Keywords

1 Introduction

On the 5th and 6th of July 2017, a heavy rainfall storm struck Northern Kyushu Island, Japan. The storm affected mainly the Northern parts of Fukuoka prefecture (Asakura City) and Oita prefecture (Hita region). The storm, which the Japanese Meteorological Agency (JMA) named “Northern Kyushu heavy rainfall in July 2017”, has caused severe damage to the mountainous area extending between Asakura city and Hita region, Fig. 1. 12-h cumulative precipitation of 511.5, 329.5, and 532 mm were recorded at Asakura meteorological agency observatory, Hita rainfall observation center, and Tsurukawauchi rainfall observatory, respectively. The latter was confirmed to significantly exceed the cumulative precipitation of Kyushu Island’s heavy rainfall events in 2009 and 2012 (JGS 2010, 2013, 2018). Consequently, geo-disasters, including mud and debris flows and landslides occurred within the affected area.

Fig. 1
figure 1

Mountainous area location of the northern Kyushu heavy rainfall-induced geo-disasters, July 2017 (Chikugo river report 2017)

Full size image

Several large-scale slope failure cases were reported, including failure of the top parts of the slopes, scouring, and failure of the beds and shores in the middle basin. It must be noted that the ground is mainly comprised of severely weathered granodiorite and metamorphic rocks. Consequently, large amounts of sediments and driftwood have flooded and accumulated in the downstream region, spreading over private houses and farmlands, causing extensive sediments, driftwoods, and water-induced damage to lives and properties, as illustrated in Fig. 2. Immediately, an investigation team was formed to investigate the affected mountainous area. The team investigated the prevailing situation and conducted various geotechnical tests to define the affected area’s soil mechanical and hydrological characteristics.

Fig. 2
figure 2

a Typical erosion at Shirakitani river. b Catchment of driftwood by check dam (JGS 2018)

Full size image

The ultimate priority in large-scale geo-disasters is to prevent injuries and fatalities. Several approaches to protect human lives were developed, such as constructing infrastructures like Sabo dams. However, considering the current rainfall patterns, the variations in the scale and type of the impacts that are felt, and the adopted countermeasures, novel and innovative approaches are needed. Those approaches are better based on soft measures such as “an evacuation warning system” and “restrictions on land use” to protect human life and define vulnerable zones.

This paper comprehensively summarizes the recent heavy rainfall-induced geo-disaster in Japan, especially in Kyushu Island. The cases are analyzed, and a group of learned lessons is delineated. It provides a fundamental database for effective preparations for future similar geo-disasters subjected to similar external forces. Furthermore, technical obstacles and new approaches to deal with such events are elaborated on from geotechnical and geological points of view. The term “geo-disaster” is used in this context to describe various slope disasters, including debris flows, driftwood, slope failures, landslides, and embankment damages.

2 Characteristics and Potential of the Geo-disasters During the Northern Kyushu Heavy Rainfall in July 2017

The historical maximum cumulative rainfall recorded at Asakura, Kita-Kouji Public hall, and the national AMeDAS and their historical rankings are illustrated in Fig. 3. The values corresponding to AMeDAS are based on the data recorded between January 1976 to July 2017. The Kita-koji Public hall, located 10 km east of Asakura, falls in the top 25 for all cumulative elapsed times. It is ranked 5th for 2-h cumulative precipitation, 4th for 3-h, 2nd for 6-h (close to the maximum record), and 1st for 12-h (exceeding the maximum record of Asakura in 2017 by 100 mm). The rainfall was unprecedented for 1–3 h and one of the largest for 6 and 12-h cumulative records. The intense rainfall lasted 9 h and is considered one of the most devastating records. Figure 4 shows the precipitation records for Fukuoka in 2017 compared to Hiroshima in 2014, where rainfall lasting more than 9 h (Fukuoka record) is considered an unusual extreme phenomenon in the region.

