthquake that occurred under Shimabara in the last stage of eruptive activities of Mt. Fugen (Unzen Restoration Office 2002). After travelling around 5 km from the top of Mt. Mayuyama, a tremendous amount of debris and rocks rushed into the Ariake Sea and generated a giant tsunami that hit both sides of the inland sea. The landslide and tsunami reportedly killed a total of 15,153 persons. Out of 15,153 persons, 10,139 persons died in the Shimabara area, 5014 persons were killed on the opposite banks by the tsunami wave.

Documents and Research about the 1792 Unzen-Mayuyama Landslide and Tsunami

Many people well reported the Unzen-Mayuyama landslide and its subsequent tsunami. Many old documents and pictures were presented in pamphlets of the Unzen Restoration Office of Japan’s Ministry of Land, Infrastructure, and Transport (MLIT) (2002, 2003). By comparing old figures and photos showing almost the same landscape of the Mayuyama, Inoue (1999) and the MLIT Unzen Restoration Office (2002) reconstructed the original ground surface before the landslide and the first sliding surface. Topographic changes before and after the 1792 Mayuyama landslide are presented in Fig. 3. There were several volcanic and seismic activities in the area of Unzen Volcano before the 1792 landslide (Unzen Restoration Office 2002). Frequent earthquakes occurred since Nov. 1791. Volcanic eruptions and lava flow with a length of 2 km occurred at Mt. Fugen, 2 km west of Mt. Mayuyama in Feb. 1792. Then the frequent earthquakes and a large landslide (scale: 1080 m in west–east, 720 m in north–south) at the southeast flank of Mt. Mayuyama occurred one month before the mega-slide on 21 May 1792.

Fig. 3
figure 3

(Modified from source: leaflet by the Unzen Restoration office, MLIT 2003)

The 1792 Mayuyama landslide before and after the event

The tsunami hit both sides of Ariake Bay, including Shimabara Peninsula, a coastal area of Kumamoto and the Amakusa Islands. From the present elevation of the stones, it is estimated that the maximum elevation of the tsunami was 57 m. The disaster investigated by the Unzen Restoration Office is shown in Fig. 4. The figure was created based on the previous investigation by Tsuji and Hino (1993) and Tsuji and Murakami (1997). The original image is written in Japanese including the historical data and the location of stone pillars (Tsunami-dome-ishi) which were installed to mark the places where the Tsunami reached. The Tsunami-dome-ishi aimed to share the experience with the future generation.

Fig. 4
figure 4

(Modified from source: Unzen Restoration Office 2002)

Records of disaster by the Unzen landslide-and-tsunami disaster

The total number of deaths is 15,153 persons. In Fig. 4, B to I indicate the followings.

B: The numbers of deaths are shown in the circles (See the legend: the largest is 500 persons), the size of which is proportional to the number of human fatalities in the area.

C: The most significant number of deaths was in Shimabara town around the castle (5251 persons).

D: The second largest deaths was in the southern part of the Shimabara Peninsula (around 3500 persons).

E, F, and G: Tsunami-Dome-Ishi (A stone showing the tsunami reaching that point) was set to record the tsunami by the communities in Kyodomari (E), Umedo (F) and Otao (G) of the Higo (Kumamoto) Han area. The Tsunami-Dome-Ishi in Kyodomari was moved to construct a road, but its former location is marked on the road retaining wall (by the regional education committee). The Tsunami-Dome-Ishi is limited in Higo (Kumamoto) Han area. These tsunami records are reliable.

H, I: Stone pillars for memorial services for deaths by tsunami in Futsu (H) and Mie (I) in Shimabara Han.

Sassa et al. (2016) interpreted the main sliding block and the secondary sliding block, which were pushed forward by the motion of the initial landslide mass. The combined profile of the landslide is presented in Fig. 5 including the initial main landslide block (red dotted mass) and the secondary sliding block (black dotted layer). The landslide’s maximum depth and total volume were estimated to be 400 m and 3.4 × 108 m3, respectively, based on previous research (Furuya 1974). The lines of slope angles for the initial landslide mass and the secondary landslide mass were estimated as 28.1 degrees and 6.5 degrees, respectively. They were used for the undrained dynamic loading ring-shear testing to simulate the initiation of the main landslide and the movement of the secondary sliding block due to the undrained loading from the displaced first landslide mass.

