Catena 183 (2019) 104238 Contents lists available at ScienceDirect Catena journal homepage: www.elsevier.com/locate/catena Rainfall-triggered mass movements on steep loess slopes and their entrainment and distribution T ⁎ Wenzhao Guoa,b, Xiangzhou Xuc, Wenlong Wanga,b, , Yakun Liuc, Mingming Guoa, Zhiqiang Cuib a State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Northwest A&F University, Yangling, 712100, Shaanxi, China b Institute of Soil and Water Conservation, Chinese Academy of Sciences and Ministry of Water Resources, Yangling, 712100, Shaanxi, China c School of Hydraulic Engineering, Dalian University of Technology, Dalian 116024, China ARTICLE INFO ABSTRACT Keywords: Mass movements are predominant geomorphic processes on steep hillslopes. However, the mechanisms gov- Mass movement erning the erosion and entrainment of mass movements remain poorly understood. In this study, experiments on Soil erosion natural loess slopes were conducted to induce a series of mass movements under simulated rainfalls in the Entrainment Liudaogou Catchment on the Loess Plateau of China. A novel topography meter was used to observe random Rainfall simulation experiments mass movements. A total of 499 mass movements in 42 rainfall events and an average of 11 mass movements for Loess Plateau each rainfall event were observed. Three mass movement types were detected: landslides (67%) > mudflows (21%) > avalanches (12%). The volume of landslides dramatically increased through the entrainment of a wet gully bed material, and the volume of landslide mass was magnified by 29% on average through material entrainment. Based on the observed data, the probability of mass movement occurrences decreased with the increasing mass movement volume in a power-law relationship. The critical rainfall amount for mass movement − failure was approximately 25.6 mm at a rainfall intensity of 50 mm h 1. These results can serve as guides to mitigate geological hazards and assess erosion processes on steep loess slopes of the Loess Plateau. 1. Introduction et al. (2015a) suggested a systematic classification of mass movements, including landslides, mudflows, and avalanches. Mudflows have ob- Mass movement, also referred as gravity erosion or mass wasting, is vious flow performance and high water content compared with land- a slope failure on hillslopes. Mass movement is not only a natural ha- slides and avalanches (Guo et al., 2019). During erosion, the failure zard but also an important means of conveying sediments from slopes to block of an avalanche completely separates from the slope surface, channels in mountainous territories, thus severely affecting the struc- whereas that of a landslide slips down as a whole along a weak belt (Xu ture and function of ecosystems and societies (Keefer and Larsen, 2007; et al., 2015a). Zhang et al. (2012) found a close relationship between Qiu, 2014; Fuller et al., 2016; Xu et al., 2017). Therefore, under- the topographic attributes of post-landslide local surface and mass standing this phenomenon is necessary to implement hazard mitigation movement types. However, the responses of different movement types and control erosion. to rainfall characteristics and the distribution of mass movements have Rainfall is the most important triggering factor of mass movements received little attention despite their importance. on the Loess Plateau of China (Xu et al., 2017). Dry loess can sustain Previous studies have shown that the entrainment of initially static near-vertical slopes; however, loess can rapidly disaggregate when lo- materials can increase the mobility of avalanches (Mangeney et al., cally saturated by rainfall (Dai and Lee, 2002). Rainfall-triggered mass 2007). Accordingly, Breien et al. (2008) suggested that entrainment movements frequently occur on soil-mantled landforms (Minder et al., usually causes the debris flow to become increasingly erosive. Debris 2009). A field investigation shows that rainfall-triggered mass move- flow can markedly increase in size and speed when materials are en- ments only occur at a depth of < 2 m, corresponding to a surface layer trained from their beds. In addition, flow deposits from the underlying of completely saturated loess (Wang et al., 2015). erodible layer are difficult to distinguish when they are composed of the Mass movements include various types, each of which has specific same materials (Mangeney, 2011). Therefore, the quantitative mea- mechanisms and conditioning factors (Cruden and Varnes, 1996). Xu surement of entrainment volumes under field conditions becomes ⁎ Corresponding author at: State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Northwest Agriculture and Forestry University, Yangling, 712100, Shaanxi, China. E-mail addresses: [email