Fig. 3
figure 3

Comparison and ranking of the recorded cumulative precipitation with time

Full size image
Fig. 4
figure 4

Comparison of precipitation records (Fukuoka and Hiroshima prefectures in Japan)

Full size image

In general, northern Kyushu is experiencing frequent occurrences of high precipitation in a relatively short time, such as the recent 2009 and 2012 events. Statistical studies by the Japanese Geotechnical Society, including data over the past 4 decades in Japan, have revealed a distinct increasing tendency for more frequent heavy rainfall events with hourly precipitation exceeding 50 mm or even over 80 mm (JGS 2010, 2013, 2018).

On a global scale, according to the report released by the Intergovernmental Panel on Climate Change (IPCC 2013), global warming is expected to cause an increase in the heavy rainfall-induced geo-disasters and affect the rainfall concentration, frequency of typhoons and tornados, and wind speed. Coping with the associated significant increase in the potential of geo-disaster occurrence requires effective adaptations and implementations, which are expected to be in high demand in the near future. Under such circumstances, geotechnical engineering is expected to play a vital role. In Japan, a research initiative was launched to establish a geo-hazard vulnerability assessment framework for areas affected by climate change. It incorporates the regional differences between Kyushu and Hokkaido islands (JSPS, Grant-in-Aid for Scientific Research A 2020).

3 Types of Heavy Rainfall-Induced Slope Failures and Sediments Movement

3.1 Geotechnical and Geological Factors

The factors causing slopes to fail can be broadly categorized into predisposition and inducing force factors. Generally, the predisposition factor depends on the geological conditions, topographical conditions, and the presence of vegetation, while the inducing forces include rainfall and earthquakes. In the case of the Northern Kyushu heavy rainfall in July 2017, the inducing force of the slope failure was the heavy rainfall. A specific slope to collapse requires a weak predisposition and a trigger to induce the failure (Iseda et al. 1982). Such a phenomenon occurs only when those conditions coexist, but the thresholds are macroscopically regional and mechanically exclusive for a specific slope.

To be more specific, slope failures can be related to several points, including (1) the increase of the sliding force and decrease of the soil strength due to the saturation, (2) the decrease in the effective stress associated with the rise of the groundwater level which results in increasing the pore water pressure, (3) the collapse of a soil layer or a bedrock stratum, (4) the generation of an osmotic pressure acting as a slipping force due to seepage flow, (5) scouring, erosion, and transport of sediments due to the surface runoff, (6) the piping phenomenon caused by preferential water pathways, (7) the difference in the precipitation and infiltration rate, where the water flowing into the ground surface at a specific rate induces a hydraulic gradient leading to developing a shearing force that contributes to the sliding force (JGS 2018; AMeDAS 2018; IPCC 2013; Iseda et al. 1982). However, such typical relationships change over time for various reasons, including the progress of weathering of the comprising soil and rock layers.

Considering the unprecedented heavy rainfall that lasted for a relatively long time during the Northern Kyushu heavy rainfall in July 2017, the 7 factors mentioned above might have coexisted. Consequently, several slopes failed simultaneously within the affected area, specifically the zone located on the right bank of the Chikugo River in Asakura. However, the intercorrelation of the factors and the occurrence of the slope failure are not well understood yet. Therefore, developing a comprehensive slope stability analysis and collapsing potential evaluation techniques based on in-situ investigations and laboratory testing is crucial and urgent.

Several attempts are being approached from a geotechnical engineering perspective in response to the urgent need. For example, supported by the JSPS and in collaboration with researchers around Japan, the following research objectives are being challenged:

  1. 1.

    Evaluation of the ground resilience against failure. The study considers the changes in the natural and man-made slopes comprising soil profile characteristics with time, adopting statistical and mechanical approaches (Yasufuku et al. 2021; Alowaisy et al. 2020). Furthermore, the development of the rainfall patterns in Fukuoka prefecture with time was analyzed using the AMeDAS rainfall data provided by the Japan Meteorological Agency starting from 1988. Consequently, it was found that the 2020 rainfall-induced road damages were located within high-risk zones, where a relatively long-lasting heavy downpour was probable (Murakami and Oda 2021).

  2. 2.