Fig. 5
figure 5

(Modified from Sassa et al. 2016)

Cross section of Mayuyama including 1792 landslide

The sampling location is shown in Fig. 2, in which, sample S1 was taken from a sand layer exposed along a torrent gully in the source area of the landslide, and sample S2 was taken from the coastal area outside the landslide area.

Triggering Factor of the 1792 Unzen-Mayuyama Landslide

Tests on the sample S1 were preformed using the undrained dynamic loading ring shear apparatus (ICL-2) to investigate the initiation and motion of the Unzen Mayuyama landslide. They consist of basic tests (monotonic undrained increasing shear stress tests, pore pressure control tests, undrained cyclic loading test) and landslide simulation tests (seismic-loading tests). Results of those tests were presented in detail in Sassa and Dang (2018).

The monotonic shear stress control tests are the tests of undrained capability, the stress control capability, the precision of stress, and pore pressure monitoring. A large excess pore pressure generation during shearing was observed in all tests suggesting the possibility of rapid landslide motion. The pore pressure control tests were performed to check which level of pore water pressure could initiate the landslide. The undrained cyclic loading test was performed to examine the shear behavior during seismic loading.

The seismic-loading ring-shear test is the most advanced and complicated test to simulate the landslide initiation by the combined effect of pore water pressure and earthquake shaking. The seismic record of the 1792 earthquake could not be obtained, so the authors decided to use a recent earthquake case. The 2008 Iwate-Miyagi Nairiku earthquake which had a similar scale and triggered a mega slide (Aratozawa landslide with a volume of 67 × 106 m3) was selected. And the Iwate-Miyagi earthquake waveform recorded in Miyagi Prefecture (MYG004) was employed for the ring-shear simulation test and the computer simulation for 1792 Unzen Mayuyama landslide. It was suggested that around 1/3 smaller earthquake shaking than the Iwate-Miyagi earthquake should have caused failure under a slope condition with a pore pressure ratio of 0.21.

Reproduction of the 1792 Earthquake-induced Unzen- Mayuyama Landslide

Based on the landslide dynamics parameters measured by ring-shear testing, a computer simulation of the Unzen-Mayuyama was conducted using LS-RAPID software. Values used in the LS-RAPID are well presented in Sassa et al. (2016). The topography before the landslide and the possible sliding surface estimated by previous researchers (Unzen Restoration Office 2002 and 2003) were used in the software. Figure 6 presents the simulation result of the 1792 earthquake-induced Unzen-Mayuyama landslide (Fig. 6a) and the investigated data (Fig. 6b) made by the Unzen Restoration Office (2002). It shows a similarity in the travel distance and the moving area between the simulated landslide and the actual case.

Fig. 6
figure 6

Simulation result of the 1792 Unzen-Mayuyama landslide (a) and data investigated by the Unzen Restoration Office (2003) (b)

Reproduction of the 1792 Unzen-Mayuyama Landslide-induced Tsunami

The Tsunami triggered by the Unzen-Mayuyama landslide was also reproduced by LS-Tsunami software (Sassa et al. 2016) (Fig. 7). After the landslide occurred and entered into the Ariake Sea, the tsunami wave was triggered. In this figure, the bright red color presents the wave more than 5 m above sea level, and the dark blue color presents the wave less than 5 m below sea level. The top parts of the moving landslide blocks are seen above sea level as dark brown color dots. The first waves struck Mie and Futsu towns (Fig. 7a). Then, the wave expanded and reached Ohtao town, Kumamoto Prefecture, on the opposite bank (Fig. 7b). The tsunami wave was reflected from the opposite bank to strike the southern part of Shimabara Peninsula (around Futsu town) again (Fig. 7c). Another reflected wave hit the northern side of the Mayuyama landslide area (around Mie town) (Fig. 7d).

Fig. 7
figure 7

(Modified from Sassa et al. 2016)

