Keywords

1 Introduction

Large volcanic landslides and debris avalanches (>0.1 km3) are significant volcanic hazards, being directly or indirectly responsible for more than 20,000 fatalities since 1600 AD (Auker et al. 2013). For example, the recent landslide at Anak Krakatau on 22 December 2018 and subsequent tsunami caused at least 431 deaths (Williams et al. 2019). Since the Mount St. Helens event on 18 May 1980 (Voight et al. 1981), this phenomenon has become a major topic of investigation in volcanology. Hundreds of scientific articles have been published on the subject in the last four decades, along with several book chapters. Most of them use a relatively close scientific terminology but some terms such as “collapse”, “caldera” and “avalanche” are ambiguous because they are also used for other volcanic phenomena that might lead to misinterpretations. Other terms such as “rockslide” and “lateral collapse” are less used inducing poor benchmarking. Nomenclature is critical in a continually evolving science such as volcanology because each scientific term must convey a message that is coherent with either the characteristics of the feature (i.e., descriptive terms) or the current understanding of the physical processes (i.e., concepts). It is also relevant, in a highly productive science, to precisely define keywords to help scientists in comparative analysis. We think that now, 40 years after the paradigm shift that was the Mount St. Helens event, is an excellent time to take a hard look at the vocabulary used and propose, if necessary, changes to our language usage. We also propose a descriptive strategy using adequate terms to make field observations less subjective and enhance comparisons between different structures and deposits worldwide. In this chapter, we investigate the definition of the phenomenon before reviewing its main physical/measurable features that are the landslide scar and the debris avalanche deposit.

2 Definitions of the Phenomena

This part is mainly based on the work of Siebert (1984), Glicken (1991), and Ui et al. (2000) but also incorporate other works that deal with more specific aspects. We first present the direct and indirect observations made on the phenomena and then discuss the different terms used to define them. The most common synonyms are also presented, but the terms which best depict the features of the processes should be adopted. In the literature, most of the definitions include two parts that describe: (1) the initiation phase, and (2) the transport phase.

2.1 The Initiation Phase

2.1.1 Eyewitness, Time-Scale, Dimension, and Definition

With about 5–6 occurrences of large volcanic landslides per century since the 1600 s (Siebert and Roverato 2020), this phenomenon is scarce at a human life-time scale, so the probability of having eyewitness and measurements on an event is very low. Although, in the last four decades, volcanologists have been able to make observations with modern techniques at three occasions, namely during the Mount St Helens (USA) eruption in 1980 (Voight et al. 1981), the Soufrière Hills (Montserrat) eruption in 1997 (Voight et al. 2002), and the Anak Krakatau (Indonesia) eruption in 2018 (Williams et al. 2019). Unfortunately, both the Soufrière Hills and Anak Krakatau events occurred at night, so there are no direct eyewitnesses of these events. During the Mount St Helens event, the initiation phase was described by Stoffel and Stoffel (1980) as follows: “everything north of a line drawn east–west across the northern side of the summit crater began to move as one gigantic mass”. In this event, the initiation consisted of three retrogressive and consecutive slides that removed 2.3 km3 from the volcanic edifice (Voight et al. 1981). This description fits quite well with the observations made at the recent Anak Krakatau flank failure in which at least two failure planes were identified on a Sentinel-1A SAR image 8 h after the event (Williams et al. 2019). It is also supported by analogue models showing coherent landslide behaviour and reproducing typical landslide scar and debris avalanche hummocky topography (Shea and van Wyk de Vries 2008; Andrade and van Wyk de Vries 2010).

The study of ground deformation before the event at Mount St Helens suggests that bulging on the north flank began between March 19 and 31, and had a relatively steady rate of 1.5–2.5 m per days (~2 to 3 × 10–5 m/s) until May 18 (Lipman et al. 1981). On May 18, only 26 s after a magnitude 5.1 earthquake, the first slide had moved about 700 m, reaching a velocity of 40 m/s (Voight et al. 1981). The acceleration of the volcanic portion is therefore estimated, if the earthquake is the triggering mechanism, between 1.0 and 1.5 m/s2. To separate the long-term volcanic deformation and the rapid mass movement, we can use the acceleration (~0.1 g) as a quantitative indication of the beginning of the landslide. If continuous monitoring of an active volcano is performed through real-time geophysical methods, such as real-time GPS or continuous Electronic Distance Measurements, such acceleration can be used as an early warning triggering mechanism.

Landslide volume can be of any size, ranging from 10–5 to 104 km3 (Legros 2002; Crosta et al. 2005). Observations made in non-volcanic environments show that events smaller than one million m3 behave as granular flows with travel distances that reflect a frictional flow regime (McSaveney et al. 2000), but this distinction is made for the transport phase, not the initiation phase. For volcanic events, some authors make a distinction between small (<0.1 km3) and large or mega (>0.1 km3) events that are usually related to the triggering mechanisms (Siebert 2002; Yoshida 2016). In this chapter, we focus our attention on the large events, but most of the terminology can also be used for smaller events.

