Subaerial volcaniclastic deposits – influences of initiation mechanisms and transport behaviour on characteristics and distributions Open Access

Volcanic eruptions, and associated volcanic processes, can generate vast amounts of sediment that can mantle, modify, or create new topography (e.g. Manville et al. 2009a). Over common lifespans of stratovolcanoes and volcanic centres (c. 10⁵–10⁶ years), deposition and reworking of volcaniclastic sediment greatly affects surrounding terrain (Fig. 1). Depending on the nature of the volcanic processes, local topography, and climate, much of that sediment can remain in subaerial storage for spans of 10⁴–10⁵ years and some will pass into the geological record over time spans of 10⁶–10⁷ years (e.g. Roche et al. 2016). Volcaniclastic sediments stored over spans of 10³–10⁴ years commonly serve as the basis for understanding pertinent volcanic eruption histories and hazards of concern to society, and for deducing influences of volcanism and climate on sedimentary responses to eruptions. Longer time spans are needed to appreciate tectonic influences on sedimentary responses and storage (e.g. Smith 1991).

In this paper, I examine the influence of the nature of volcanic processes, and their initiation mechanisms and transport, on the character, storage, and preservation of volcaniclastic sediment. It is important to understand how volcanic processes, in conjunction with external influences such as glaciation and fluvial, colluvial, and aeolian reworking, affect volcaniclastic sediment because deposit characteristics and preservation strongly affect our perceptions of the eruptive histories of volcanoes and the hazards they pose. There are many excellent reviews of volcanic processes and deposits (e.g. Fisher and Schmincke 1984; Cas and Wright 1987; Fisher and Smith 1991; Branney and Kokelaar 2002; Ayris and Delmelle 2012; Bonadonna et al. 2015; Brown and Andrews 2015; Dufek et al. 2015; Houghton and Carey 2015; Vallance and Iverson 2015; Lube et al. 2020; Thouret et al. 2020; Roverato et al. 2021) as well as reviews of the hydrogeomorphic and sedimentologic responses to eruptions (e.g. Manville et al. 2007, 2009a; Manville 2010; Pierson and Major 2014). Those reviews delve far more deeply into processes and deposits than is possible here. My intention with this paper is to provide a high-level synthesis of major volcanic processes and deposits that compares and contrasts their characteristics and examines how initiation mechanisms and transport can affect the morphology, sedimentology, and distributions of that sediment. Despite extensive work done on submarine eruptions and their deposits, I focus solely on subaerial volcanism and deposition. Furthermore, I focus on processes and deposits associated with eruptions of stratovolcanoes v. eruptions and deposits from more distributed effusive volcanism and mafic cinder cones. Neither do I delve into hydrovolcanism or eruptions of maars, except for occasional brief mention. In the sections to follow, I discuss the major volcanic processes that generate and deliver volcaniclastic sediment during explosive and effusive eruptions – namely volcanic debris avalanches, pyroclastic density currents (PDCs – a generalized term for pyroclastic flows and surges), lahars, and tephra fall – as well as redistribution of that sediment following an eruption.

Several, and sometimes contradictory, terminological schemes have been proposed and used to describe volcanic sediment (see Fisher and Schmincke 1984; White and Houghton 2006; Manville et al. 2009a). I use the term volcaniclastic in a broad sense to refer to primary deposits that result from eruptive processes (see White and Houghton 2006; Manville et al. 2009a). Pyroclastic refers to primary clastic particles formed by volcanic explosions or other fragmentation processes. Some particles in volcaniclastic deposits are eroded from volcanic-conduit walls and, although they are not juvenile particles related to fresh magma, they can be volcanic particles. I do not assign any specific term to such particles but note their potential presence as needed. Other particles in volcaniclastic deposits are eroded along transport paths, and they may be of volcanic or non-volcanic origin. I call out such particles as necessary, but again assign no specific term to them. Primary deposits are those resulting directly from a specific eruption-related volcanic process without having been temporarily stored and subsequently remobilized. During and after deposition, primary deposits may be reprocessed or reworked (e.g. Sohn and Sohn 2019). Deposits of reworked sediment can have characteristics similar to primary deposits, making distinctions difficult. Where deposits have clearly resulted from processes not directly related to eruptions, yet retain characteristics of primary eruption deposits, I refer to them with the prefix ‘secondary’ (for example, secondary lahars). Several authors use the term epiclastic to refer to deposits that result from sedimentation by water regardless of the origin of the sedimentary fragment (see Manville et al. 2009a), but in this chapter I avoid this term. Here, I restrict discussion to subaerial volcanism and sedimentation (thus avoiding complications that may arise from terminology involving explosive submarine eruptions), and I further use descriptive phrasing, such as fluvially reworked and deposited sediment, rather than more ambiguous phrasing like epiclastic. The reader is referred to Fisher and Schmincke (1984), White and Houghton (2006), Manville et al. (2009a), Sohn and Sohn (2019), and Di Capua et al. (2022) for more detailed discussions of terminology. Specific volcanic processes also have their own nomenclature (e.g. block facies, bulking, hyperconcentrated flow, megaclasts, surge, etc.), which is defined and used as needed.

