Elsevier

Engineering Geology

Volume 79, Issues 3–4, 11 July 2005, Pages 185-214
Engineering Geology

Small fast-moving flow-like landslides in volcanic deposits: The 2001 Las Colinas Landslide (El Salvador)

https://doi.org/10.1016/j.enggeo.2005.01.014Get rights and content

Abstract

Fast-moving landslides are one of the most significant dangers deriving from slope instabilities. Landslides involving large volumes can develop in rock or debris avalanches with extreme mobility and enormous destructiveness. Nevertheless, a relevant number of casualties and damages derive from small, fast-moving landslides with flow-like behaviour.

The Las Colinas landslide occurred at Santa Tecla (El Salvador, Central America) during a strong earthquake. It slid off the northern flank of the Bálsamo ridge, and resulted in almost 500 casualties and can be considered one of the most destructive landslides ever known. Earthquake shaking was amplified by the rock mass and the steep ridge topography.

We collected original geological, geomorphological and geophysical data in the Cordillera del Bálsamo area. The involved materials, ranging from lapilli to tuff layers of different strength, have been mapped and characterized.

Slope stability analyses have been performed both under static and dynamic conditions through limit equilibrium and finite element methods.

Hazard zonation for this type of landslides requires the forecast of the movement velocity and final deposition area. We used a fully two-dimensional FEM model to simulate landslide spreading downslope. The developed code allows the use of different constitutive models and yield rules with the possibility to model and study internal deformation of the landslide mass, as well material entrainment and deposition.

Introduction

Fast-moving, flow-like landslides are the most destructive among all slope instability phenomena and can be quite dangerous both for human lives and built areas. Flow-like landslides (e.g., rock and debris avalanches, debris flows) are complex phenomena involving large volumes of dry to saturated rock and/or debris, and/or earth (Varnes, 1978). These phenomena present different types of initial movement (fall, slide, etc.) followed by a flow-like movement of a disintegrating and fragmenting rock mass or debris, and an abnormal mobility (Hungr et al., 2001). During this last stage, the moving debris is able to flow around major obstacles, or destroy and remove them, or to be channelled by morphological features. The wide spectrum of observed behaviours and the occurrence of long runout distances and large areas covered by deposits are among the consequences of the natural variability of these phenomena.

Landslides involving large volumes can develop in rock or debris avalanches with extreme mobility and enormous destructiveness (e.g., Frank slide, Cruden and Krahn, 1978, Cruden and Hungr, 1986; Nevados Huascaran, Plafker and Ericksen, 1978; Vajont rockslide, Hendron and Patton, 1985; Val Pola, Crosta, 1991, Erismann and Abele, 2001). Nevertheless, a relevant number of casualties and damages derive from small fast-moving landslides (e.g., Gero-Barcone, 1762, 115 casualties; Aberfan, 1966, 144 casualties; Stava, 1985, 269 casualties; Sarno, 159 casualties). As a consequence, destructiveness is related to the size, typology and mobility of the phenomenon, and local conditions and must be quantified through an estimate or an evaluation of the expected or actual damages. This definition suggests the possibility that consequences of or damages induced by small landslides can be larger than those due to very large ones.

Landslides in volcanic materials are usually characterized by a higher mobility than those involving non-volcanic rocks (Hayashi and Self, 1992, Legros, 2001). The reasons for such a behaviour can be found in the granular nature of these materials, their collapsibility, degree of weathering and alteration, layering and grain size variability, low strength and variable water content, type of depositional process and conditions. Starting from a statistical analysis of 34 non-volcanic and 40 volcanic debris avalanche deposits, as well as 15 pyroclastic flow deposits, Hayashi and Self (1992) show that pyroclastic flow and volcanic debris avalanche data are statistically indistinguishable from each other but can be discriminated from non-volcanic debris avalanches. This seems to suggest that pyroclastic flows and volcanic debris avalanches develop under similar boundary conditions or that material properties characterize flow behaviour and evolution.

Following this idea, we prepared a dataset of flow-like landslides in different environments (subdivided in 132 subaerial non-volcanic, 50 volcanic, 47 submarine, 29 Martian landslides and 45 debris flows) starting from the one presented by Legros (2001). We observe (Fig. 1) that the maximum runout length for equal volumes is larger for volcanic landslides than for non-volcanic ones. This difference is slightly less evident when plotting landslide runout area versus landslide volume. This is true for small and medium-size landslides, and it becomes much more evident for very large volumes (>1 km3). This could be due to the larger size of volcanic landslides and to different characteristics of volcanic materials: grain size, weak cementing and low strength collapsible particles, easily fragmented and comminuted during shearing, and the presence of water or gas. On the contrary, very large non-volcanic landslides can include very stiff and strong lithologies with lower water content. Debris flows present a more evident tendency to spread over larger areas and this can be associated to their water content, that is usually being higher than in rock and soil slope instabilities. This behaviour is much more evident if we examine submarine landslides. These differences can be the results of various controlling factors.