Accordingly, the initiation phase can be described as the translation, rapid and mostly horizontal, of one (or more) portion(s) composed of multiple volcanic units over a slide surface produced by a failure in the volcanic edifice (Fig. 1; Table 1). This definition excludes phenomena such as vertical collapses (such as pit and caldera collapses), rock falls, and long-term volcanic deformation such as creeping or spreading.

Fig. 1
figure 1

Modified from Bernard (2008)

Sketch of a volcanic landslide and debris avalanche with a the initial state; b the initiation; c the transport; and d the final state.

Full size image
Table 1 Definition and terms used to describe large volcanic debris avalanches
Full size table

2.1.2 Terminology

The most popular term used to name the initiation phase is probably “sector collapse” (Boudon et al. 1987; Kokelaar and Romagnoli 1995; Richards and Villeneuve 2001; Ponomareva et al. 2006; Bernard et al. 2008; Kervyn et al. 2008; Delcamp et al. 2016) which highlights the fact that the failure removes away “a sector” of the volcano. Nevertheless, this term is ambiguous as it can be confused with partial vertical collapse. In order to distinguish this rupture from a purely vertical collapse and to specify that the mass movement is toward one particular direction, several authors proposed the terms “lateral collapse” (Marti et al. 1997; Day et al. 1999; Tibaldi et al. 2006; Romagnoli et al. 2009) and “flank collapse” (Vincent et al. 1989; Day et al. 1997; Elsworth and day 1999; Leyrit 2000; Le Friant et al. 2004; Arce et al. 2008; Andrade and van Wyk de Vries 2010; Macías et al. 2010). Capra et al. (2002) and Scott et al. (2005) distinguish “sector collapse” from “flank collapse”, suggesting that the latter is typically smaller and does not involve the volcano summit and core. Some authors prefer the term “rockslide” because it takes into account the description of the rupture. This term was first used to describe the Mount St. Helens event (Voight et al. 1981) and has also commonly been used ever since (McEwen et al. 1989; Cruden and Lu 1992; Glicken 1996; Shea et al. 2008; Paguican et al. 2014). However, the definition of a rockslide, as taken from the slope movements classification (Varnes 1978; Hungr et al. 2014), implies the slide of (multiple) rock units, but a volcanic edifice is made of both massive rock units (lavas) and loose volcaniclastic and epiclastic formations. The terms “slope failure”, “flank failure”, “slope instability”, and “flank instability” have been frequently used for volcanic islands or submarine events (Johnson 1987; McGuire 1996; Hürlimann et al. 2000; Lipman et al. 2003; Coombs et al. 2007; Paris et al. 2011; Williams et al. 2019), but could relate to longer-term mass wasting processes such as creeping. We considered that the best term describing the initiation phase is probably “volcanic landslide” (Duffield et al. 1982; Stoopes and Sheridan 1992; Iverson 1995; Carracedo et al. 1999; Legros et al. 2000; Hürlimann et al. 2000; Watt et al. 2009; Delcamp et al. 2017) since it includes the dynamic aspect (slide) and the material involved (volcanic land) without being too specific. The rupture can involve or not the volcano summit and may be caused by different edifice weakening processes and triggering mechanisms. Volcanic landslides might also provoke various secondary phenomena such as directed blast, debris flow, or tsunami (Siebert et al. 1987). In the field, the evidence of a volcanic landslide is a breached depression with steep walls at the source (Fig. 1).

2.2 The Transport Phase

2.2.1 Eyewitness, Time-Scale, Dimension, and Definition

Very few direct observations exist on the transport phase of this phenomenon. During the Mount St. Helens event, this stage was hidden by the lateral blast cloud created by the rapid decompression and explosion of the cryptodome (Stoffel and Stoffel 1980). Emplacement times of less than 3 min and about 10 min have been estimated for Soufrière Hills (runout 4 km) and Mount St Helens (runout 26 km) events, respectively. Velocity values obtained by measurements or calculations are generally between 20 and 100 m/s (Voight et al. 1981, 2002; Siebert et al. 1987). These mass movements can travel a long distance (up to 10 s of km) and affect considerable area (up to 100 s of km2) (Siebert 1984; Ui et al. 2000). Transport mechanisms are generally studied through deposit analysis, but the emplacement mode remains highly controversial (Ui et al. 2000). However, it is possible to give a general description of the transport phase as a rapid water-unsaturated gravity-driven mass movement of multiple volcanic units (Fig. 1; Table 1). This definition excludes other volcanogenic flows such as lava flows, pyroclastic density currents, and lahars.