Volcanology has a common, but non-intuitive, language for describing and classifying grain-size characteristics of primary volcaniclastic sediment (White and Houghton 2006). In general, particles are classified as volcanic ash, lapilli, blocks and bombs (Table 1). Many sedimentologists, however, use the Wentworth (1922) classification system, which describes particle sizes as clay, silt, sand, granule, pebble, and cobble with various descriptive modifiers (e.g. fine, coarse, etc.). Modern descriptions of debris-avalanche and lahar deposits commonly use sedimentological grain-size terms, whereas descriptions of PDC and tephra-fall deposits typically use volcanological terms. Herein, I retain these conventions for consistency with the literature even though it may make some discussions appear inconsistent. Table 1 lists common volcanological and sedimentological grain-size terms.

In addition to grain-size characteristics, deposit sorting (σ), or the degree of segregation of grains of different sizes, is also used to characterize volcaniclastic deposits. Deposits that consist largely of grains of uniform size are considered very well sorted, whereas those that consist of a broad mixture of grain sizes are considered very poorly sorted. As with grain sizes, sometimes there are differences in the ways that volcanologists and sedimentologists describe deposit sorting (e.g. Cas and Wright 1987). Sorting is based on the deviation of the grain-size distribution about a mean size. Two sets of definitions for this parameter are commonly used: (1) the Inman (1952) graphical standard deviation, defined as σ_φ = (φ₈₄ − φ₁₆/2), where φ₈₄ and φ₁₆ are the 84th and 16th percentile phi-scale grain sizes, respectively, and (2) the inclusive graphic standard deviation defined by Folk and Ward (1957) as σ_I = ((φ₈₄ − φσ₁₆/4) + (φ₉₅ − φσ₅/6.6)), where φ₉₅ and φ₅ are the 95th and 5th percentile phi-scale grain sizes, respectively. Regardless of the method used, sedimentologists (and some volcanologists, e.g. White and Houghton 2006) commonly consider deposits having sorting values less than 1 to be very well to moderately sorted and those greater than 1 to be poorly to extremely poorly sorted (Folk and Ward 1957). In contrast, volcanologists sometimes broaden those descriptive categories such that sorting values less than 2 are considered well sorted and those greater than 2 are considered poorly to very poorly sorted (e.g. Cas and Wright 1987) (see Table 2). In this paper, I have adopted the sedimentologists’ perspective.

Volcanic processes that generate volcaniclastic sediment occur on a variety of scales. Debris-avalanche and PDC deposits can mantle, modify or create new topography. Lahar deposits can bury valley floors and lowland alluvial fans or thinly drape narrow corridors along river channels. Tephra-fall deposits mantle topography but can do so over thousands to tens of thousands of square kilometres. This variety of fill and areal coverage influences the effects of these events on the landscape as well as their subsequent erosion and preservation potential.

Contrary to outward appearances, many volcanoes are perched delicately on the landscape. They commonly consist of stratigraphic mixtures of volcaniclastic sediment, lava flows, and lava domes, which are sometimes highly altered. In some settings the stratigraphic mixtures include glacial sediment. As a result, many volcanoes have grown and collapsed repeatedly over millennia (e.g. Hausback and Swanson 1990; Belousova and Belousov 1995; Cronin et al. 1996; Calvari et al. 1998; Belousov et al. 1999; Cantagrel et al. 1999; Tibaldi 2001; Alloway et al. 2005; Pareschi et al. 2006; Waitt and Begét 2009; Roverato et al. 2011; Zernack et al. 2011; Tost et al. 2015; Cortés et al. 2019; Dufresne et al. 2021a; Zernack and Procter 2021). This tendency for volcanoes to collapse is ubiquitous (e.g. McGuire 1996; Dufresne et al. 2021a); nearly 600 volcanoes worldwide exhibit evidence of large-scale collapse (Siebert and Roverato 2021). Volcanic collapses can range from segments of lava domes to large sectors of a volcano (Fig. 2). The latter, which can range in size from tenths to hundreds of cubic kilometres, are referred to as volcanic debris avalanches (Voight 1981; Siebert 1984; Glicken 1996; McGuire 1996; Collot et al. 2001; Dufresne et al. 2021a; Siebert and Roverato 2021).

Debris avalanches can greatly alter the morphology of a volcano (Fig. 3). In general, the volume of material composing a debris-avalanche deposit is comparable to the volume of material ‘missing’ from a volcano. As a result, debris avalanches typically form an ‘avalanche crater’ (Siebert 1984), commonly manifest as a large, horseshoe-shaped breach of the volcano (Fig. 3a, b). These craters are typically a few kilometres wide, slightly longer in the direction of the breach, and many hundreds of metres deep (Siebert 1984). Ancient volcanic debris avalanches may not easily correlate with present volcano morphology as subsequent eruptions may fill and erase the scar (Fig. 3c, d).

Debris avalanches are exceptionally mobile and can affect large swaths of landscape (Fig. 4). Typically, the horizontal runout distance (L) of a debris avalanche is 5 to 10 times that of its vertical (H) drop (H/L ∼ 0.1–0.2) (Siebert 1984). Depending on failure size, local topography, and source material characteristics, debris avalanches can travel as little as a few kilometres from a volcano to many tens of kilometres and cover many tens to many hundreds of square kilometres (e.g. Voight 1981; Voight et al. 2002; Crandell et al. 1984; Wadge et al. 1995). They can surmount topographic barriers many tens to hundreds of metres tall (e.g. Voight 1981; Stoopes and Sheridan 1992) and flow across shallowly sloping terrain (e.g. Kelfoun et al. 2008). Estimated emplacement velocities are many tens of metres per second (e.g. Voight 1981; Stoopes and Sheridan 1992).