    Comprehensive analysis of the regional characteristics of the landslide hazard criterion lines (CLs). The CLs are established in each region, considering the ground, topography, geological characteristics, and rainfall information while expanding the existing database. A real-time slope failure evaluation was proposed by combining machine learning with a physically-based slope stability model that considers the strength parameters, water retention curve, topography, geological characteristics, and rainfall conditions (Sakamoto et al. 2022).

  3. 3.

    Update the existing CLs records to include undefined areas not covered by the Landslide Disaster Warning Information System or categorized as non-disaster rainfall. The attempt aims at enhancing the accuracy of the CLs estimation to evaluate the ground resilience reflecting the regional characteristics of rainfall and ground using Artificial Intelligence (AI). Furthermore, precise determination of the dangerous spots on slopes and releasing warning was investigated, focusing on the runoff, infiltration, and snowmelt simultaneously in a wide and narrow area (He and Ishikawa 2022). The post-rainfall earthquakes’ loading on the stability of embankments was investigated by Nguyen and Kawamura (2022). Moreover, Kasama et al. (2021) investigated the effect of spatial variability on the stability of the 3D slope subjected to earthquakes. A stochastic response surface method (VRSRSM) combined with a variance reduction method that considers the spatial averaging of soil parameters in three-dimensional space was proposed to reduce the computational cost. The VRSRSM was applied to evaluate the deep-seated landslides caused by the 2016 Kumamoto earthquake together with the inverse determination of the strength parameters.

  4. 4.

    Propose a multi-scale slope stability analysis method that considers infiltration and runoff during heavy rainfall events. Furthermore, the results are then used for identifying slope hazard risk areas based on the surface water generation associated with rainfall probabilities, which generally cannot be handled using the standard rainfall design criteria (He and Ishikawa 2022).

　Ultimately, these results are expected to be integrated into an engineering assessment of ground immunity, a modified definition of ground resilience to disasters that considers the changes with time. The study focuses on defining risk areas based on the changes in the rainfall and snowmelt patterns while upgrading the risk assessment methods for natural and man-made slopes in Kyushu and Hokkaido regions as disaster vulnerable areas affected by climate change. Furthermore, the project promotes prioritizing the order of the countermeasures for establishing disaster risk maps that consider the change over time.

3.2 Rainfall Patterns and Forms of Slope Failure

Figure 5 shows the collapsed slope and floods distribution during the 2017 rainfall. Figure 5a corresponds to 3-h cumulative rainfall contours, while Fig. 5b illustrates the 12-h contours. The highlighted blue regions represent the flooded areas, while the red highlights correspond to the failed slopes. It can be observed that for 3-h of cumulative rainfall, most of the collapses occurred in regions with rainfall exceeding 200 mm. While, in the Shirakidani River and the Sozu River, collapses associated with rainfall less than 150 and 200 mm were reported, Fig. 5a. On the other hand, for 12-h rainfall exceeding 300 mm, collapses extended over vast regions while exceeding 400 mm became more localized.

Fig. 5
figure 5

Slope failure distribution (red highlights) and flood reach range. a 3 h rainfall contour. b 12 h rainfall contour

Full size image

Figure 6 summarizes the recorded cumulative precipitation versus the collapsed area ratio of the slopes at the right bank of the Chikugo river basin for several durations (Jitozono 2017; Committee of Chikugo river 2017). The collapsed area ratio was determined based on field surveys and investigations for different rainfall durations at various slopes within the affected zone. It can be observed that when the precipitation exceeds 100 mm for 1 h, 250 mm for 3 h, 350 mm for 6 h, 400 mm for12 h, and 450 mm for 24 h, the collapsed area ratio increases rapidly. Establishing a robust statistical-based system that considers the regional characteristics for comprehensively analyzing the rainfall and investigation records is expected to serve as an effective tool for risk assessment, disaster prevention, and mitigation.