Simulation result for the 1792 Unzen-Mayuyama landslide-induced tsunami

4 Landslides by the 2016 Kumamoto Earthquake

Outline of the Kumamoto Earthquake and Associated Landslide Disasters

An earthquake (Mj = 6.5) occurred at 9:26 PM on April 14, 2016, in the central part of Kumamoto Prefecture (epicenter: N32° 44.5′, E:130° 48.5′, focal depth: ca. 11 km) in Central Kyushu. The second earthquake (Mj = 7.3) occurred at 1:25 AM on April 16, 2016 (epicenter: N32° 45.2′, E:130° 45.7′, focal depth: ca. 12 km) (Fig. 8). The total fatality by these earthquakes was 120 people (Japan Meteorological Agency (JMA) 2016). Both epicenters are located near the junction of the Futagawa and the Hinagu active fault zones, extending NE–SW. The seismic faults, i.e., right lateral strike-slip faults, were traced along these fault zones where aftershocks have occurred. The earthquakes were named the 2016 Kumamoto Earthquake by the JMA. Figure 9 shows the three orthogonal components of the acceleration time history of the Kumamoto Earthquake at the K-net KMM005 station (NIED 2016a), about 10 km away from the Aso-Ohashi and Takanodai landslide sites. Aso-Ohashi and Takanaodai landslides will be described below.

Fig. 8
figure 8

Locations of the epicenters of the 2016 Kumamoto Earthquake and the Futagawa and Hinagu fault zones. a Distribution of active faults (red line after Nakata and Imaizumi 2002; brown line after HERP 2013) and epicenters of the M6.5 foreshock and the M7.3 mainshock of the series of earthquakes that occurred in April 2016 (pink cross after JMA 2016). Solid and dashed lines indicate shorelines and prefectural boundaries, respectively. b Distribution of landslides (black polygons) induced by the 2016 Kumamoto Earthquake and river channels (white lines) around the western rim of the Aso Caldera (after NIED 2016b)

Fig. 9
figure 9

Time history of acceleration in three directions for the mainshock of the Kumamoto Earthquake at KMM005 (NIED 2016a)

The earthquake fault has been traced toward the east up to the western part of the Aso Caldera (Geospatial Information Authority of Japan 2017) with the diameters of 18 km and 25 km in E–W and N–S directions respectively. The maximum PGA was recorded at 1316 cm/s2 at Kawayo in Minami-Aso Village (JMA 2016), which is located nearer than the KM005 site.

A Large-scale Slope Failure Near the Aso-Ohashi Bridge and Its Countermeasures

A large-scale slope failure near the Aso-Ohashi Bridge occurred from a slightly convex slope around a ridge of the Aso Caldera rim on April 16, 2016, with a width of 200 m, a length of 700 m, a relative height of 325 m, and a depth of 5–10 m (Fig. 10). The slope failure destroyed National Route 57, the JR Hohi Line, and collapsed the Aso-Ohashi Bridge on Route 325 over the Kurokawa River (Fig. 10). The Kyushu Development Bureau, MLIT, had carried out investigations and countermeasures since 2016 as national projects for the emergency and permanent remedial measures. They were to prevent secondary disasters caused by unstable sediment remaining at the upper slopes where many cracks have been formed and to perform permanent slope stabilization, respectively (Yamagami 2018; Matsumoto 2020; Aso Sabo Office, Kyushu Regional Development Bureau, MLIT 2021) (Fig. 11).

Fig. 10
figure 10

(Modified from the Kyushu Regional Development Bureau, MLIT 2016)

A slope failure induced by the mainshock near the Aso-Ohashi Bridge

Fig. 11
figure 11

The head part of the slope failure and the distribution of cracks around the failure (upper) and the cracks around the failure (white line: lower) (Photo: Kyushu Regional Development Bureau, MLIT 2016)

The geology around the slope failure is composed of andesite, auto-brecciated lava, and tuff breccia alternately deposited horizontally with some open fractures. The topsoil of the ridge behind the slope failure consists of debris deposits, including surficial Kuroboku soil (3 m in depth) and the underlying loosened rocks (Fig. 12). The geological structure shows a slightly reverse-dip** slope composed of volcanic deposits (Fig. 12).

Fig. 12
figure 12

(Modified from source: Kyushu Regional Development Bureau, MLIT 2016) (The location of the section is shown in Fig. 13)

Geological section of the slope failure near the Aso-Ohashi bridge

Based on the topography before the earthquake and the distribution of elastic wave velocity, slope materials with Vp = 1.0 km/s or less in the shoulder part of the slopes were considered to have collapsed during the earthquake (Fig. 13) (Kyushu Regional Development Bureau 2016-2020).