Historical debris avalanches have commonly been associated with magmatic or phreatic eruptions, although some may have occurred without an associated eruption (Siebert 1984; Aguila et al. 1986; Siebert et al. 1987; Paguican et al. 2012). If magma intrudes high into a volcanic edifice or is otherwise near the surface of the failure plane when a debris avalanche occurs, the decapitation of the volcano can trigger a magmatic explosion. Such decapitation can produce a violent, laterally directed or low-angle explosion and consequent PDC commonly referred to in the literature since 1980 as a lateral blast, directed blast, or blast PDC (Gorshkov 1959; Hoblitt et al. 1981; Waitt 1981; Siebert et al. 1987; Belousov 1996; Hoblitt 2000; Sparks et al. 2002; Belousov et al. 2007, 2020). Laterally directed PDCs associated with such events (see the section ‘Directed explosion’) can be highly energetic, sweep broadly across rugged topography, and devastate hundreds of square kilometres of landscape (e.g. Lipman and Mullineaux 1981, pl. 1; Bogoyavlenskaya et al. 1985; Belousov et al. 2007). On the basis of deposit characteristics and stratigraphic associations (see the sub-section ‘Deposit characteristics’ in the ‘Pyroclastic density current’ section), directed blasts have been inferred to have occurred in association with some prehistoric debris avalanches (Boudon et al. 1984; Francis et al. 1985; Siebert et al. 1995). Even in the absence of a directed blast, magmatically involved eruptions associated with debris avalanches have produced (Plinian) eruption plumes, PDCs, and subsequent lava domes (e.g. Katsui and Yamamoto 1981; Belousov 1995; Belousova and Belousov 1995; Belousov et al. 1999, 2020; Cutler et al. 2022). Variations in eruptive behaviour following collapse-driven unloading reflect variations in amounts of eruptible magma, magma storage, and modifications to pressurization of the magma reservoir (Pinel and Albino 2013; Watt 2019). Debris avalanches can also be associated with phreatic explosions that do not involve a magmatic component (Sekiya and Kikuchi 1889; Siebert 1984; Johnson 1987; Siebert et al. 1987; Yamamoto et al. 1999; Pinel and Albino 2013; Day et al. 2015). In those instances, phreatic eruptions occurred when pressure on hydrothermal fluids within the volcano was released suddenly by the debris avalanche. These types of debris-avalanche-triggered phreatic eruptions may also induce laterally directed explosions and PDCs that can devastate the surrounding area.

Precursory activity at some historical eruptions involving debris avalanches has included elevated seismicity, deformation, and minor eruptions (Siebert et al. 1987). However, some volcanoes may show little precursory activity or departure from prior styles of activity before an avalanche occurs. At Bandai volcano (Japan), its 1888 debris avalanche and eruption were preceded by substantial seismic unrest, which may have been related to deep magma movement. But it has been inferred that the debris avalanche was triggered by an earthquake well before magma had moved into the edifice (Siebert et al. 1987; Yamamoto et al. 1999). In 2018, Anak Krakatau (Indonesia) had been active for 6 months prior to flank collapse, but there were no changes in eruptive behaviour that might have signalled incipient failure immediately prior to collapse and consequent formation of a devastating tsunami (Perttu et al. 2020; Cutler et al. 2022).

Not all debris avalanches induce eruptions. For example, although Unzen volcano (Japan) was active in the late 1700s CE, a debris avalanche occurred on an older part of the volcanic complex at nearby Mayu-yama volcano during an earthquake (in 1792 CE), and this avalanche is not known to have been associated with any explosive activity (Katayama 1974; Siebert et al. 1987).

Volcanic debris avalanches pose multiple threats. (1) Because of their size and mobility, debris avalanches pose a severe hazard near volcanoes. They can also pose a hazard far from volcanoes if they are particularly mobile. Their mass and momentum are likely to crush all infrastructure within their paths, and their ability to surmount tall physical barriers minimizes topographic protections. (2) They are commonly associated with explosive eruptions and can trigger complementary volcanic processes (e.g. Hoblitt 2000; Belousov et al. 2020), such as PDCs that can sweep beyond the boundaries of the debris avalanche. (3) Debris avalanches from coastal, island, or submarine volcanoes can generate devastating tsunamis (Clark 1977; Katsui and Yamamoto 1981; Moore and Moore 1984; Siebert 1984; Johnson 1987; Tsuji and Hino 1993; Siebert et al. 1995; Belousova and Belousov 1995; Satake and Kato 2001; Ward and Day 2003; Giachetti et al. 2011; Tinti et al. 2011; Paris et al. 2014; Day et al. 2015; Paris 2015; Sassa et al. 2016; Grilli et al. 2019; Ye et al. 2020). (4) Owing to variations in size, volcanic debris avalanches can mantle or modify existing topography. If sufficiently large they can completely bury existing topography and form new topography. They can block tributary channels and lake outlets, enlarging existing lakes or forming new ones where none existed (Janda et al. 1984; Siebert 1984; Meyer et al. 1986; Lagmay et al. 2000; Pulgarín et al. 2001; Waythomas 2001; Capra 2007, 2011; Capra et al. 2002; Capra and Macías 2002; Clavero et al. 2002). (5) Some debris avalanches, particularly those containing substantial quantities of hydrothermally altered material or those that are particularly wet, can transform directly to lahars. Such transformations, although rare, can extend destruction far from a volcano (e.g. Carrasco-Núñez et al. 1993; Vallance and Scott 1997; Capra and Macías 2000; Detienne et al. 2017). (6) Subsequent to emplacement, debris avalanches are a source of sediment that, when reworked, poses severe and lasting societal consequences downstream (e.g. Lehre et al. 1983; Schuster 1983; Major et al. 2000, 2018, 2020; Major 2020). Although debris avalanches are infrequent events at individual volcanoes, on a global basis they have occurred historically a few times per century (Siebert 1984; Dufresne et al. 2021a).