2.2.2 Terminology

Most of the authors use the term “debris avalanche” for the transport phase (Crandell et al. 1984; Glicken 1991; Hayashi and Self 1992; Ui et al. 2000; van Wyk de Vries and Delcamp 2015). This term may sometimes be employed to imply a certain degree of the incoherence of a flow made of debris (Fisher et al. 1987). There is widespread evidence in the deposits showing that a significant part of the mass moves in a nearly coherent manner, for example, the presence of toreva (Belousov et al. 1999; van Wyk de Vries et al. 2001; Clavero et al. 2002; Roa et al. 2003; Bernard et al. 2011; see Sect. 4 for definition) and large panels with original stratigraphy preserved even several tens of kilometers away from the source (Glicken 1991; Takarada et al. 1999; Bernard 2008; Fig. 2). To illustrate these characteristics, some authors prefer the terms “rockslide avalanche” (Voight et al. 1981; Shea et al. 2008) and “rockslide-debris avalanche” (Glicken 1996; Paguican et al. 2014). Studies of those deposits show that a layer, containing elements from both the landslide source and the substratum, is formed at the base of the moving mass due to intense shearing and mixing (Schneider and Fisher 1998; Takarada et al. 1999; Bernard et al. 2008) so the basal contact is mainly not a slide surface. Exceptions exist, such as the 8 ka Parinacota deposit (Chile), where some blocks have slid several 10 s of meters away from the main deposit body due to a slippery fluvio-lacustrine substratum (Clavero et al. 2002). To have a consistent terminology, we propose to use the term “volcanic debris avalanche” that best describes the transport phase. In the field, the evidence of a debris avalanche is a hummocky deposit at the foot of the volcano (Fig. 1).

Fig. 2
figure 2

Descriptive elements and dimensions for a volcanic landslide scar. a Cross-sections and plan view with the main descriptive terms and geometrical parameters defined in Table 2. b Mayuyama landslide scar, Japan

Full size image

3 Descriptive Strategy for the Volcanic Landslide Scar

3.1 Terminology

In the literature, the most common morphological term used to describe this structure is “caldera,” preceded by a genetic qualifier such “sector collapse” or “avalanche” to distinguish it from collapse caldera (Siebert 1984). Nevertheless, “caldera” derives from the Spanish word “Caldero” (cauldron in English, caldeirão in Portuguese) and depicts a sub-circular depression with vertical walls and an almost flat, horizontal floor. This term accurately illustrates structures formed by vertical collapse due to large ignimbrite eruption (e.g., Crater Lake caldera, USA) or large basaltic lava outpouring (e.g., Mokuaweoweo caldera at Mauna Loa, Hawaii) but does not fit the description of the volcanic landslide rupture zone that is opened in the direction of the landslide. The terms “horseshoe-shaped” or “U-shaped caldera” have been commonly used to distinguish the volcanic landslide source from the other vertical structures (Siebert et al. 1987, van Wyk de Vries and Delcamp 2015). Some authors prefer the term “collapse amphitheater” (Ui et al. 2000; Coombs et al. 2007) and use it to describe some of the most typical rupture structure (e.g., Mount St Helens, USA) but “amphitheater” might not be an adequate term as it refers to a circular or elliptical open-air venue (e.g., Colosseum amphitheater in Rome). In the literature concerning non-volcanic landslides, the terms “scar” and “scarp” have been used interchangeably but some authors (Strasser and Schlunegger 2005; Yoshimatsu and Abe 2006) propose that a scar is a mark left by a rupture whereas a scarp is a steep slope formed by various processes such as faulting, erosion or deposition (e.g., scarps form a high viscosity lava front). The term “scar” has already been used in the volcanic context (Oehler et al. 2008; Shea et al. 2008; van Wyk de Vries and Delcamp 2015), and we consider that the best term to depict the source should be “volcanic landslide scar.” The depression formed during a volcanic landslide can thus be described using three geomorphological elements (Fig. 2): (1) the “wall”, which corresponds to the steep limit of the scar; (2) the “floor”, which stands for the relatively flat, but not necessarily horizontal, the interior of the depression; and (3) the “aperture”, which is the region between the lowermost points of the wall. We may differentiate the “head-wall”, that faces the aperture of the depression (Siebert 1984), and the “side-walls”, which are more parallel to the general landslide direction. Thus, the aperture can be defined as the section between the side-walls’ extremity.

3.2 Metrics

The morphology of the rupture zone can be described using common metrics presented in Fig. 2a and defined in Table 2 (Siebert 1984; Legros 2002; Bernard 2008; Dufresne 2009). The scar evolves through time due to subsequent erosion of the wall and to filling of the depression by the material of different origin (e.g., post-landslide volcanic activity, glacial erosion, rockfall from the walls) so the reconstruction of the original scar size might be the result of an interpretative work based on geologic and topographic data. Thus, any parameter estimated or measured after interpretation must be moderated with a margin of error (Siebert et al. 2004; Yoshida and Sugai 2007; Bernard et al. 2008). To compare scars of different sizes, we propose to complete their description with dimensionless parameters such as elongation, aspect ratio, and closure factors (Table 2).