Fig. 13
figure 13

(Modified from source: Kyushu Regional Development Bureau, MLIT 2016) (The location of the section is shown in Fig. 13)

Results of geological investigation around the slope failure near the Aso-Ohashi Bridge by seismic survey

Based on the slope micro-topography formed by the earthquake, and geological survey results, the slope failure and surrounding unstable slopes were divided into six areas (blocks), considering the assumed mass movement types, and the basic concepts for countermeasures were then determined (Fig. 13). The remedial measures for the slope failure were: (1) to prevent rockfalls, surface failure, erosion, and weathering in the head part of the slope failure; and (2) to protect the lower part from erosion and surficial failures of debris deposits of the slope failure.

As emergency measures, the unstable slopes in and around the slope failure were firstly rounded to remove the unstable materials at the head. These works were carried out using an elevated slope excavator with a movable camera and network-enabled unmanned construction method to avoid accidents (Fig. 14). The construction method enabled up to 14 unmanned construction machines to operate on the slope failure site (Kitazawa and Motomura 2021). As a result, rocks and soils in the volume of 17,000 m3 were removed in 70 days.

Fig. 14
figure 14

(Modified from source: Kyushu Regional Development Bureau, MLIT 2020)

Removal of unstable sediment by unmanned construction

Then, using manned construction of slope stabilization as the permanent measures in the middle and lower parts of the slope failure, eight rows of steel-reinforced retaining walls 200 m long each had been constructed with re-vegetation since June 2017 (Fig. 15). A strong wire net has protected at the upper part of the slope failure with the support of steel bars installed into the ground. Anchor works were adopted to stabilize the loosened rock slopes at the side of the slope failure.

Fig. 15
figure 15

Countermeasures adopted for the stabilization of each block (Embarkment and steel-reinforced soil at the lower slope and earth removal and soil sha** at the head slope) (Aso Sabo Office, Kushu Regional Development Bureau, MLIT 2020)

Given many cracks around the exposed bare earth, extensometers, GNSS, and ground inclinometers were installed to monitor the ground surface movement. In contrast, borehole inclinometers and pipe strain gauge were used to observe the underground movement. The in-site observation of cracks was also carried out.

These countermeasure works enabled National Highway No. 57 to be open to traffic in the summer of 2020. The management of the protected slopes has been handed over to the Kumamoto Prefectural Government.

Landslides in Takanodai

Over 1000 individual landslides occurred in the Aso Volcano area by the mainshock of the 2016 Kumamoto Earthquake (NIED 2016b). The slopes were widely covered with fallout tephra layers (Higaki et al. 2019). The Takanodai landslide in the Minami-Aso Village is an example of landslides which occurred in such tephra layers (Figs. 8 and 16). Several landslides were induced by the mainshock on the hill slopes of the area where the Kyoto University Institute of Volcanology stands atop (blocks A to E in Figs. 16, and 17). This hill originated from the Takano-Obane lava dome formed in 51 ka (Matsumoto et al. 1991), and it has been covered with tephra layers.

Fig. 16
figure 16

Landslides at Takanodai caused by the earthquake. A-E: Landslide block (area) (Area of landslide block: Higaki et al. 2016)

Fig. 17
figure 17

a Long run-out landslide (A block), b Head scarp with the exposure of fallen tephra layers indicated as a red arrow (B block), c Run-out materials (Toe of B block), d Main scarp (back) and slide body (front) (E block)

The source area of the largest landslide (block A) at the shoulder part of the hill had a width of 100 m and a depth of 5–10 m with a slope gradient of ca. 15° (Fig. 17a). Exposure at the head cliff show tephra and humic soil layers cover the gentle hill slopes cumulatively (Fig. 17b). The displaced mass of block A moved more than twice as long as the length of its source area (Fig. 16). The other landslides moved far toward the southwest, west-northwest, and then north. Since a telephone pole standing at an in-situ position has been buried by moving materials at the toe of block B (Fig. 17c), and parallel striation remains on the head scarp remain at the block E (Fig. 17d), both disruptive slide and translational slide occurred in the Takanodai landslide. The fact that the landslide of the block E moved 65 m with a house on it without the windows broken (Higaki et al. 2016) indicates the translational slide.

After forming the present Aso Caldera with the outflow of the Aso-4 pyroclastic flow (90 ka), the fallen tephra, mostly from the post-caldera central cones, reached a thickness of 100 m (Miyabuchi et al. 2003). The Kusasenrigahama Pumice (Kpfa), which has formed the slip surface of some of the Takanodai landslides, is a pumice layer fallen in 31 ka and is interbedded at a depth of 3–8 m on the slope (Fig. 18).