Volcanic debris avalanches can form in many ways. Principal initiation mechanisms include: (1) failure owing to magmatic intrusion; (2) failure caused by an earthquake; (3) failure resulting from gradual weakening of an edifice caused by hydrothermal alteration; (4) failure from gradual weakening caused by slope loading; (5) failure of the volcano's substrate; or (6) failure caused by peripheral erosion or debuttressing of a volcano (e.g. Elsworth and Voight 1996; McGuire 1996; van Wyk de Vries and Francis 1997; Voight and Elsworth 1997; van Wyk de Vries et al. 2001; Reid et al. 2001, 2010; Reid 2004; Paguican et al. 2012; Siebert and Roverato 2021). Magmatic intrusions elevate pore-fluid pressures both mechanically and thermally, thus weakening the strength of volcanic rock. Deformation caused by intrusion can steepen volcano flanks, increasing the shear stress exerted on potential slip surfaces within the edifice. These complementary processes can destabilize an edifice and increase the possibility of a large, deep-seated failure (Day 1996; Elsworth and Voight 1996; Voight and Elsworth 1997; Reid 2004). Magmatic intrusions also increase the rate and magnitude of earthquakes, which can generate transient stresses that exceed weakening rock strength and precipitate failure. Over the longer term, hydrothermal alteration gradually weakens segments of a volcano making it more susceptible to failure (e.g. van Wyk de Vries and Francis 1997; Capra and Macías 2000; Reid et al. 2001, 2010; Vallance 2005). Stresses on and within a volcano can increase through gradual loading of a volcano slope by repeated extrusion of lava, and the increasing stresses can lead to edifice failure. Gravitational loading of a weak substrate beneath a volcano owing to the weight of a volcano can lead to failure. Loading of substrate materials composed of older, weathered materials, hydrothermally weakened materials, or weak sediments (such as lake sediments) can lead to slow lateral spreading and eventual structural failure of the substrate, which can cause catastrophic failure of the overlying volcano (van Wyk de Vries and Francis 1997; van Wyk de Vries et al. 2001; Cecchi et al. 2005). Peripheral erosion at the land–sea interface of coastal and island volcanoes and debuttressing of volcanoes by deglaciation can lead to gradually increasing stresses in an edifice (e.g. Roberti et al. 2018). Changing sea levels (in particular sea-level rises) can affect pore-fluid pressures within an edifice and lead to gradual weakening of rock strength (e.g. McGuire 1996).

The surfaces of volcanic debris-avalanche deposits commonly are dotted with numerous distinctive mounds known as hummocks or hillocks (e.g. Ui 1983; Siebert 1984; Crandell 1989; Palmer and Neall 1989; Glicken 1996; van Wyk de Vries and Davies 2015; Dufresne et al. 2021b; Siebert and Roverato 2021) (Fig. 5). These mounds can range in height from a few metres to more than 100 m and have diameters from a few to a few hundreds of metres (e.g. Siebe et al. 1992; Glicken 1996; Romero et al. 2022). Over time, these mounds may become more gently rounded and surfaces between mounds filled and flattened. The origin of this distinctive morphology was debated for many decades prior to the 1980 eruption of Mount St Helens (USA). Before 1980, these mounds were hypothesized to represent glacial deposits, phreatic ‘blisters’ on the surfaces of gas-rich lava flows, independent volcanic vents, results of landslides or lahars, or even anthropogenic features (Siebert 1984). Following the 1980 Mount St Helens eruption, it became clear that such mounds are a distinctive feature of volcanic debris avalanches. Voight et al. (1981) attributed them to horsts developed in an extending sheet of debris. Glicken (1996) attributed some, especially in distal parts of the deposit, to transport of individual large clasts. Paguican et al. (2014) conducted scaled analogue experiments and showed hummocks formed initially by extensional faulting during avalanche motion. They also showed that hummock size depends on position within the flowing mass, and that size can be modified as hummocks disintegrate and coalesce during motion. In their experiments, small hummocks tended to populate the avalanche front whereas larger hummocks remained closer to source, consistent with characteristics of hummocks in natural deposits (Glicken 1996; Yoshida et al. 2012; Yoshida 2013). Field studies by Shea et al. (2008) of debris-avalanche deposits emplaced over nearly flat, unconfined topography revealed a predominance of extensional structures among hummock fields consistent with experiments by Paguican et al. (2014). Particularly large blocks that preserve stratigraphy that is intact but tilted relative to its source, known as toreva blocks, may be present at the bases of failure scars (e.g. Clavero et al. 2002; Paguican et al. 2014; Dufresne et al. 2021b; Romero et al. 2022).