Fig. 6
figure 6

Relationship between the cumulative precipitation and the collapsed area ratio (Committee of Chikogu river 2017)

Full size image

Estimating the amount of generated sediments and driftwoods due to extreme heavy rainfall-induced slope failures is essential when determining the type and scale of the preventive measure structure (such as Sabo dam) built around rivers’ facilities. Therefore, it is crucial to properly evaluate the scale of the slope failure under a specific pattern of heavy rainfall. However, the scale and failure pattern vary significantly depending on many factors such as geological conditions, topographical conditions, vegetation cover, and rainfall intensity. An accurate method to estimate the amount of generated sediments does not exist for the time being. Therefore, a method utilizing a statistical approach based on measurable mechanical factors is needed.

In an attempt to establish such a system, the records of the sediment transport were analyzed, focusing on the heavy rainfall-induced slope failures for the Akatani river basin in Asakura. Figure 7 illustrates the relationship between the cumulative precipitation and the generated amount of sediments (Jitozono 2017). It can be observed that the larger amounts of generated sediments are generated for higher cumulative precipitation, which is expected to follow a power formula trend. Besides, a typical collapse mode is expected to occur once the cumulative precipitation reaches 200–500 mm. Although this figure does not include the actual measured amount of generated sediments, reflecting the history of such disasters might be efficient for predicting the type of failure and the amount of generated sediments under specific cumulative precipitation at the considered basin. When considering such heavy rainfall-induced geo-disasters in mountainous areas, enhancing the resiliency of the basin is inevitable. Therefore, a reliable evaluation approach is needed to analyze the existing risks and factors for future mitigation and countermeasure plans. Such attempts are expected to contribute to various geotechnical engineering aspects, including improving the accuracy of flood inundation analyses that directly affects the generated volume of driftwood and sediment.

Fig. 7
figure 7

Relationship between the cumulative precipitation and the amount of generated sediments (Jitozono 2017)

Full size image

4 Shapes and Patterns of Slope Failure

The geometry and the pattern of failure, including the collapsed height and depth of the heavy rainfall-induced collapsed slopes in 2017, were analyzed by Kasama (2018). The studied area included the basin where slope failures were reported in Asakura city, Fukuoka prefecture (Akatani river, Otoishi river, Shirakidani river, Sozu river, Kita river, and Naragaya river). A comparison utilizing laser scanning results of the area profile before and after the collapse was carried out. A detailed description of the adopted methods and the geological formation can be found in Kasama (2018).

Figure 8 shows the slope inclination angle, for the studied basins in Asakura city, versus the collapse frequency and cumulative frequency distribution before and after the failure. Furthermore, Japan’s average cumulative frequency of slope failures is also illustrated (Koyamauchi et al. 2009). It can be observed that the highest frequency in the Northern Kyushu rainfall disaster of 2017 corresponds to an inclination angle of around 40°, while Japan’s average failure inclination angle is around 60°. Moreover, 80% of slope failures occurred at an inclination angle of 38° or less, which can be considered a unique feature of this region. This angle is 20° smaller than the average angle corresponding to the average in Japan. Similarly, Fig. 9 shows the slope collapsed height versus the collapse frequency and cumulative frequency distribution before and after the failure. By conducting cumulative frequency comparative analysis and correlating them to the geological and geotechnical characteristics of the comprising profiles, it is expected that the collapsed slopes can be categorized from a statistical perspective and accumulated as a record.

Fig. 8
figure 8

Slope’s angle distribution of collapsed slopes (Kasama 2018)

Full size image
Fig. 9
figure 9

Slope’s height distribution of collapsed slopes (Kasama 2018)

Full size image

Figure 10 summarizes the average collapsed depth for the studied slopes in Asakura city. Furthermore, the collapsed geological formation and area ratio for each catchment basin, defined as the ratio of the failed area to the catchment area, are illustrated in Fig. 11. It can be seen from Fig. 10 that the average landslide and slope failure depth in all basins ranges from 0.4 to 1.3 m. However, several large landslides from deep layers with an average depth exceeding 8 m have occurred. In addition, from Fig. 11, it can be seen that the collapse area ratio in each basin under the prevailed rainfall falls in the range of 3–8%. Integrating various indices reflecting the rainfall and the generated soil volume into a statistical record of the failed slopes is expected to serve as an efficient risk assessment tool for prevention, mitigation, and estimating the scale of impact on society under such events.