Fig. 18
figure 18

Slip surface formed in the Kusasenrigahama Pumice fall layer

Kasama et al. (2018) pointed out that the Kpfa deposited in the loose condition indicated by the high void ratio was compacted quickly by repeated seismic shearing so that the shear strength was decreased and the landslides with large run-out displacement occurred.

5 Landslides Induced by the 2018 Heavy Rain in Hiroshima

Encroachment of Urban Areas Toward Slo** Terrain

In Hiroshima City, with a population of 1.2 million, slope failures and debris flow often occur even with less rainfall than in areas such as Shikoku and Central and Southern Kyushu. One of the reasons for this is geology and rock types (Kaibori et al. 2018). Granitic rocks are widely distributed in and around Hiroshima City (Fig. 19) and are often highly weathered. In addition, many houses are located near the outlet of mountain streams and steep foot slopes, which are prone to sediment disasters. In the outskirts of Hiroshima City, catastrophic sediment disasters were also caused by heavy rain in 1999, 2014 (Tsuchida et al. 2019), and 2018 (Kaibori et al. 2018). Figure 20 shows a debris flow disaster in Hiroshima City that occurred in 2018. The geographical conditions of mountainous terrain encroaching on the plains where large city areas develop, and the spread of residential areas into the mountain sides cause an increase in the number of areas at risk of sediment disasters.

Fig. 19
figure 19

(Modified from source: Hiroshima West Mountain Range Sabo Office, Chugoku Regional Development Bureau, MILT 2019)

Locations of recent sediment disasters and geological distribution around Hiroshima City

Fig. 20
figure 20

Debris flow at Kuchita-minami 3 Chome in the Asa-kita Ward, Hiroshima City (Sediment Control Division, Public Works and Construction Bureau, Hiroshima Prefecture 2019)

In 2000, the Government of Japan enacted the Sediment Disaster Prevention Act to prevent people from living in hazardous areas of sediment disasters, often without knowing the dangers. Under the Act, “sediment disaster hazard areas” are to be designated into two categories based on the investigation of the topography, geology, and land use in the sediment disaster-prone areas (Sabo Department, MLIT 2016). An area of exceptionally high risk is designated as Sediment Disaster Special Hazard Area, where housing and building developments are controlled. In contrast, an area prone topographically to sediment disaster is designated as Sediment Disaster Hazard Area.

In addition, non-structural measures are being developed to provide warning and evacuation from the hazard zones in conjunction with sediment disaster warning information issued by the local government based on local rainfall conditions. Hazard areas are set for debris flow, slope failure, and landslide, each likely to cause sediment disaster in Japan.

The 2018 Heavy Rainfall Disaster in Hiroshima City

During the July 2018 heavy rains, after Typhoon No. 7 passed between Kyushu and the Korean Peninsula on July 3–4, a front stalled between the cold Okhotsk High in the north and the subtropical Pacific High in the southwest of the Japanese archipelago, causing a remarkable inflow of water vapor from the southwest to continue, resulting in heavy rainfall over a wide area in Western Japan (Hiroshima West Mountain Range Sabo Office 2019). The total amount of rain from July 5 to 7 reached 430 mm, 1.7 times larger than the monthly rainfall in July, at Nukushima, Higashi-ku, Hiroshima City (Fig. 21). The precipitation of accumulated rainfall was 1.6–2.6 times larger than previous sediment disasters in 1999 and 2014, and continuous rain was 3.4–20.5 times longer in 2018 (Kaibori et al. 2018). Figure 21 shows the occurrence of slope failures and debris flows that are particularly concentrated in the southern part of Hiroshima Prefecture during the July heavy rains (Sediment Control Division, Public Works and Construction Bureau, Hiroshima Prefecture 2019). The number of debris flows was particularly high. The locations of the slope failures were widely distributed along the Seto Inland Sea in a zonal pattern from southwest to northeast, with a total sediment discharge of 8.1 million m3 (Kaibori et al. 2018) and densities of debris flows and slope failures of 3.01 and 0.30 locations/km2, respectively (Fig. 22: Hiroshima University 2018).