Different data (geological, geomorphological, mechanical, etc.) are required to understand the preparatory and triggering causes of such slope instabilities and of their potential evolution, motion and impact on the environment and existing or planned structures. Data limited to the observation of the accumulation geometry and grain size are not sufficient for a full comprehension of the mechanisms. Pre-, co- and post-failure observations are essential for the analysis of such events and for the correct choice, and for the calibration and validation, of the adopted modelling approach. As a consequence, an incomplete and merely qualitative modelling is possible. This difficulty is clearly represented by the large number of theories and models in the literature (e.g., Heim, 1932, Shreve, 1968, Scheidegger, 1973, Abele, 1974, Habib, 1975, Hsu, 1975, Melosh, 1979, Davies, 1982, Hutter and Savage, 1988, Sassa, 1988, Campbell, 1989, Kilburn and Soresen, 1998, Erismann, 1986, Davies et al., 1999, Legros, 2001).

The definition of the hazard and terrain zonation are therefore complicated and require the determination of the size or volume of the unstable mass, the geometrical and geological and physical–mechanical constrains, the triggering probability of the initial failure, the change in behaviour, the geometry of the runout area and of the deposit, the intensity of the phenomenon along the path, the probability that a specific point in space will be reached (probability of reach) and the time needed to reach any specific point along the path.

In this paper, we analyse the 2001 Las Colinas landslide (Santa Tecla, El Salvador, Central America) that occurred during a major earthquake. This landslide, involving about 180,000 m3 of volcanic deposits, had a total runout of about 800 m, and resulted in about 500 casualties; it can be considered one of the most destructive landslides ever known.

We collected original geological, geomorphological and geophysical data in the Bálsamo Cordillera area, impacted by this and other slope instabilities. These data allowed us to reconstruct the geometry of the failure surface and the controlling factors. Slope stability analyses have been performed both under static and dynamic conditions through limit equilibrium and finite element methods. Hazard zonation of this type of landslides requires the forecast of the movement velocity and final deposition area. We used a fully two-dimensional FEM model to simulate landslide spreading downslope.

Section snippets

Geologic and tectonic settings

El Salvador is located on the Pacific coast of the Central America. The Central America tectonics are characterized by the interaction between the Caribbean and Cocos plates. The Cocos plate is converging, from the southwest towards the northeast, and is subducting, with a steep angle, beneath the Caribbean plate in the Middle American Trench (rate of convergence of 92 mm/year).

The main geological terrains cross El Salvador with a rough east trend parallel to the coast (Schmidt-Thomé, 1975,

Soil properties

Characterization of loose or weakly cemented volcanic soil and rock is a difficult task. Sampling techniques, sample treatment and conservation, and testing conditions make the characterization more difficult. Some authors have tested similar materials or samples coming from this same landslide. Suction pressures up to 796 kPa were measured in the Tierra Blanca deposits (Bernal Riosalido, 2001, Rolo et al., 2004) for moisture content ranging from 7.6 to 36%. Konagai et al. (2002) using a ring

Slope stability analyses

Detailed geometrical, stratigraphical and mechanical description of the Las Colinas slope, as well as of the earthquake triggering conditions, allowed us to perform slope stability analyses by applying classic limit equilibrium methods and finite element techniques.

No piezometric level or pore water pressure have been inserted in the models because of complete absence of water from field surveys and boreholes.

Modelling landslide flow

In Fig. 1, we observed that the Las Colinas landslide always occupies an abnormal position even for the category of landslides in volcanic materials and environments. As already suggested by many authors (Sousa and Voight, 1991, Hungr, 1995, Calvetti et al., 2000, Erismann and Abele, 2001, Legros, 2001, Chen and Lee, 2000, Crosta et al., 2004a, Crosta et al., 2004b), the analysis of flow-slide runout and spreading must be accurately evaluated and modelled quantitatively.