Surface morphologies of volcanic debris-avalanche deposits may also be characterized by ponds, marginal levees, and ridges (Fig. 6; see also Siebe et al. 1992; Romero et al. 2022). Extensional faulting and spreading of a debris avalanche not only isolate mounds but can form closed graben that can fill with water. Deposit extension and internal shearing can form linear ridges along the deposit surface. Along valley margins, lateral levees may form, some of which may block outlets of tributary channels and form new lakes.

Debris-avalanche deposits typically are composed of a chaotic mix of particles, from large intact or deformed blocks of a volcano to an extensively disaggregated mixture of volcanic material as well as sediment, rock, and organic debris entrained during emplacement. Deposits can range widely in thickness, from a few metres to more than 150 m. Depending on pre-existing topography and avalanche volume, average thicknesses commonly range from a few to many tens of metres.

A wide range of terminology has been applied to debris-avalanche deposits. Summaries of terminology are provided in Siebert (1984), Glicken (1996), van Wyk de Vries and Davies (2015), and Dufresne et al. (2021b). In broadest terms, units within volcanic debris-avalanche deposits are categorized and mapped according to their sedimentological and lithological facies (e.g. Glicken 1996). Debris-avalanche deposits are broadly characterized according to one of two principal facies: block or mixed (sometimes called matrix) facies (Fig. 7). Block facies (also referred to in the literature as megablock and axial facies) refers to those characteristics indicative that the deposit represents material transported largely intact from the source. The block facies can be highly brecciated and pervasively shattered such that no individual particle is larger than about a metre, but this facies clearly preserves material transported intact, such as original source stratigraphy (Palmer and Neall 1989; Glicken 1996; Fig. 7a, b). In contrast, the mixed facies represents material that cannot be shown to have been transported completely intact. Instead, this material is highly fragmented and disaggregated, may contain multiple clast lithologies, is generally unsorted, lacks stratification or grading, and typically appears to consist of isolated large particles (clasts) surrounded by or floating in a matrix of finer-grained sediment (dominantly sand) (Fig. 7e, f). The mixed facies is largely inferred to be the result of disaggregation and mixing of material during transport (Glicken 1996; van Wyk de Vries and Davies 2015). At intermediate distances, outcrops can exhibit textural characteristics of incomplete mixing, revealing a transition from block to mixed facies (Fig. 7c, d). If the flow was locally saturated, facies characteristics may show evidence of transition from debris avalanche to lahar, such as thinning of the deposit into a cohesive, matrix-supported flowage deposit with a progressive decrease in primary clasts and increase in secondary entrained particles (Zernack et al. 2011).

Many particles within debris-avalanche deposits are highly shattered (Fig. 8). Evidence commonly indicates that particle shattering occurs largely during the initial release of the avalanche, and that subsequent downstream transport disaggregates fragments but causes little additional shattering. For example, Voight et al. (1983) and Glicken (1996) found that bulk density of the Mount St Helens debris-avalanche deposit varied little from proximal to distal reaches, indicative that the material was already highly shattered in proximal areas. Inspection of individual clasts shows that in proximal areas, many clasts are shattered but still relatively coherent – called jigsaw-cracked clasts (e.g. Siebert 1984; Glicken 1996; van Wyk de Vries and Davies 2015; Dufresne et al. 2021b). With distance, debris-avalanche deposits show features of clast disaggregation and separation, internal faulting, smearing of material by internal shearing, fingers of material injected amongst particles, and other evidence to indicate that material is entrained, disaggregated, and mixed with transport distance (e.g. Siebe et al. 1992; Glicken 1996; van Wyk de Vries and Davies 2015; Dufresne et al. 2021b) (Fig. 8). Even at the microscopic level, particles sampled from proximal to distal reaches of debris-avalanche deposits exhibit features indicative of shattering and disaggregation, such as microcracks, hackly fractures, and fracture separation (e.g. Komorowski et al. 1991; Dufresne et al. 2021b).

Quantitative sedimentological characteristics of debris-avalanche deposits overlap with other volcaniclastic deposits. Analyses of median grain size, sorting, and grain shape show substantial overlap with those from lahar and PDC deposits (e.g. Siebert 1996; Dufresne et al. 2021b). Indeed, some outcrops of mixed facies of debris-avalanche deposits may be difficult to distinguish from those of PDC and lahar deposits, and correct interpretation may rely on broader morphologic and sedimentary context.

Debris-avalanche-deposit textures are likely to be similar despite the nature of the initiation mechanism. Key characteristics of a debris-avalanche deposit are evidence of mass transport from a volcano. This evidence includes distinctive surface morphology (hummocks), transport of intact stratigraphic sections, preservation of jigsaw-cracked clasts, and mottled coloration indicative of incomplete blending of stratigraphic units (Figs 5, 7, 8). There may also be evidence of localized internal shearing and pressured loading (clastic dykes) (Fig. 8). These features, however, do not discriminate a specific initiation mechanism, as the transport process, more than the initiation mechanism, controls these deposit textures. Stratigraphic associations, however, can lend insights into probable, or at least possible, initiation mechanisms. Debris-avalanche deposits with clear association to a magmatic eruption – such as close temporal association with tephra-fall or PDC deposits, especially with deposits from directed explosions rich in juvenile material – provide strong evidence that edifice failure was associated with magmatic intrusion. Debris-avalanche deposits containing substantial amounts of highly deformed subvolcano substrate are suggestive of substrate failure as a precipitating cause, especially if there is structural evidence of failure reaching below the edifice. Debris avalanches caused by earthquakes, slope loading, or debuttressing may or may not be associated with magmatic intrusions or phreatic explosions. Those induced by hydrothermal weakening of an edifice may or may not be associated with an eruption, but close association with deposits containing juvenile material is indicative of magma involvement. Evidence of direct transformation to a lahar along with evidence of inclusion of hydrothermal material may signal that hydrothermal weakening of an edifice was a significant contributing factor to the debris avalanche.

