Elsevier

Engineering Geology

Volume 169, 4 February 2014, Pages 133-146
Engineering Geology

Characteristics of rainfall-induced landslides in Miocene formations: A case study of the Shenmu watershed, Central Taiwan

https://doi.org/10.1016/j.enggeo.2013.11.020Get rights and content

Highlights

  • We explain the triggers for the frequent landslide in the Shenmu watershed in Taiwan.

  • The analysis of relationship between the density of fault and landslide.

  • We do the analysis of landslide SOC.

  • We explain the bank-erosion landslide in the Shenmu watershed.

Abstract

Fractured and high-permeability Miocene formations, exhibiting shallow soil layers and dense faults, and an increasing number of heavy rainfall events have caused severe and frequent landslides in the Shenmu watershed of Central Taiwan. The Shenmu watershed exhibited a landslide ratio that was typically greater than 1.0% during heavy rainfall events between 1996 and 2009, and can be considered one of the most landslide-prone areas in Taiwan. The landslide ratio in the lower hillslopes is approximately 2.8 to 43.3 times greater compared with that in the upper hillslopes. Analyzing based on a self-organized criticality (SOC) perspective indicated that the β value of the Shenmu watershed (1.24–1.26) was the lowest among landslide-prone areas, implying that small landslides dominated the distribution of landslides documented in available records. The six large landslides that occurred include four shallow landsides and two deep-seated landslides (with a landslide depth greater than 10 m). Approximately 82.3% of the small landslides constituted bank-erosion landslides in the downstream watershed, whereas 96.2% of the large landslides occurred in upstream areas with dense faults. Large landslides occurred during only heavy rainfall events in which the accumulated rainfall was greater than 1000 mm in fractured and high-permeability strata with dense faults. The large landslides deposited considerable amounts of sediment into streams, inducing significant bank-erosion landslides in the downstream watershed. Bank erosion in the downstream watershed caused additional small landslides along the river. Large landslides were typically caused by a fault density of 693.6 m/km2, fractured strata with a permeability of 10 3 to 10 4 m/s, and high levels of accumulated rainfall exceeding 1000 mm. The apparent increase in bank erosion, which increased from 103% to 136% between 1996 and 2009, resulted from an increasing number of heavy rainfall events and the amount of sediment deposition caused by major landslides. Frequent landslides result from the increasing number of heavy rainfall events, fractured rock exhibiting dense faults, and steep and shallow-soil morphology in the Shenmu watershed.

Introduction

Sediment-related disasters, including debris flow and landslides, have frequently occurred in Taiwan during the past two decades, particularly following the 1999 Chichi earthquake (ML = 7.3). The most well-documented recent debris-flow events were those caused by typhoons, including Typhoon Toraji in 2001 (Cheng et al., 2005), Typhoon Mindulle in 2004 (Chang et al., 2007), and Typhoon Morakot in 2009 (Wu et al., 2011). The average annual number of deaths attributed to typhoons and heavy rainfall in Taiwan was 134 persons (including missing persons) from 2001 to 2010, and this figure was higher compared with the reports of past decades (Directorate General of Budget, Accounting and Statistics, 2012). The Shenmu watershed was approximately 36 km from the epicenter of the 1999 Chichi earthquake, during which the peak ground acceleration (PGA) reached 250–400 gal (Chung, 1999). The frequency of debris-flow events in the watershed was approximately 0.75 per year during the past 16 years (Table 1) and approximately one per year in the past decade; thus, the Shenmu watershed is a representative debris flow-prone watershed in Taiwan.

Table 2 shows the geomorphological characteristics and deposition materials of the three most debris flow-prone watersheds in the world, namely the Jiangjia Valley watershed in Southwestern China (Liu et al., 2009), the Moscardo Torrent in Northeastern Italy (Marchi et al., 2002), and the Shenmu watershed in Taiwan. The typical geomorphological characteristics of debris flow-prone watersheds have been widely described in the literature (Spalletta et al., 1979, Marchi et al., 2002, Cui et al., 2005, Liu et al., 2009, Lenzi et al., 2011, Yu et al., 2012). Debris flow-prone watersheds are typically small watersheds exhibiting average slopes close to or greater than 25°, high reliefs (> 1000 m), highly concentrated rainfall during the wet season, numerous and frequent landslides, fragile strata composed primarily of highly fractured and weathered sandstone and shale, and abundant colluvium on the upstream hillslopes.