Table 2 Definition of the quantitative parameters for the volcanic landslide scar
Full size table

3.3 Morphology

The description of the landslide scar is generally done using plan view and profile representations. In plan view the terms “horseshoe”, “amphitheater”, and “U-shaped” are typical (Siebert 1984; Belousov et al. 1999; Ui et al. 2000; Riggs and Carrasco-Nunez 2004) completed by terms like “deep-seated” and “thin-skinned” in cross-sections to denote the depth of the failure (Siebert 1984; Petley and Allison 1997; Coombs et al. 2007). Horseshoe scars have an aperture width smaller than their maximum width (e.g., Mayuyama scar; Siebert et al. 1987). Amphitheater scars are described as semi-circular structures. However, this term might not be etymologically correct (see Siebert and Roverato 2020, this volume) and, therefore, we propose to use “semi-circular” as a descriptive term for such structure that is also characterized by its width being close to twice its length (e.g., Duau scar; Johnson 1987). U-shaped scars have a semi-circular head-wall and parallel side-walls, and their length is generally larger than their width (e.g., Reventador 19 ka scar, Bernard and Andrade 2019). Triangular scars in plan view, such as the one at Socompa (van Wyk de Vries et al. 2001), are scarcer with clearly divergent linear side-walls. Deep-seated structures, such as Mount St. Helens landslide scar, generally have steep high walls and an almost horizontal floor while thin-skinned structures, such as Mayuyama one (Ozeki et al. 2005), have much steeper floors and smaller walls (Fig. 3). Irregular scars, such as Guagua Pichincha (Robin et al. 2010), are also frequent, in particular in compound volcanoes and volcanic complexes.

Fig. 3
figure 3

Common shape of volcanic landslide scar in plan view and cross-section

Full size image

The analysis of the scar shapes could be used to understand the long-term weakening process, short-term instability, and triggering mechanisms. Deep-seated scars are frequently associated with deep processes such as cryptodome intrusions (Siebert 2002) or highly altered volcano cores (Bernard 2008), whereas thin-skinned scars are generally associated with external processes such as heavy rainfall (Kerle and van Wyk de Vries 2001). Triangular and large U-shaped scars can be created when the failure initiates in or propagates to the volcanic substratum/basement (van Wyk de Vries et al. 2001), whereas horseshoe scars are usually associated with eruptions (Voight et al. 1981). Nested scars associated with recurrent volcanic landslides are also a common feature that can influence the shape of the scar (Belousov et al. 1999; Bernard and Andrade 2019). Nevertheless, no systematic relationship has yet been presented between the shape of the landslide scar and its origin. Such a study should be considered as highly relevant in the future.

3.4 Geological Elements and Distinction from Other Volcanic Depressions

To study the origin of the landslide, a geological description of the scar is necessary. Structural elements such as dikes and faults intersecting the walls can provide insights into local stress regime and might be related to the direction of aperture (Siebert 1984; Lagmay et al. 2000; Paguican et al. 2012; Andrade et al. 2018). Long-term weakening processes and instability factors can also be assessed when studying the nature and physical and chemical alteration of the wall units (Voight et al. 2002; Oehler et al. 2008; Roverato et al. 2020, this volume).

Volcanic landslide scars can be distinguished from most other volcanic depressions due to their shape and their dimensions (Siebert 1984). Evidence of an aperture is a major criterion to differentiate them from collapse calderas or craters, but sometimes those can be breached by eruptions or erosion processes. Nonetheless, those secondary apertures tend to be much narrower compared to landslide scar apertures and could be discriminated using the closure factor (Table 2). The aperture can also be hidden by following eruptive activity, and then more geological and geophysical investigations are required. Glacial erosion structures, such as glacial cirques, can also look like landslide scars but are generally shallower and filled with glacial deposits (Karátson et al. 1999). Fluvial erosion can be assessed through the analysis of the shape and depth of the fluvial network (Paris and Carracedo 2001).

4 Descriptive Strategy for the Volcanic Debris Avalanche Deposit

4.1 Terminology of the Fundamental Elements

The volcanic debris avalanche deposits (VDAD) are epiclastic, unsorted, heterogeneous, and heterometric breccias composed of pieces of the edifice source and the transport path (Siebert 1984; Glicken 1991; Ui et al. 2000; Leyrit 2000). Such deposits have multiple facies and complex structures, hence the terminology used to describe them has become somewhat confusing. The component terminology in volcanic or non-volcanic debris avalanche deposits is a controversial subject that has led to an extensive literature (Ui 1983; Ui and Glicken 1986; Glicken 1991; Palmer et al. 1991; Yarnold 1993; Friedmann 1997; Belousov et al. 1999; Takarada et al. 1999; Nehlig et al. 2001; Shea et al. 2008).

Glicken (1991) distinguishes two kinds of fundamental elements: clasts and debris avalanche blocks (Fig. 4). This distinction has been used almost unanimously ever since. A clast is defined as “rock of any size that would not break if passed through a sieve or immersed in water”. Except for particularly fragile particles such as prismatically jointed blocks, this term is unequivocal. A debris avalanche block (DAB) is defined as “an unconsolidated (or poorly consolidated) piece of the old mountain transported(?) to its place of deposition”. The definition mentions that all DABs come from the source-edifice, but it has been undoubtedly verified that large coherent portions of VDADs are made of substratum material (Palmer et al. 1991; Belousov et al. 1999; van Wyk de Vries et al. 2001; Clavero et al. 2004). This term can be employed in the sense of Glicken (1991) terminology, but it should be mentioned that DABs may also be derived either from the transport path or the edifice substratum. DABs are commonly found in other deposits such as secondary slides and cohesive lahars (Capra and Macías 2000; Bernard et al. 2009), which helps to assess their primary source.