A pyroclastic density current (a generalized term for a pyroclastic flow or surge) – or PDC – is a hot, gravity-driven, heterogeneous mixture of air, gases, and volcanic particles, which is denser than the ambient atmosphere and flows away from a volcano (Wilson 1986; Druitt 1998; Branney and Kokelaar 2002; Sulpizio et al. 2014; Dufek et al. 2015; Lube et al. 2020) (Fig. 9). It can occur suddenly during an eruption, travel many kilometres across the landscape at high speeds (many tens of metres per second), and burn, bury, and destroy everything in its paths. Its mass, momentum, and temperature (to many hundreds of degrees Celsius) make it particularly destructive and hazardous.

A PDC can have variable particle concentration and be strongly stratified. Dynamic relations among grain interactions, particle densities and settling velocities, turbulent eddies, and fluid drag on solid particles promote particle segregation. Such segregation leads to stratification within the current, creating a dense, high-concentration flow phase, dominated by particle interactions modulated by pore-fluid pressures, underlying a dilute, low-concentration turbulent phase in which particle settling is affected dominantly by fluid drag (e.g. Wilson 1986; Druitt 1998; Burgisser and Bergantz 2002; Sulpizio et al. 2014; Dufek et al. 2015; Lube et al. 2020) (Fig. 10). Although recent experimental work has shown that PDCs can span a spectrum of solids concentrations (Lube et al. 2020), two end-member styles of currents are representative of common conceptual models. A coarse-grained, high-concentration current, known as a pyroclastic flow, commonly has a particle concentration of about 50% by volume (e.g. Lube et al. 2020). This dense flow generally hugs the ground and is funnelled along topographically low areas. This phase of a PDC commonly constitutes the basal part of a flowing current. In contrast, a swiftly moving, relatively fine-grained, low-concentration flow, known as a pyroclastic surge, commonly has a particle concentration of about 1% by volume or less (e.g. Lube et al. 2020). A surge is less constrained by topography and can sweep broadly across the landscape, even surmounting topographic ridges. As a result of these different concentrations and solid–fluid interactions, a PDC develops into a stratified current that can be many tens to hundreds of metres thick (e.g. Druitt 1998; Burgisser and Bergantz 2002; Sulpizio et al. 2014; Dufek et al. 2015; Lube et al. 2020). Because of particle segregation, a PDC can evolve with travel distance. The basal flow can increase in concentration through both substrate entrainment and internal particle settling, and a surge of fine sediment and gas can detach from the underlying coarser flow and travel as a separate entity (e.g. Fisher 1995). Thus, a single PDC can evolve into multiple currents involving different behaviours, sedimentation regimes, and hazards (Branney and Kokelaar 2002; Fisher 1995; Sulpizio et al. 2014; Dufek et al. 2015; Lube et al. 2020).

In addition to flow stratification that forms within the initial flow mass, hot gases and heating of air ingested into the flow cause buoyant rise and elutriation of fine particles into a cloud above the main flow. If this cloud maintains substantial lateral momentum, it can produce a deposit with evidence of a flowage component, such as cross-bedding or lateral interactions with the substrate and vegetation. In contrast, if deposition from this cloud is predominantly from vertical fall, the deposit will exhibit characteristics of a fall process (see ‘Tephra fall’ section). Ash clouds emanating from PDCs (called co-PDC ash clouds) can be substantial and produce abundant fall deposits (e.g. Hoblitt 2000) (Fig. 11). They can emerge from a widespread footprint (in contrast to volcanic plumes that rise from discrete vents) and rise many tens of kilometres in altitude (e.g. Sparks et al. 1986; Waitt 2015). Condensation and electrostatic forces within a co-PDC ash cloud may cause aggregation of fine ash and formation of wet accretionary lapilli or dry-ash aggregates (see sub-section ‘Deposit characteristics’ in the ‘Tephra fall’ section).

Multiple initiation mechanisms generate PDCs (Figs 9 & 12). The most common initiation mechanisms include: (1) collapse of parts or all of an eruption column; (2) sustained fountaining of volcanic ejecta that do not fully rise into a lofting eruption plume; (3) laterally or low-angle-directed explosions; (4) collapses of columns from fissures, ring fractures, or multiple vents during caldera collapse; (5) collapses of growing lava domes; (6) collapses of fronts of advancing lava flows; and (7) interactions of magma and water during hydrovolcanic eruptions. In addition, PDC sediment deposited precariously on steep hillsides can spontaneously remobilize and form secondary PDCs (e.g. Hoblitt et al. 1981; Fisher 1990). A less common mechanism is slumping of valley-fill PDC deposits, which result in secondary PDCs (Scott et al. 1996; Robinson et al. 2017).