Fig. 21
figure 21

(Modified from source: Kaibori et al. 2018)

Comparison of rainfalls between the past disasters in Hiroshima Prefecture and the 2018 Hiroshima disaster

Fig. 22
figure 22

(Modified from; Base map: http://www.gsi.go.jp/BOUSAI/H30.taifuu7gou.html#6)

Locations of slope failures and debris flows in the southern part of the Hiroshima Prefecture, the total number, and the total amount of sediment discharge volume Sediment Control Division, Public Works and Construction Bureau, Hiroshima Prefecture and Hiroshima University 2019

This extremely heavy rain caused 1242 sediment disaster locations in Hiroshima Prefecture, and the number of sediment disasters is larger than the recent national annual average of about 1100 sediment disasters (2008-2017) (Sabo Planning Division, MLIT 2018). The sediment disasters mainly occurred as steep slope failures and debris flows. Out of the 120 people killed or missing in Hiroshima Prefecture due to this calamity, 87 fatalities were caused by the sediment disasters (Hiroshima West Mountain Range Sabo Office 2019).

As an emergency response to this disaster, the Chugoku Regional Development Bureau of MLIT implemented emergency measures to prevent secondary disasters in the nine districts severely damaged by debris flows (Hiroshima West Mountain Range Sabo Office 2018). The adopted countermeasures included: (1) channel stabilization works using large sandbags to channel water downstream safely; (2) installation of warning devices such as wire sensors to detect the occurrence of debris flows (Fig. 23); and (3) installation of strong wire net barriers to supplement small debris flows as emergency measures (Fig. 24).

Fig. 23
figure 23

(Modified from source: Hiroshima West Mountain Range Sabo Office, MLIT 2019)

Wire sensor for early detection of a debris flow occurrence a wire sensor, b warning device with a speaker and a pilot lump

Fig. 24
figure 24

Wire net barrier to capture sediment (Hiroshima West Mountain Range Sabo Office, MLIT 2019)

On the other hand, MLIT constructed Sabo dams and other structures as permanent countermeasures in 28 streams that urgently need to be addressed since 2014 (Fig. 25, Hiroshima West Mountain Range Sabo Office, MLIT 2020).

Fig. 25
figure 25

(Modified from source: Hiroshima West Mountain Range Sabo Office, Chugoku Regional Development Bureau, MLIT 2020)

Installation plan of Sabo dams for disaster recovery in the catchment of debris-flow affected streams

The Sabo dam constructed in Catchment No.19-299 (Fig. 24) trapped the sediments during the heavy rain in August 2021, protecting the downstream area (Fig. 26). In addition, 20 Sabo dams have been completed after the 2018 disaster for the debris flow-affected areas shown in Fig. 24).

Fig. 26
figure 26

(Modified from source: Hiroshima West Mountain Range Sabo Office, MLIT 2019)

Sediment capture behind a Sabo dam in the catchment 1-9-299 in Fig. 25

6 Discussions

Western Japan has high precipitation in Japan because the region is susceptible to typhoons and rainy season fronts and is surrounded by waters where warm currents enter. In the 2018 Hiroshima heavy rain disaster, the East China Sea supplied water vapor, resulting in high precipitation. In addition, its proximity to a large plate border causes frequent earthquakes and creates many Quaternary volcanoes. Furthermore, geologically and topographically, the region has a wide distribution of mountains with medium to low relief terrain, where erosion proceeds slowly and weathered granitic rocks are easily formed. Sedimentary and metamorphic rocks that have undergone deformation and fracturing of the accretionary complex are distributed in the Shikoku and southern half of the Kyushu regions. There are many slopes covered with fallen tephra layers in and around the volcanic areas.

Compared to Eastern Japan, excluding the Tokyo metropolitan area, Western Japan is densely populated, with residential lands concentrated in the foothills. Because of these natural and social conditions, sediment disasters, such as steep slope failures, landslides, and debris flows, frequently occur in Western Japan. Historically, a tsunami generated by the mega-slide at Unzen-Mayuyama due to an earthquake in 1792 killed ca. 15,000 people.

In Japan, various non-structural and structural measures have been taken to reduce the risk of sediment disasters (Kanbara and Imamori 2020). Stone pillars at various locations in the Ariake Sea, which mark the arrival points of the tsunami caused by the Mayuyama mega-slide, are still in place to remind the future occurrence of landslide tsunami risk. Previous documents, including drawings, are also helpful for recognizing the phenomena that induced catastrophic disasters in the ancient times (Inoue 1999). In addition, simulations of the dynamic motion of the earthquake-induced mega-slide and subsequent tsunamis are helpful for hazard map** from a multi-hazard perspective. Hazard map** and the resulting warnings and evacuations are necessary non-structural measures.

