Assessing landslide characteristics in a changing climate in northern Taiwan
Introduction
Landslides are a common natural hazard in mountainous areas worldwide. Even though the mechanism of landslides is very complicated, the most common triggering factor is extreme rainfall (Larsen and Simon, 1993; Crozier, 1999; Glade et al., 2000; Khan et al., 2012; Saito and Matsuyama, 2012; Oku et al., 2014). Therefore, many previous studies have tried to establish the relationship between rainfall conditions and landslide characteristics and used empirical models to assess a possible landslide scale under different conditions of heavy rainfall. Among the early studies, Uchiogi (1971) compared the average cumulative rainfall of a catchment with the landslide-area ratio to establish the relationship between them, thereby obtaining the critical rainfall that caused landslides. In recent years, many studies in Taiwan have also used this kind of empirical model to assess landslides in a catchment. For example, Shou et al. (2011) investigated the behavior of landslides in Central Taiwan after 1999 for the Ta-Chia River, the Wu River and the Chuo-Shuei River. They considered landslides triggered by four typhoons, Toraji in 2001, Mindulle in 2004, Sinlaku in 2008, and Morakot in 2009, to establish the empirical relationship between landslide-area ratio and average cumulative rainfall for those catchments. They found that the impact of the 1999 Chi-Chi earthquake on landslides triggered by subsequent typhoons decayed to 50% in about one to three years and to 10% in about ten years. In addition, Liu et al. (2013) further discussed the impact of the 1999 Chi-Chi earthquake by combining an empirical model with a statistical model to analyze changes in landslides from 1996 to 2008 in the Chuo-Shuei River catchment. Chou et al. (2017) analyzed the coastal alluvial fan in Suao, Taiwan, formed by landslide materials transported during Typhoon Megi in 2010, to establish an empirical model to assess the proportions of new landslides in the upstream area under different cumulative rainfalls. Similar empirical models have been widely used all over the world. Zhu et al. (2011) analyzed loose materials sourced from landslides caused by the 2008 Wenchuan earthquake in China and transformed into debris flows by subsequent torrential rain events, and explored the relationship between rainfall conditions and landslide area using an empirical model combined with rainfall recurrence intervals. Their results show that the torrential rain events of the 50-year and 100-year recurrence intervals will likely generate 0.87 and 1.67 km2 of new landslide area, respectively. Although the simplicity of the empirical approach neglects important hydrological controls, it offers a straightforward means for identifying regional-scale landslide characteristics based solely on rainfall data. The above-mentioned previous studies all established the empirical relationship between rainfall conditions and landslide characteristics based on cumulative rainfall; however, many studies have pointed out that mean rainfall intensity and peak rainfall intensity are also closely related to the occurrence of landslides (Aleotti, 2004; Dahal and Hasegawa, 2008; Chen et al., 2015). In addition, the relationship between rainfall intensity and duration, called the I–D relationship, is commonly used as an indicator to assess the occurrence of landslides in studies of rainfall-induced landslides (Caine, 1980; Guzzetti et al., 2007, Guzzetti et al., 2008; Brunetti et al., 2010; Saito et al., 2010; Chen et al., 2015). In these studies, the I–D relationship was used to find the rainfall threshold for landslides, such as 2% or 5% of the lower bound determined by statistical methods including the Bayesian inference method, the quantile-regression method, and the frequentist method (Guzzetti et al., 2007; Brunetti et al., 2010; Saito et al., 2010).
Another important aspect of landslides is their frequency of occurrence. The recurrence interval of rainfall conditions that trigger landslides is often used to determine the frequency of landslides. Iida (2004) studied the recurrence intervals of rainfall conditions associated with shallow landslides in the area east of Hamada in Shimane Prefecture, Japan. Heavy rainfall (about 350 mm in 6 h) on 15 July 1988 triggered many shallow landslides in that area, and he found that rainfall events with recurrence intervals of <500 years and <1000 years respectively trigger about 50% and 65% of the total number of landslides over the long term. On the other hand, Saito and Matsuyama (2015) produced a preliminary high-resolution probable hourly precipitation and probable Soil Water Index (SWI) for a 50-year recurrence interval over the Japanese archipelago from 5-km grid-cell radar/raingauge-analyzed precipitation with a 26-year time series (1988–2013). Their results revealed detailed spatial patterns of the probable hourly precipitation and SWI with a 5-km grid-cell, which enabled nationwide landslide hazard assessment.
In addition to the triggering factor, geomorphological and geological conditions are also very important factors in the landslide occurrence (Yalcin, 2008) and are often used to determine the susceptibility of an area to landslides (e.g., Ayalew and Yamagishi, 2005). The bedrock geology is highly related to the occurrence of landslides, which may or may not favor displacements along well-defined shearing surfaces (Migoń et al., 2017). Among the geomorphological factors, elevation, slope, and aspect are commonly used in landslide research as they have different significances in the occurrence of landslides (e.g., Guzzetti et al., 1999; Paudel et al., 2016) Varied geomorphologic and geologic conditions in different areas cause differences in the effects of typhoons on landslide occurrence and characteristics.