The largest and most hazardous PDCs form during explosive eruptions. PDCs that form during explosive eruptions commonly result from gravitational collapse of an eruption column, although some are the result of directed explosions. Although initiation of PDCs from eruption columns spans a gradational spectrum of processes, this spectrum is captured by three principal variations: eruption-column collapse from a tall column, volcanic fountaining, and boiling over (Fig. 12a–c).

Explosive eruptions commonly thrust volcanic particles vertically into the atmosphere forming an eruption column. Magmatic water content, vent radius, particle size and density, exit velocity, mass flow rate of the ejected material, and entrainment of ambient air influence the height to which ejected material is thrust, the behaviour of the column, and the partitioning of mass between convection and collapse (e.g. Sparks and Wilson 1976; Wilson et al. 1980; Neri et al. 2002; Shea et al. 2011; Carey and Bursik 2015) (Figs 9a & 12a, b). As ejected material rises, there are two conceptual end members that characterize its behaviour: (1) the column can efficiently entrain ambient air, become fully buoyant, convect to great height, and the ejecta are dispersed largely downwind and return to the ground surface as tephra fall, or (2) the column fails to entrain, or entrains insufficient, ambient air to become fully buoyant and collapses gravitationally, forming PDCs (e.g. Sparks and Wilson 1976; Wilson et al. 1980; Neri et al. 2002). In reality, convection and collapse usually attain some intermediate hybrid state, such that some of the eruption column becomes buoyant while parts of it collapse (e.g. Taylor 1958; Moore and Melson 1969; Nairn and Self 1978; Hoblitt 1986; Clarke et al. 2002; Saucedo et al. 2002; Neri et al. 2007; Dufek et al. 2015; Miyabuchi et al. 2018; Roche et al. 2021). Depending on how much of the eruption column collapses, the geometry of the collapse, and the topography of the volcano, PDCs can be spatially focused or radial in nature and affect many drainages of a volcano simultaneously (e.g. Moore and Melson 1969; Sigurdsson et al. 1984, 1985; Hoblitt et al. 1996; Scott et al. 1996; Neri et al. 2007).

Collapse of large portions of an eruption column can transfer large fluxes of mass and momentum into a highly energetic PDC (e.g. Taylor 1958). However, even partial collapse of an eruption column can generate a substantial PDC. Individual collapses generate unsteady, single-pulsed currents whereas sustained, nearly continuous collapses feed prolonged, quasi-steady currents (Fig. 12a, b). A collapse-triggered PDC can consist of multiple phases and undergo a transformation as it travels (e.g. Fisher 1983). A column-collapse PDC can form a pyroclastic surge at its leading edge followed by pyroclastic flow, which is then followed by a surge from higher in the flow (e.g. Fisher 1979; Sigurdsson et al. 1984, 1985). Surges spawned by column collapses commonly outrun the footprint of pyroclastic flows and expand the area of impact.

If an eruption column is unsteady or not particularly energetic, ejected material may rise to only a modest height above the vent before it collapses. Modest column rise is especially common in low fountains of ejecta that contain relatively little magmatic water content (less than 1% by volume; e.g. Sparks and Wilson 1976; Neri et al. 2002). As a result, the column lacks buoyancy or rises only weakly and collapses quickly into a PDC. Eruption columns can also be laden with dense juvenile clasts and entrained wall-rock material, which inhibit buoyancy (e.g. Shea et al. 2011). Such eruptions may occur in pulses or may be sustained for prolonged periods of time. Thus, volcanic fountaining can feed a transient or sustained PDC that can travel several kilometres from its source. This process of PDC initiation has been observed and modelled at several volcanoes (e.g. Hoblitt 1986; Voight et al. 2000a; Clarke et al. 2002; Cole et al. 2005; Maeno et al. 2016) (Fig. 9b, e).

Sluggish, transient eruption columns that barely rise above the vent before collapsing have been described as appearing like a pot that is boiling over (e.g. Wolf 1878; Hoblitt 1986; Clarke et al. 2002; Hall and Mothes 2008; Rader et al. 2015). Such boiling over is merely a gradation of volcanic fountaining (e.g. Hoblitt 1986; Clarke et al. 2002).

Rapid exposure of volatile-rich magma to the atmosphere or sudden reduction of confining pressure on a shallowly emplaced magmatic body can lead to an explosion that is directed laterally or at low angle (Gorshkov 1963; Voight 1981) (Figs 9c, d & 12d). A directed explosion occurs most commonly following a debris avalanche or failures of a lava dome. Rapid decompression permits swift exsolution of magmatic volatiles and generation of a fragmentation wave, perhaps augmented by flashing of heated groundwater to steam, which results in rapid fragmentation of magma and surrounding wall rock and an energetic explosion that spawns a PDC (e.g. Woods et al. 2002). The PDC is initially propelled by decompression of the magmatic system, but swiftly becomes negatively buoyant and forms a gravitationally driven flow as the ejected material collapses (e.g. Esposti Ongaro et al. 2012; Figs 9c, d & 12d). Perhaps the most renowned laterally directed PDC occurred during the 18 May 1980 eruption of Mount St Helens (Hoblitt et al. 1981; Hoblitt 2000; Voight 1981). In that eruption, a debris avalanche rapidly decompressed dacitic magma that had risen high into the volcano. The debris avalanche (Fig. 2), ensuing directed blast, and consequent PDC were observed and well documented by many eye-witnesses (e.g. Voight 1981; Waitt 2015; Figs 9c, d & 13). The PDC spread rapidly (Fig. 13), swept over multiple ridges, and devastated nearly 600 km² of rugged landscape in a 180° arc north of the volcano. A similar event occurred during the 1956 eruption of Bezymianny volcano (Russia) and devastated some 500 km² of rugged landscape (Gorshkov 1959).