Fig. 10
figure 10

Distribution of the average collapsed depth (Kasama 2018)

Full size image
Fig. 11
figure 11

Comparison of the collapsed area ratio (Kasama 2018)

Full size image

5 Technical Issues Related to Geo-disasters Prevention and Mitigation

Based on the lessons learned from the geo-disasters history in Kyushu, geotechnical and geological-related issues and concerns are summarized in Fig. 12, considering enhancing the geo-disasters prevention and mitigation protocols.

Fig. 12
figure 12

Geotechnical and geological considerations for slope-related Geo-disasters mitigation and prevention system

Full size image

5.1 Preparation of the Past Disaster Records as a Database for Effective Usage

The Northern parts of Kyushu island have experienced several heavy rainfall-induced geo-disasters that have occurred repeatedly with different scales over the past few decades. Although many valuable records exist for Northern Kyushu and all of Japan, they are not efficiently organized for risk management and mitigation usage. Those records are preserved separately in each department, such as the administrative office. The records are not effectively organized as an integrated geo-disasters database due to the lack of efficient, accessible, and easy-to-use digital storage format. Therefore, developing an organized system to collect the geo-disasters history records in cooperation with the national and local governments under the leadership of academic and professional engineers specializing in geo-disaster prevention and mitigation is essential. The efficient and accessible database and system are expected to be conveyed to the next generation as a reference for future geo-disasters.

5.2 Screening Technology Considering the Stratigraphy, Topographical, and Geological Interpretations

Recently, the accuracy of the laser profile scanning data and the C-X synthetic band radar has improved dramatically. Besides, image analysis technologies using drones have improved remarkably. During the last decade, many organizations have been utilizing the latest technologies in developing topological interpretations and screening techniques based on the analysis of the geological and stratigraphic conditions. Utilizing such technologies, if the vulnerable slopes can be defined using practical indices with high accuracy, it is expected to significantly improve the quality and efficiency of geo-disasters prevention and mitigation. Such academic and practical integrated approaches are highly needed.

5.3 Enhancement of the Ground Information Database in the Mountainous Areas

Kyushu branch of the Japanese Geotechnical Society has created and published a database including geotechnical and geological information with more than 80,000 boring data in seven prefectures within Kyushu Island. The database is open for access to public users through the JGS society. These data are enormous in number and include both urban and coastal areas. However, it is essential to enrich the data of the mountainous regions to efficiently contribute to the recovery and restoration in the case of geo-disasters in such regions. There are various cases and ways where geotechnical engineers can efficiently enrich the database during their response to the requests of the state and the local governments. For example, when collecting data from ground surveys and field investigations in the case of a specific geo-disaster and its restoration, establishing a system for checking the collected data is of significant value. Especially for recent geo-disasters, it is necessary to develop a reliable database for disaster response that can be effectively used when deciding on a countermeasure while analyzing the characteristics of that specific region. The system should be easy-to-understand and consider the risk analysis from an academic perspective to objectively explain the reason for deciding on a specific group of actions in response to a geo-disaster.

5.4 Geo-disasters Periodicity

The periodicity of landslides or slope failures in areas comprised of granite was studied in the 1980s (Shimokawa et al. 1984). The elapsed time since the occurrence of the previous landslide, the changes in the soil surface layer thickness, and the rainfall catchment area were analyzed using surveys and field investigation records. The study aimed to define the average cycle time for a landslide. It was reported that for an average failure depth of about 0.7 m, the periodicity is approximately 200 years. Establishing a system to accurately estimate the risk of a slope failure based on its periodicity requires extensive research to investigate the surface soil weathering and restoration rates as functions of time, geotechnical and geological characteristics of the comprising profiles. In Kyushu, several high-risk zones are mainly comprised of granite. Thus when considering geo-disasters, the surface soil weathering and restoration due to sediment deposits over time are vital factors. Therefore, it is not sufficient to specify a landslide hazard based only on an ordinary topographical analysis. Instead, it is crucial to investigate the distribution of sediments on slopes and mountain streams, vegetation at the site, the degree of weathering of granite at the head and toe of slopes, and the periodicity of the past slope failures and debris flow at that relevant site, to efficiently reduce the risk of slopes related geo-disasters. Finally, based on such practical detailed knowledge, society is urged to accelerate the efforts to define landslide and slope failure hazard areas.