Fig. 4
figure 4

Modified from Bernard (2008)

Volcanic debris avalanche deposit descriptive elements. Photo a and drawing b of an outcrop from the 2.4 Ma Mont Dore VDAD, France.

Full size image

When describing large pieces of the edifice source in the deposit, some authors use the terms “megablock” or “megapanel” that can preserve their original bedding (Ui 1983; Mehl and Schmincke 1999). The minimum size and the dimension (great axis length, surface, and volume) used to describe them can vary from one author to another. These terms can look appealing, but they are redundant with the term DAB, which has no upper size limit.

The term “matrix” should be employed in its sedimentological meaning that refers to the fine grains between the larger particles (Mehl and Schmincke 1999). Glicken (1996) distinguishes the intraclast matrix (inside one particular DAB, often monolithologic) and the interclast matrix. According to the definitions of clast and DAB mentioned above, the terms “intraclast” and “interclast” should be replaced by “intrablock” and “interblock”, respectively (Fig. 4). It might be of interest to distinguish the intrablock matrix present at the source and the intrablock matrix produced during transport through shattering of the debris avalanche blocks or shearing at the basal contact. Although such distinction is complicated for reworked deposits with pre-existent matrix (e.g., in autobreccia, pyroclastic, and epiclastic deposits), it could be easily done for initially massive units such as lava flows. The study of the interblock matrix and its components could help to understand the evolution of the avalanche downstream and bulking processes.

4.2 Deposit Facies

VDADs have the particularity of presenting a greater diversity of facies than any other volcanic deposit. Their description is fundamental in deciphering the complexity of this phenomenon. The variety of rock types, structures, and textures observed in VDADs is responsible for the multiplication of facies terms in the literature: block facies (Ui 1983), matrix facies (Ui and Glicken 1986), mixed facies (Glicken 1991), marginal facies (Palmer et al. 1991), sheared facies (Mehl and Schmincke 1999), bulldozer facies (Belousov et al. 1999), basal facies (Ui et al. 2000), hybrid facies (Roverato et al. 2011) and others (see Dufresne et al. 2020, this volume). Despite several attempts to build coherent terminologies, it remains difficult to compare the facies observations made in different deposits by different authors. Facies must be described more precisely to get meaningful information on the landslide origin and the flow mechanisms. A facies is generally defined by a unique character that distinguishes one rock-body from another (Cas and Wright 1987). In the following sections, we propose to split the facies analysis of VDADs into two stages that provide different information on the event.

4.2.1 First-Order Classification

Based principally on the work of Glicken (1991, 1996), we propose to use a first-order classification that allows distinguishing the mostly intact material and the newly formed material, which would be the block facies and mixed facies, respectively. Although there is a wide range of intermediate facies between them, these terms are the most widely used and are useful for mapping purposes (e.g., Bernard et al. 2008). To improve the relevance of this classification, we propose to add an adjective to the block facies to separate the material coming from the volcanic edifice from the material coming from the substratum if possible (Bernard 2008).

The edifice-block facies (EBF) is almost exclusively made of material from the landslide source and is comparable to the original definition of the block facies in Glicken (1991). The EBF is made of the apposition of DABs from ancient layers of lava flows, pyroclastic density current, and fall deposits, and even epiclastic material such as debris-flow and moraines deposits (Fig. 5a). The result is a coherent unconsolidated or poorly consolidated facies with low interblock matrix that can be stratified or not, depending on its origin and the exposure size. It can be highly deformed or just a little shattered. Jigsaw cracking is a common feature in this facies. It is generally polylithological but can appear monolithological, depending on the outcrop size.

Fig. 5
figure 5

Typical facies of volcanic debris avalanche deposits. a edifice-block facies (EBF) with jigsaw cracks from the 4.5 ka Cotopaxi VDAD, Ecuador. b substratum-block facies (SBF) made of Toya ignimbrite and associated reworked parts from the 20 ka Zenkoji VDAD, Japan. c mixed facies (MF) with a prismatic jointed block from the 60–65 ka Chimborazo VDAD, Ecuador. d, e Combination of facies from the > 30 ka Imbabura VDAD, Ecuador

Full size image

The substratum-block facies (SBF) is made of DABs coming from the transport path or the edifice basement that have been incorporated almost intact in the debris avalanche (Fig. 5b). The amount of interblock matrix is generally higher compared to the EBF. This facies is partially similar to the bulldozer facies of Belousov et al. (1999). It generally appears at the base or the front of the VDAD but can also be fluidized and incorporated up to the top of the deposit (Bernard et al. 2008). The SBF can be made of sedimentary, volcaniclastic, and epiclastic deposits, even soil. In some cases, it can also include basement material such as altered leucogranite (Schneider and Fisher 1998). The coherence of this facies depends on the material incorporated. The degree of deformation in the SBF is extremely variable, but folding and faulting are particularly common. The distinction between SBF and EBF facies can be difficult when the former is made of volcaniclastic or epiclastic material. Good knowledge of the edifice and substratum nature can help to distinguish between them. In the case of failures due to a weak substratum, the SBF can represent most of the VDAD (van Wyk de Vries et al. 2001).