Long-term debris flow observation systems have been established in mountainous areas in several countries over the past 50 years to analyze the characteristics of debris-flow or serve as early warning systems for possible catastrophic debris-flow events; such systems have been established in Southwestern Canada (VanDine, 1985), Southwestern China (Hsiao et al., 2007), Northeastern Italy (Berti et al., 1999), Japan (Kamei and Nowa, 1985), and Taiwan. Currently, 22 long-term debris flow observatories are located in Taiwan, including one in the Shenmu watershed that was established in 2002 and equipped with digital video cameras, geophones, wire sensors, and rainfall gauges. Monitoring data from the Shenmu long-term observation station has primarily been used by the Taiwanese government to develop a “Debris Flow Disaster Prevention and Evacuation System,” and protect the lives and properties of the residents in the Shenmu watershed. Whereas 14 casualties were associated with Typhoon Toraji in 2001, zero was caused by Typhoon Morakot in 2009, indicating the effectiveness of the long-term observation station.

The frequent debris flow events in the Shenmu watershed result from the frequent landslides that have occurred in the past two decades; thus, it is worth discussing why this area has been prone to landslides during that period. The frequent landslide events can be analyzed on the basis of long-term observation data, rainfall data, landslide distribution, and field survey data. How the landslide characteristics of the watershed and macrosediments deposited in the Shenmu watershed affect the morphology of the downstream region can provide insightful morphological information for evaluating other watersheds in which copious sediment is deposited in the upstream region.

Section snippets

Study area

The Shenmu watershed, including the Aiyuzi (area 4.0 km2, average slope 39.3°), Housha (area 26.2 km2, average slope 35.1°), and Chushui (area 8.6 km2, average slope 35.8°) watersheds, is located in the upstream basin of the Chenyoulan River in Central Taiwan, and encompasses an area of 38.8 km2 (Figure 1). The elevation of the Shenmu watershed ranges from 1153 to 2859 m, exhibiting steep mountainous terrain and an average slope of 36°. Land sloped > 30° comprises approximately 70.8% of the watershed

Rainfall type

Rainfall events were classified based on their hourly rainfall intensity and rainfall duration. An average hourly rainfall intensity of > 30 mm/h was considered a high-intensity rainfall event, and rainfall durations of > 24 h, 12–24 h, and < 12 h were considered long-duration, mid-duration, and short-duration rainfall events, respectively. For example, a rainfall event with an average rainfall intensity of 35 mm/h and a duration of 36 h was considered a long-duration, high-intensity rainfall event. In

Landslide proneness of the Shenmu watershed hillslopes

The landslide distribution in the Shenmu watershed is closely related to the geomorphological and geological characteristics of the watershed, particularly the slope and formation lithology. In this study, five landslide inventories were adopted for further analysis, namely the inventories generated before Typhoon Herb (1996) and after Typhoon Herb, Typhoon Mindulle (2004), Typhoon Kamelgi (2008), and Typhoon Morakot (2009).

Fig. 4 shows the distribution of each landslide inventory regarding the

Discussion

The aforementioned analysis provides a clear framework for landslides in the Shenmu watershed. A detailed landslide classification can elucidate the causative factors of these landslides. In this study, the landslides that occurred in the Shenmu watershed were classified as bank-erosion and nonbank-erosion landslides. Based on our field survey experiences, bank-erosion landslides in the Shenmu watershed should demonstrate three characteristics: the lower boundary of the landslide is extremely

Conclusion

The frequent landslides in the Shenmu watershed located in Central Taiwan can be attributed to seismic forcing caused by the 1999 Chichi earthquake, the increasing amount of heavy rainfall and flooding events, steep hillslopes, fractured strata exhibiting a high fault density, and the large-scale sediment input affecting the rivers. These landslides caused severe bank-erosion landslides downstream. Small landslides located in the lower area of the watershed dominated the landslide distribution

References (43)

  • P. Meunier et al.

    Topographic site effects and the location of earthquake induced landslides

    Earth Planet. Sci. Lett.