protected] (W. Guo), [email protected] (W. Wang). https://doi.org/10.1016/j.catena.2019.104238 Received 25 October 2018; Received in revised form 19 August 2019; Accepted 23 August 2019 0341-8162/ © 2019 Elsevier B.V. All rights reserved. W. Guo, et al. Catena 183 (2019) 104238 (a) (b) (c) Rainfall simulators Gully Steep slope Mass movement Gully Mass movement T1 Fig. 1. Study area and sampling sites. (a) Location of the Liudaogou Catchment on the Loess Plateau of China; (b) Topography of typical mass movement; (c) Mass movement experiment in the Liudaogou Catchment. T1: Topography meter. Table 1 Table 2 Experimental summary of the initial slope landform and rainfall. Error between design rainfall intensity and experiment rainfall intensity in experiment F1. Test Lower slope configuration Rainfall − number Rainfall events Rainfall intensity (mm h 1) Error Height (m) Gradient (°) Intensity Duration Runs −1 (mm h ) (min) Experiment Design F1 1.0 70 50 60 6 1 47.4 50 5.2% F2 1.0 80 50 60 6 2 46.8 50 6.4% F3 1.0 60 50 60 6 3 48.8 50 2.4% F4 1.5 70 50 60 6 4 49.2 50 1.6% F5 1.5 80 50 60 6 5 50.4 50 −0.8% F6 1.5 60 50 60 6 6 47.4 50 5.2% F7 1.5 70 100 30 6 Average 48.3 50 3.3% complicated. Furthermore, the mechanisms that govern the growth of Table 3 landslides remain unclear, hampering efforts to assess natural hazards Soil physical properties. (Iverson et al., 2011; Mangeney, 2011). − Initial water content/ Dry density/g cm 3 Primary particle size (%) Recently, numerous scholars have conducted laboratory experi- % ments on mass movements to understand their processes and mechan- Clay/mm Silt/mm Sand/mm isms. For instance, Terajima et al. (2014) conducted a flume experiment < 0.002 0.002–0.05 > 0.05 to examine slope subsurface hydrology and found that seepage forces 9.3–13.6 1.44–1.66 2 30 68 affect the promotion of shallow landslide initiation. Xu et al. (2015b) tested the stability of different slope geometries and rainfalls to explore the triggering mechanisms of mass movements on a remolding slope. landslides. Kharismalatri et al. (2019) conducted a flume experiment to Yuliza et al. (2016) prepared a small-scale landslide experiments to evaluate factors for controlling sediment connectivity of landslide ma- determine the soil characteristics and water content that induce terials. However, these laboratory experiments used remolded soil, 2 W. Guo, et al. Catena 183 (2019) 104238 (a) (c) (b) (d) Fig. 2. Comparison of a mass movement and the three-dimensional vector model. (a) Crevice was created and expanded, which indicated that a mass movement was occurring. (b) Failure block was fragmentized and stacked in the main channel. (c) and (d) are 3D surface models reconstructed with ArcGIS corresponding to (a) and (b), respectively. Table 4 Summary information on mass movement in experiments F1–F7. Test number Number of mass movements Amount of mass movements/103 cm3 Avalanche Landslide Mudflow Total Avalanche Landslide Mudflow Total F1 12 41 5 58 11.0 53.3 4.2 68.6 F2 7 14 1 22 3.7 8.4 0.4 12.5 F3 1 3 3 7 0.3 9.5 1.5 11.4 F4 7 126 45 178 5.3 173.4 40.9 219.6 F5 10 89 18 117 5.6 92.2 13.2 111.0 F6 9 56 32 97 6.8 56.1 20.6 83.6 F7 16 4 0 20 49.4 3.1 0.0 52.5 Summation 62 333 104 499 82.2 396.1 80.7 559.0 Percentage 12% 67% 21% 100% 15% 71% 14% 100% destroyed the mechanical structure of the original soil, and could not contribution of avalanches, landslides, and mudflows to the amounts of truly reflect the changes of the stress field on natural slopes. Further- mass movements. In addition, the distribution of mass movements in more, few experiments have focused on the distribution of mass terms of failure volume and rainfall was explored. Different from la- movements in terms of failure volume and rainfall. boratory experiments, the experiment on the segment of unscaled rea- Therefore, this study conducted a series of mass movement experi- lity on natural loess slopes retains the loess scale and natural char- ments on segments of unscaled reality on natural loess slopes on the acteristics (such as the internal structure and vertical joints) while Loess Plateau of China. The objective of this study was to investigate controlling for the location and timing of mass movement occurrence the characteristics and distribution of mass movements and the (Guo et al., 2019). This characteristic in our study is an important 3 W. Guo, et al. Catena 183 (2019) 104238 80% Number landslides, mudflows, and avalanches that contribute large amounts of 71% Volume 67% sediment yield by conveying soil into valleys. 70% 60% 3. Materials and methods 50% To analyze the failure mechanism of mass movements, a series of 40% experiments (F1–F7) were conducted on natural loess slopes in the Liudaogou Catchment of Shenmu County in the summer of 2014 Percentage 30% (Fig.