The mixed facies (MF) was defined by Glicken (1991) and referred to the completely mixed part of the VDAD, where the material has lost its original primary structures (e.g., stratifications). MF is mostly made of interblock matrix and is generally highly polylithological, light-brown colored, sometimes consolidated, unsorted, and unstratified (Fig. 5c). It may contain juvenile material and wood fragments. MF is highly heterometric. It may be possible to distinguish between the elements coming from the source edifice and the substratum when there are distinct differences such as geochemical character and clast shape (e.g., pebbles are more likely coming from the substratum). Due to their shared characteristics, MF can be confounded with landslide-triggered cohesive lahars. Detailed investigation of the internal structures, such as trapped air bubbles, can help distinguish between them (Bernard et al. 2009).

A single outcrop might expose all the facies at once (Fig. 5d, e). The quantification of each facies helps to estimate the original landslide volume, the amount of erosion, and total volume increase during transport.

4.2.2 Lithology

This part is based on the work of Cas and Wright (1987) on facies analysis. The lithological description may include the physical constituents (lava, pyroclastic, autoclastic, epiclastic, and non-volcanic; Fig. 6), their composition (geochemical, mineralogical, and petrological characters) and their texture (grain size, sorting, rounding, shape, and fabric). The identification of physical constituent can sometimes be complicated in high-deformation zones. Nonetheless, it is required for any textural analysis intended to explore the flow dynamics because the origin of the material and its primary texture influence the final texture. The physical constituent analysis also permits to study the origin of the landslide. The composition analysis is particularly interesting in two cases. If the source zone is still exposed, a comparison between the compositional maps of the scar and the deposit can provide valuable information on the emplacement processes (Kelfoun et al. 2008). If the source region is hidden, the composition analysis can help to understand the development of the volcano before the failure. A complete lithological description gives information on the volcano history, the pre-collapse state, the triggering mechanism, and the transport path nature.

Fig. 6
figure 6

Modified from Bernard (2008). MF: mixed facies

Photo a and sketch b of an outcrop of the >30 ka Imbabura VDAD with facies lithologies.

Full size image

4.3 Deposit Structures

The most typical structures used to identify a VDAD are jigsaw cracks (Fig. 5a) and hummocky topography (Ui 1983; Siebert 1984). These features also occur in non-volcanic rock avalanches (Dufresne et al. 2016 and references therein). Nonetheless, the variety of observed structures is much more comprehensive (Glicken 1991). In VDADs, there are innate structures such as magmatic joint, stratification, and prismatically jointed blocks. Structures existed in the volcanic edifice, before the failure, such as fractures due to tectonic stress or mechanical weathering. Additionally, there are structures acquired during the event, and finally, others developed after the deposition, due to faulting or reworking. It is essential to distinguish among those structures during fieldwork. The literature on VDAD structures is extensive and is presented here in three complementary sections: the basal contact, the internal structures, and the surface morphology.

4.3.1 Basal Structures

Even if debris avalanches, either volcanic or non-volcanic, are highly erosive (Yarnold 1993), basal contact descriptions are scarce in the literature. Some authors use the term “sole” to describe a particular space placed between the unaffected substratum and the VDAD body (Schneider and Fisher 1998). This space seems to concentrate the most extreme features from flat contact without particular structures (Clavero et al. 2002; Fig. 7d), highly deformed DABs (Fig. 7a, b), up to the occurrence of frictionite (Legros et al. 2000). Abrasion features such as striae and channels are described (Schneider and Fisher 1998; Mehl and Schmincke 1999) as well as deformation structures such as boudinage, folding, and shearing of the substratum (Clavero et al. 2004; Bernard et al. 2008). It is possible to observe wood pieces incorporated in the VDAD and oriented slightly parallel (Belousov 1995; Takarada et al. 1999) or perpendicular (Bernard et al. 2009) to the flow direction. Substratum injections in the deposit body are also observed (Bernard and van Wyk de Vries 2010; Fig. 7c). Fine-grained injections descending from the VDAD down into the substratum are more rarely observed (Schneider and Fisher 1998).

Fig. 7
figure 7

Examples of basal contacts. a, b sheared basal contact in the 60–65 ka Chimborazo VDAD, Ecuador (modified from Bernard et al. 2008). c undulating contact with substratum injection (pumice) in the 2.4 Ma Mont Dore VDAD, France. d almost planar contact in the 8 ka Parinacota VDAD, Chile

Full size image

It is important to note that most of the VDAD particular facies (e.g., sheared facies, basal facies, and bulldozer facies) are related to the incorporation of the substratum in the deposit (Friedmann 1997; Belousov et al. 1999) which can contribute to a large proportion of the total deposit volume (Palmer et al. 1991; Clavero et al. 2004). Sometimes, the whole base of the VDAD is composed of substrata incorporated during failures, such as Socompa and Mombacho examples (van Wyk de Vries et al. 2001; Shea et al. 2008).