5.5 Assessment of Time-Dependent Geo-disaster Immunity

Road cut slopes are man-made and common infrastructures usually constructed with surface reinforcement to prevent erosion and weathering. However, it has been observed that decades after construction, the ground behind the cut-and-fill surface comprised of soil and rocks deteriorates due to the weathering, resulting in deformation and finally collapse.

Recently, a practical and simple method for assessing slopes’ stability considering the degree of weathering and the aging factor is being developed (Yasufuku et al. 2021).

It is not economically feasible to implement preventive measures covering all the existing hazardous areas of the existing slopes. Therefore, it is necessary to incorporate a social impact index, for example, the traffic volume or the distance of diversion roads and detours, to prioritize the cut slope surfaces that need to be managed first. The method proposes a term called slope ‘disaster immunity’, defined as the product of the ‘disaster resistance’, which is an index of the physical stability of the slope, and the ‘disaster resilience’, which reflects the social impact. Supposing the geo-disaster immunity can be objectively and concretely evaluated, it is expected to be used as a system to support the administrative evaluation for prioritizing management and measures of slopes, taking into account the changes over time. Ultimately, it is expected to be utilized for both natural and man-made slopes.

$$I(t)=S(t) \times R(t)$$
(1)

where I(t): the expected value of time-dependent disaster immunity of the target slope; S(t): the probability of soundness of the target slope (disaster resilience), taking into account the time-dependent deterioration, R(t): the ability to mitigate social losses in the event of a geo-disaster.

6 Conclusions

Recently, heavy rainfall events have induced various geo-disasters, including floods, sediments, and debris flows in Kyushu Island and all around Japan, which have caused severe damage to lives and properties. Table 1 summarizes the geo-disasters that occurred in Kyushu island from 1969 to 2020, with the 2017 event analyzed through this paper in bold. Figure 13 illustrates the numbers and the corresponding percentages of the slope geo-disasters all around Japan in the period ranging from 1967 to 2012 (Ministry of Land, Infrastructure, Transport and Tourism 2013). The slope-related geo-disasters in the Kyushu-Okinawa region are remarkably higher than the other regions, accounting for approximately 31%, where on average, 390 slope-related geo-disasters occur annually.

Table 1 History of geo-disasters mainly in Kyushu from 1969 to 2020
Full size table
Fig. 13
figure 13

Slopes related geo-disasters frequency since 1967 (Ministry of Land, Infrastructure, Transport and Tourism 2013)

Full size image

According to the Intergovernmental Panel on Climate Change (IPCC 2013), the frequency and intensity of localized torrential rainfall events are expected to increase. Through this study, the anticipated increase in the geo-disasters inducing forces due to climate change (rainfall), deterioration of the infrastructure, and the decline in the labor forces, limiting the prevention and restoration capabilities due to the increasing difference in the death to birth ratio, were discussed and several lessons and recommendations for dealing with future similar events were introduced.

By carefully comparing and analyzing the situation of the repeated geo-disasters and reflecting the obtained results to the geo-disasters mitigation and prevention practice, developing innovative systems and techniques that integrate the academic disciplines in collaboration with the residents and government is now strongly needed more than ever, see Fig. 12.

Finally, A new concept called ‘disaster immunity’, which can reflect the past rainfall and earthquake histories and the characteristics of particular landforms that change over time was introduced. The concept is being tested and used in a comparative study of Kyushu and Hokkaido islands, where both regions include soil profiles comprised of volcanic ash and weathered residual soils that are vulnerable to climate change-related disasters. For the time being, the obtained results indicate that it can be used as a system to support the administrative evaluation for prioritizing management and measures of slopes, taking into account the changes over time to save lives reliably.