Climate change is an important issue that has attracted growing attention in recent years. The report of the International Panel on Climate Change (IPCC) noted that rises in temperature and sea level are ongoing and that appropriate adaptations to reduce disaster risk are necessary (IPCC, 2007). In the current changing environment, both the probability of the occurrence of a strong typhoon and the rainfall intensity during a typhoon event will increase (Emanuel, 2005; Webster et al., 2005; Tu et al., 2009). Many studies have explored the potential effects of different climate change scenarios on slope disasters (Melchiorre and Frattini, 2012; Turkington et al., 2016). According to the Taiwan climate change report, the frequency of extreme rainfall events has also increased in the past 20 years in Taiwan (Hsu et al., 2011). Taiwan is extremely susceptible to landslides because of its steep mountainous topography and frequent heavy rainfall and earthquakes (Chen et al., 1999; Chen and Su, 2001; Dadson et al., 2003; Chuang et al., 2009). Under the impact of climate change, extreme rainfall events will become more frequent, and the possibility of landslides in the upstream area will increase significantly. These disasters will result in considerable losses of life and property. However, there is almost no study addressing the effect of climate change on landslides in Taiwan. Therefore, the objectives of this study are to establish the empirical relationship between rainfall conditions and landslide characteristics for two catchments in northern Taiwan using historical cases and to assess changes in the characteristics and frequency of landslides from the base period (1979–2003) to the end of the 21st century (2075–2099) according to the climate change scenario and dynamical downscaling of rainfall data in Taiwan. We also discussed differences in geomorphologic and geologic conditions between the two catchments, relating to landslide characteristics at present and under future climates. These results not only provide a reference for areas having similar climate change trends over the world, but also open the possibility to analyze characteristics and frequency of landslides under climate change scenarios from triggering and predisposing factors.
Section snippets
Study area
The Tamsui River Basin covers the metropolitan regions of Taipei City, New Taipei City, Keelung City, and Taoyuan City (Fig. 1a). The Tamsui River is the third longest river in Taiwan, having a total length of about 159 km and a drainage area of 2726 km2. The total population in this metropolitan area is over seven million. The Dahan River and Xindian River flow toward the northeast and northwest respectively before successively merging into the Tamsui River, which is the major river in this
Landslide-area characteristics
The Forestry Bureau of Taiwan identifies the landslide inventory for all of Taiwan using satellite images in the first half (January to June) of each year (https://www.tgos.tw/TGOS/Web/Metadata/TGOS_MetaData_View.aspx?MID=CDBB21ACB7275BDB56E94ED8BA9E9546&SHOW_BACK_BUTTON=false&keyword=). The landslide inventory is automatically interpreted according to the operating standards set by the Forestry Bureau, and the minimum area of a landslide is 0.1 ha. According to the classification of landslides
Rainfall conditions and area characteristics of landslides
Table 1 shows the landslide-area characteristics of the two catchments after each typhoon event. For the Shihmen Reservoir catchment, Typhoon Aere in 2004 had the largest effect, resulting in 769 landslides, a maximum landslide area of 6.06 × 105 m2, a total landslide area of 4.89 km2, and a landslide-area ratio of 0.65%. The smallest effect was during Typhoon Sinlaku in 2008, which resulted in 251 landslides, a maximum landslide area of 9.36 × 104 m2, a total landslide area of 1.29 km2, and a
Conclusions
This study used landslides triggered by historical typhoon events to establish the empirical relationship between rainfall conditions and landslide characteristics and identified the frequency of landslides for two adjacent catchments in northern Taiwan. The Shihmen Reservoir catchment has high-relief topography, and it is highly prone to landslides at present. During past typhoon events, the average landslide-area ratio was 0.4%, and landslides mainly occurred under rainfall conditions of less
Acknowledgments
We would like to thank the theme C of the Program for Risk Information on Climate Change of Japan (SOUSEI-C) for providing the MRI-AGCM data. We are also very grateful to Dr. Ai-Ti Chen in the Department of Geosciences, National Taiwan University, discussing with us and providing many helpful comments. This research was supported by the Taiwan Climate Change Projection and Information Platform (TCCIP; MOST106-2621-M-865-001).
References (59)
A warning system for rainfall-induced shallow failures
Eng. Geol.
(2004)- et al.
The application of GIS-based logistic regression for landslide susceptibility mapping in the Kakuda-Yahiko Mountains, Central Japan
Geomorphology
(2005) - et al.
Estimation of sediment volume of debris flow caused by extreme rainfall in Taiwan
Eng. Geol.
(2011) - et al.
Rainfall intensity–duration conditions for mass movements in Taiwan
Prog. Earth Planet Sci.
(2015) - et al.
Application of radar data to modeling rainfall-induced landslides
Geomorphology
(2009) - et al.
Increase in basin sediment yield from landslides in storms following major seismic disturbance
Eng. Geol.
(2009) Deciphering the effect of climate change on landslide activity: a review
Geomorphology
(2010)- et al.
Representative rainfall thresholds for landslides in the Nepal Himalaya
Geomorphology
(2008) - et al.
Landslide hazard evaluation: a review of current techniques and their application in a multi-scale study, Central Italy
Geomorphology
(1999) A synthesis of the geologic evolution of Taiwan
Tectonophysics
(1986)