4.3.2 Internal Structures

The most common and most described internal structure is the jigsaw cracking (Ui 1983; Fig. 8a) that corresponds to a chaotic fracture network characterized by small displacements of the resulting fragments and associated with the dilation without disaggregation of a rock unit initially not fractured (Glicken 1996). These fractures are observed at all scales (Komorowski et al. 1991; Belousov et al. 1999; Roverato et al. 2015), and they are present from the proximal to the distal part of the VDAD and can change from jigsaw crack (cracks without intrablock matrix) to jigsaw fit (open cracks with intrablock matrix) (Ui and Glicken 1986; Ui et al. 1986). It is common to observe the superimposition of jigsaw cracking over innate or acquired fractures, such as thermal and tectonic fractures, but it is quite easy to differentiate among them because jigsaw cracking is generally much more random.

Fig. 8
figure 8

Examples of internal structures. a jigsaw cracking and mixed facies injection in a DAB in the >30 ka Imbabura VDAD, Ecuador. b mixed facies and fluidized substratum injections in the 60–65 ka Chimborazo VDAD, Ecuador. c faulting in lacustrine deposits incorporated in the 30 ka Tungurahua VDAD, Ecuador. d boudinage of pyroclastic units in the >30 ka Imbabura VDAD, Ecuador. e mingling structure between different colour matrices in the 2.4 Ma Mont Dore VDAD, France. (f) vortex structure of a substratum DAB in the 2.4 Ma Mont Dore VDAD, France

Full size image

There are many more internal structures in VDADs, including the following non-exhaustive list of the most common structures. These structures can also help to distinguish VDAD from other volcanoclastic or epiclastic deposits.

  • Mixed facies and fluidized substratum injections in DABs (Fig. 8a, b);

  • Faulting or boudinage of stratified units due to stretching of the avalanche mass (Fig. 8c, d);

  • Mingling between different color matrix/facies (Fig. 8e);

  • Vortex structure of DABs due to shearing (Fig. 8f);

  • Impact features on block surface associated with the vibration and collision of the blocks during transport (Clavero et al. 2002);

  • Hackly texture on millimeter size grains that are similar to microscopic-scale jigsaw cracks (Komorowski et al. 1991; Belousov et al. 1999);

  • Ramp structure similar to thrusting (Schneider and Fisher 1998).

4.3.3 Topography

One of the most prominent features of VDADs is their characteristic surface morphology called “hummocky topography” made of widespread fields of hills (hummocks) and depressions (Siebert 1984; Fig. 9). The geomorphological term “hummock” is also used for the description of massive ice bodies and cryoclastic formations (Grab 2005). Hummock shapes and sizes are highly variable. For example, at Cotopaxi volcano, hummocks from the 4.5 ka VDAD range from less than 4 m high and 100 m3 in volume up to 190 m high and 18 million m3 (Encalada and Bernard 2018; Fig. 9f, h). Some hummocks have a rounded base, and their slope slowly decreases towards their summit (Crandell et al. 1984; Fig. 9b, f) while others are more conical with a constant slope and a limited summit area (Siebert 1984; Fig. 9a). Some hummocks also have polygonal bases and are called pyramidal hummocks, while others have complex irregular shapes (Fig. 9f). When hummocks are strongly elongated, they are often called ridges (Fig. 10e) and can be kilometer-long (Siebert 1984; Fig. 9c). The major ridge axis can be parallel (longitudinal ridge) or perpendicular (transversal ridge, Fig. 9g) to the flow direction (Bernard et al. 2008; Kervyn et al. 2008; Dufresne and Davies 2009; Dufresne et al. 2010). Border ridges or natural levees can mark the limits of the deposit (Ui et al. 2000). Hummocks can be qualified as “torevas” when they correspond to very large (up to >1 km-long) almost intact pieces of the landslide source that had only suffered a relatively short translation from the source and, sometimes, a slight rotation through a horizontal axis (Reiche 1937; Fig. 9d, e, h). Some torevas do not go further than the landslide scar aperture (van Wyk de Vries et al. 2001; Bernard et al. 2011), but others reach the foot of the volcanic edifice (Belousov et al. 1999). The bedding in torevas is generally inclined toward the volcano (Clavero et al. 2002). In some cases, torevas were previously interpreted as lava domes (Bernard et al. 2011).

Fig. 9
figure 9

Typical morphologies of the debris avalanche deposit topography. a Conical hummock from the proximal area of the 9–25 ka Taapaca VDAD (Chile). b Rounded hummock from the distal area of the 60–65 ka Chimborazo VDAD (Ecuador). c Ridge from the medial area of the 60–65 ka Chimborazo VDAD (Ecuador). d Torevas from the proximal area of the 4.5 ka Cotopaxi VDAD (Ecuador) with lava flow filling the inter-hummock depression. E Landscape of the proximal area of the 8 ka Parinacota VDAD (Chile). f Digital Elevation Model (DEM, 4 cm spatial resolution, 2 m isohypse interval) of rounded and irregular hummocks from the distal area of the 4.5 ka Cotopaxi VDAD (Ecuador). g DEM (4 m spatial resolution, 20 m isohypse interval) of transversal ridges from the medial area of the 60–65 ka Chimborazo VDAD (Ecuador). h DEM (4 m spatial resolution, 25 m isohypse interval) of torevas from the proximal area of the 4.5 ka Cotopaxi VDAD (Ecuador). Red arrows in the DEMs show the debris-avalanche direction