    (2008)
  • D.L. Turcotte et al.

    Landslides, forest fires, and earthquakes: examples of self-organized critical behavior

    Physica A

    (2004)
  • C.H. Wu et al.

    Geomorphologic characteristics of catastrophic landslides during typhoon Morakot in the Kaoping Watershed, Taiwan

    Eng. Geol.

    (2011)
  • G.A. Yu et al.

    Rehabilitation of a debris-flow prone mountain stream in southwestern China—strategies, effects and implications

    J. Hydrol.

    (2012)
  • P. Bak et al.

    Self-organized criticality

    Phys. Rev. A

    (1988)
  • F. Brardinoni et al.

    Representing the landslide magnitude-frequency relation: Capilano River Basin, British Columbia

    Earth Surf. Process. Landf.

    (2004)
  • S.C. Chen et al.

    Analysis of the slope stabilization and the landslide size on Mt. Chiu-Chiu after the Chichi earthquake

    Environ. Geol.

    (2006)
  • C.Y. Chen et al.

    Discussion of landslide self-organized criticality and the initiation of debris flow

    Earth Surf. Process. Landf.

    (2007)
  • S.C. Chen et al.

    The characteristics of landslides and landslide size in Yu-fong river watershed

    J. Chin. Soil Water Conserv.

    (2010)
  • S.C. Chen et al.

    The influence of macro-sediment from mountainous areas on the river morphology in the HesheRiver

    J. Chin. Soil Water Conserv.

    (2013)
  • J.K. Chung

    Report on Chi-Chi Earthquake of September 21, 1999

    (1999)
  • Cited by (28)

    • Dynamic response of a physical anti-dip rock slope model revealed by shaking table tests

      2020, Engineering Geology
      Citation Excerpt :

      In Taiwan, landslides occur frequently and constantly change the natural topography, damage roads and bridges, and threaten human life and property. The magnitude and effect of landslides are related to the topography, geological condition, and triggering factors such as earthquakes, rainfall, and human activities (Hung, 2000; Shou and Wang, 2003; Lin et al., 2006; Lin et al., 2011; Lee, 2014; Wei et al., 2014; Chen et al., 2014; Li et al., 2018; Shou and Lin, 2019). For instance, the 1999 Chi-Chi earthquake caused a landslide on Tsaoling Mountain that killed 32 people.

    • Rainfall-triggered mass movements on steep loess slopes and their entrainment and distribution

      2019, Catena
      Citation Excerpt :

      The frequency distribution of mass movements decreased with the increasing mass movement volume. The probability distribution is consistent with that determined by Densmore et al. (1997) and Chen et al. (2014). The relative frequency of landslide size also increases with the decreasing landslide area in an inverse power-law relationship up to a limit of a few hundred square meters (Lu and Godt, 2013).

    • Rainfall threshold for initiation of channelized debris flows in a small catchment based on in-site measurement

      2017, Engineering Geology
      Citation Excerpt :

      The precipitation parameters that are most often used in determining thresholds for debris flow are antecedent rainfall, antecedent effective rainfall, precipitation duration, precipitation intensity, cumulative rainfall, and maximum hourly rainfall intensity (Guzzetti et al., 2008). These proposed thresholds have been applied in different regions, including the studies conducted near San Francisco, CA, USA (Baum and Godt, 2010; Godt et al., 2006), the Alps in Europe (Guzzetti et al., 2007), and Taiwan (Chen et al., 2006b, 2014; Huang, 2013). Most of the aforementioned critical thresholds were empirical thresholds that were based solely on the statistical relationship between rainfall and the occurrences of debris flows.

    • Rainfall thresholds for the occurrence of debris flows in the Jiangjia Gully, Yunnan Province, China

      2015, Engineering Geology
      Citation Excerpt :

      The influence of rainfall on landslides differs substantially depending upon the landslide type and materials involved (Guzzetti et al., 2008). Shallow failures are usually triggered by short, intense storms, while deep-seated failures are triggered by long-term antecedent effective rainfall (Li et al., 2008; Piacentini et al., 2012; Chen et al., 2014). Debris flows can be initiated by either shallow or deep-seated landslides.

    View all citing articles on Scopus
    View full text