We performed some fully

Conclusions

The January 2001 El Salvador earthquake triggered many landslides some of which are peculiar in terms of typology and evolution. We investigated in detail the area of the Las Colinas landslide. The main Las Colinas landslide transformed into a flow-slide despite the low water content and the relatively low potential energy. Constrains to the landslide development were the weakly bonded material subject to grain crushing and fragmentation, the slope steepness, the strong earthquake shaking and

Acknowledgements

The authors acknowledge the Ministerio del Medio Ambiente, San Salvador, David A. Gómez Toledo, from Ministerio de Medio Ambiente/Operación y Soporte Informático, Eng. Alessandro Correra and Lotti Ass. for the support and data availability. Comments by the two reviewers, J. Bommer and E. Krinitzsky, and by O. Hungr strongly improved the paper. The research has been partially funded by the LESSLOSS EC Project, and by the Italian Ministry of Education, University and Research_(MIUR and FIRB).

References (58)

  • G.V. Burbach et al.

    Seismicity and tectonics of the subducted Cocos plate

    Journal of Geophysical Research

    (1984)
  • F. Calvetti et al.

    Numerical simulation of dry granular flows: from the reproduction of small-scale experiments to the prediction of rock avalanches

    Rivista Italiana di Geotecnica, AGI

    (2000)
  • C.S. Campbell

    Self-lubrication for long runout landslides

    Journal of Geology

    (1989)
  • R.R. Castro et al.

    S-waves response estimates using Nakamura's technique

    Bulletin of the Seismological Society of America

    (1996)
  • J.M. Cepeda et al.

    Strong-motion characteristics of January and February 2001 earthquakes in El Salvador

  • H. Chen et al.

    Numerical simulation of debris flows

    Canadian Geotechnical Journal

    (2000)
  • CIG (Centro de Investigaciones Geotécnicas,El Salvador). Historical Seismicity at El Salvador...
  • Crosta, G., 1991. Studio di movimenti in massa. Modellizzazione teorica e sperimentale con osservazioni e rilievi di...
  • G. Crosta et al.

    Numerical modeling of large landslide stability and runout

    Natural Hazard and Earth System Sciences

    (2003)
  • G. Crosta et al.

    Continuum numerical modelling of flow-like landslides

  • G.B. Crosta et al.

    The 2001 Las Colinas landslide, El Salvador: from geology to numerical modeling

  • D.M. Cruden et al.

    The debris of the Frank Slide and theories of rock-avalanche mobility

    Canadian Journal of Earth Sciences

    (1986)
  • T.R.H. Davies

    Spreading of rock avalanche debris by mechanical fluidization

    Rock Mechanics

    (1982)
  • T.R. Davies et al.

    A fragmentation-spreading model for long-runout rock avalanches

    Canadian Geotechnical Journal

    (1999)
  • R. Denlinger et al.

    Flow of variably fluidized granular masses across three-dimensional terrain: 2. Numerical predictions and experimental tests

    Journal of Geophysical Research

    (2001)
  • T.H. Erismann

    Flowing, rolling, bouncing, sliding: synopsis of basic mechanisms

    Acta Mechanica

    (1986)
  • T.H. Erismann et al.

    Dynamics of rockslides and rockfalls

    (2001)
  • S.G. Evans et al.

    The Las Colinas landslide, Santa Tecla: a highly destructive flowslide triggered by the January 13, 2001, El Salvador earthquake

  • W.H. Grebe

    Zur Geologie der altvulkanischen Gebirge in El Salvador (Mittelamerika)

    Geologisches Jahrbuch, Beihefte

    (1963)
  • Cited by (95)

    • Pre-failure suction-induced deformation to inform early warning of shallow landslides: Proof of concept at slope model scale

      2022, Engineering Geology
      Citation Excerpt :

      Rainfall-induced shallow landslides in coarse-grained volcanic fall deposits often evolve into debris flows causing significant damage and fatalities worldwide. Catastrophic events have been recorded in the last two decades in El Salvador, Hong-Kong, Indonesia, Italy, Japan, Mexico, Russia, Taiwan, and Venezuela (Fuchu et al., 1999; Cascini and Ferlisi, 2003; Olivares and Picarelli, 2003; Capra et al., 2003; Crosta et al., 2005; Chen et al., 2006; Pagano et al., 2010; Mizuyama and Egashira, 2010; Santo et al., 2012; Yamao et al., 2015; Chávez et al., 2016; Shimizu and Ono, 2016; Perov et al., 2017; Kusumawardani et al., 2017; Zhang et al., 2022). Volcanic soils are non-plastic and are characterised by high porosity maintained by the suction generated by the partially saturated state.

    • Analysis of slope fracturing under transient earthquake loading by random discrete element method

      2022, International Journal of Rock Mechanics and Mining Sciences
    View all citing articles on Scopus
    View full text