Full size image
Fig. 10
figure 10

Main descriptive terms and geometrical parameters for volcanic debris avalanche deposits defined in Table 3. Black stars: gravity center for the source and the deposit

Full size image

It is typically possible to identify hundreds to thousands of hummocks on a single deposit (Crandell et al. 1984; Conway et al. 1992; Clavero et al. 2004; Shea et al. 2008; Encalada and Bernard 2018). Glicken (1991) proposes a classification based on their representative facies which we modify slightly using the facies classification proposed in this chapter (Sect. 4.2.1): (a) EBF hummocks (Siebert et al. 2004); (b) MF hummocks with a DAB core (Glicken 1991); (c) MF hummocks (Crandell et al. 1984); (d) SBF hummocks (Clavero et al. 2002). In some cases, like Parinacota or Mombacho VDADs, an evolution of the hummocks type from the proximal (mostly type a) to the distal area (mostly types c and d) is described (Clavero et al. 2002; Shea et al. 2008). Clavero et al. (2002) also distinguish between individual and compound hummocks, the latter being formed by the amalgamation of individual hummocks. The spatial organization and distribution of surface morphologies is a controversial theme, but recent analogue modeling has improved our understanding of these structures. In general, they are interpreted as the results of extension and compression of the avalanche body during transport (Shea et al. 2008; Paguican et al. 2014), and its lithology probably influences the hummock shape and size. In the field, the hummocky topography can be considerably modified by post-avalanche erosion and covering processes (Fig. 9e, h).

While the positive reliefs are commonly described, the depressions are often ignored. Inter-hummocks depressions are present between hummocks and might sometimes be filled by water (Fig. 9E). Furthermore, VDAD can present sub-circular depressions called “Kettle holes”, supposed to be left by the melting of large glacier blocks transported during the event (Glicken 1996; Clavero et al. 2002). Multiple grooves oriented parallel to the flow direction have also been observed in some cases, such as the 1964 Shiveluch VDAD (Belousov et al. 1999), the Pleistocene Llullaillaco VDAD (Richards and Villeneuve 2001), and the 218 BP Tutupaca VDAD (Valderrama et al. 2016) and may be associated with granular fingering (Valderrama et al. 2018) or high-velocity emplacement (e.g., “herringbone” structures at Lastarria volcano; Naranjo and Francis 1987).

4.4 Metrics and Morphology

The deposit limits have not been formally named, and we, therefore, propose to use the term “front” for the distal limit, “margin” for the lateral sides of the deposit and “tail” for the proximal boundary (Fig. 5). It is possible to characterize the dimensions of the deposits and the transport phase with numerous geometrical parameters (Table 3). The deposit volume is a crucial parameter that can be estimated using different methods. Crandell et al. (1984) use the average thickness for seven different segments of the 300–380 ka Mount Shasta VDAD. Clavero et al. (2002) calculate the volume of hummocks of the 8 ka Parinacota VDAD to obtain a minimum value. For old deposits, it is necessary to reconstruct the pre-deposit topography. Erosion and covering of the VDAD induce large uncertainties in this calculation. Thus the volume obtained for old deposits may not be relevant. The measurement of the deposit area is generally much more reliable than volume estimates. Deposit shape parameters are poorly studied yet, but they can give valuable information on the transport constraints and the flow dynamics (Bernard 2008). When the geometries of the landslide scar and the debris avalanche deposit are well known, it is possible to quantify the transport phase (Fig. 10; Table 3).

Table 3 Definition of the quantitative parameters for the volcanic debris avalanche deposit
Full size table

So far, few studies have included the shape of the debris avalanche deposits in their description. However, some terms have appeared in the literature such as “fan-shaped” when unconfined with a large concave front (Siebert 2002; Bernard 2008), “wedged” when widest at the tail volcano and narrowing near the front (Clavero et al. 2002), “bifurcated” when the deposit is divided by a topographic obstacle (Richards and Villeneuve 2001), and “shoestring” or “elongated” when confined by valley walls (Siebert, 2002; Bernard 2008). The shape of the deposit can indicate if and how the debris avalanche was confined, which can affect its mobility (Bernard 2008).

5 Conclusions

Four decades of research have increased our knowledge on volcanic landslides and debris avalanches, but, at the same time, the complexity of the scientific terminologies has also increased. Re-defining the key terms used to describe these phenomena and their geological features, coupled with a descriptive strategy, helps to standardize observations and enhance comparative analysis. The two main elements that indicate the occurrence of a volcanic landslide are the scar and the debris avalanche deposit. The volcanic landslide scars can be characterized using metrics, and their origin can be assessed through detailed geological observations. The description of volcanic debris avalanche deposits can be partitioned to obtain meaningful information on the emplacement mechanisms. Typical features are described and named, and the biggest challenge is now to produce a physical model that takes into account all the available data to fully understand the mobility of volcanic debris avalanches. Further analysis of their frequency, origin, and size are required to better understand and assess the associated hazard in volcanic settings.