In Japan, the Sediment Disaster Prevention Act legally designates areas susceptible to rain-induced sediment disasters as hazardous areas of different types of sediment disasters, such as slope failures, debris flows, and landslides that can affect important protection objects. In the designated areas, warning and evacuation systems are under development (MLIT 2016).

On the other hand, as structural measures, check dams similar to the Sabo dams (see Fig. 26) have been constructed for a long time to control sediment discharge from mountain streams and to stop debris flows. The oldest masonry check dams, constructed by the Fukuyama-han (Edo era domain) 320 years ago, still remain (Takanashi et al. 1997). After World War II, concrete dams became mainstream. The Sabo dams constructed in the wake of the 2018 Hiroshima heavy rain disaster captured a large amount of sediment and prevented damage to the downstream residential areas.

Recent developments in information technology have significantly advanced the prevention of secondary disasters, such as the search for missing persons and workers for disaster prevention work, and construction in watersheds and slopes where unstable sediments still remain after sediment disasters. Here, we introduced sensors for detecting debris flows and unmanned construction on slopes with unstable sediments.

However, temporal and spatial prediction of earthquake-induced sediment disasters is still difficult. The large-scale slope failure near the Aso-Ohashi Bridge due to the Kumamoto Earthquake occurred on a slope of the gentle convex ridge, covered with a thick weathered layer mainly of volcanic rocks. Because the restoration of roads, railways, and bridges connecting the prefectural capital was urgently required, the potential slope failure was stabilized by removing unstable sediment with Vp = 1.0 km/s or less on the upper slope, and by filling and earth retaining at the foot of the slope failure.

Concentrated co-seismic landslides in tephra mantled slopes, especially in hilly areas, have been reported for many earthquakes in Japan (Higaki and Abe 2012; Osanai et al. 2019). Long run-out distances in comparison with the length of source areas are also reported. Wang et al. (2021) pointed out that hydrated halloysite formation due to weathering in the pumice layer liquefied in the fallen tephra deposits during the 2018 Eastern Iburi Earthquake in Hokkaido. In the hills or slopes of dissected terraces around or on the eastern side of Quaternary volcanoes, such weak layers in fallen tephra tend to remain on gentle rather than steep slopes. This also inferred geomorphologically landslide-favorable conditions. A further mechanism of slip surface formation and long run-out movement of earthquake-induced landslide in tephra layers is required to make susceptibility map** of seismic landslide hazard zonation.

7 Conclusions

Western Japan is climatically prone to heavy rainfall and frequent earthquakes due to its proximity to a plate boundary. In Hiroshima Prefecture, debris flows are frequent due to the distribution of weathered granitic rocks. On the other hand, due to the dense population, many residential lands are located at the foot of mountains, and this tendency has become more pronounced with the recent development of residential lands. Many areas in Western Japan are at high risk of sediment disasters due to these natural and social conditions.

Here, in Western Japan, where sediment disasters occur frequently, we introduced some case histories. They were: (1) The large-scale landslide at Unzen-Mayuyama caused by the 1792 earthquake that generated a tsunami,

(2) The large-scale slope failure in weathered volcanic rocks, (3) Landslides in tephra layers, and

(4) The debris flows in residential areas close to mountainous areas in Hiroshima City. In addition, disaster prevention measures were described mainly for the 2018 Hiroshima City disasters and the slope failure near the Aso-Ohashi Bridge.

The 1792 Unzen-Mayuyama mega-slide alerts the tsunami risk caused by a mega-slide of over 108 m3. Dynamic numerical simulation can be an effective tool for its hazard zoning. Stone pillars indicating the landslide-induced tsunami-affected areas are indigenous awareness creation means to the future generation.

In the 2016 Kumamoto Earthquake disaster, slope stabilization works were constructed by removing unstable sediment at the head part and by filling and earth-retaining the areas at the foot of the slope for the early restoration of important traffic routes. Since the Hiroshima heavy rain disasters in 1999, 2014, and 2018, debris flows prevention projects such as the construction of Sabo dams have been promoted, and their effectiveness has been monitored.

In these cases, recent advances in information and construction technologies and structure materials have made unmanned construction possible; early warning system installations are also progressing based on the designation of hazardous areas by the Sediment Disaster Prevention Act. On the other hand, earthquake-induced landslide damage in volcanic tephra-covered hills has occurred frequently. Identifying slopes where such landslides may occur is to be resolved as soon as possible.