Phreatic explosions may also generate directed PDCs. The 1888 eruption of Bandai volcano included an earthquake-triggered debris avalanche that rapidly decompressed a pressured hydrothermal system. Rapid decompression triggered several brief phreatic explosions and spawned a PDC that travelled about 6 km and devastated 13 km² (Yamamoto et al. 1999). The PDC that resulted from these explosions had been thought to result from a directed explosion, but Yamamoto et al. (1999) suggested the PDC resulted from fountaining of dense, transient, vertical explosion columns rather than from a directed explosion. Nevertheless, directed explosions can occur during phreatic eruptions.

Directed explosions have been reported for other eruptions, but they are largely inferred based on deposit characteristics (discussed in the sub-section ‘Deposit characteristics’ in the ‘Pyroclastic density current’ section). Those deposit characteristics, sometimes showing close association in time with debris-avalanche deposits, have been used to infer directed-blast origins for PDC deposits (Boudon et al. 1984, 1990; Ritchie et al. 2002; Sparks et al. 2002; Komorowski et al. 2013; Major et al. 2013; Pallister et al. 2019).

Distinguishing a directed-blast origin of a deposit from one resulting from an otherwise energetic PDC can be complex. Even a close association with a debris avalanche and significant impact on vegetation do not necessarily indicate a low-angle-blast origin. At Lamington volcano (Papua New Guinea), characteristics of volcanic deposits and impacts on vegetation were used to infer that its 1951 eruption included a directed-blast PDC (Taylor 1958). Belousov et al. (2020) reanalysed the deposits and eyewitness accounts and concluded a debris avalanche likely decapitated a magmatic body that nearly reached the surface. Because decapitation exposed the top of the magmatic body rather than its steep side, the directed explosion was high-angle rather than lateral or low-angle. This high-angle-directed explosion generated a dense eruption column that collapsed and formed a transient but very energetic PDC having many hallmarks of a laterally directed PDC. The directionality of the PDC was controlled largely by the pre-existing topography of the volcano rather than being horizontally propelled. At Mont Pelée (Martinique, French Department Territory), eyewitness accounts and distributions of deposits have been debated as to whether the devastating 8 May 1902 PDC was caused primarily by a directed blast or column collapse (LaCroix 1904; Fisher and Heiken 1982, 1990; Boudon et al. 1990). Numerical modelling of the event by Gueugneau et al. (2020) concluded there were likely elements of both. They concluded a sudden decompression of the lava dome growing in the volcano's crater led to a brief blast-like event and a low, dense vertical eruption column that quickly collapsed into an energetic PDC that was directionally focused owing to the geometry of the crater and downslope topography. As this PDC moved downslope it developed into a powerful pyroclastic surge that spread widely. These analyses and reinterpretations of the Lamington and Pelée volcanic eruptions show that there is clear gradation among initiation mechanisms of PDCs and that inferences of a specific initiation mechanism from deposit characteristics must be approached with caution.

The largest PDCs result from collapses of columns or intense fountaining formed during eruptions leading to caldera collapses. Some of these PDCs are formed during voluminous eruptions from single, central vents prior to caldera collapse as vent-wall erosion increases vent diameter and air entrainment is insufficient to produce a buoyant plume (e.g. Wilson et al. 1980). Others occur from eruptions along caldera ring vents that form as the caldera roof founders (e.g. Bacon 1983; Wilson 1985; Self 1992; Wilson and Hildreth 1997; Kandlbauer and Sparks 2014). Multiple-pulsed or prolonged and sustained PDCs resulting from caldera collapses produce deposits that can fill and smooth rugged topography (e.g. Punongbayan et al. 1996). Cumulative accumulations of PDC deposits from caldera-collapse eruptions have volumes that range from as little as a few cubic kilometres dense rock equivalent (DRE) to thousands of cubic kilometres (e.g. Bacon 1983; Wilson 1985; Hildreth and Mahood 1986; Self 1992; Scott et al. 1996; Allen 2001; Christiansen 2001; Cas et al. 2011; Chesner 2012; Hildreth and Fierstein 2012; Kandlbauer and Sparks 2014; Marti et al. 2016; Takarada and Hoshizumi 2020; Valentine and Cole 2021).

PDCs produced during caldera-collapse eruptions affect widely ranging areas, are emplaced within minutes to days, and produce deposits having widely variable thicknesses. These PDCs have commonly spread radially from volcanic centres, covered hundreds to a few tens of thousands of square kilometres in area, and reached travel distances of many tens to nearly 200 km from source. PDCs associated with caldera collapses are commonly sustained and emplaced over hours or days’ duration (e.g. Scott et al. 1996; Wilson and Hildreth 1997; Hildreth and Fierstein 2012). However, they can be emplaced more quickly. Wilson (1985) estimated that the 10 km³ (DRE) deposit emplaced by the 186 CE eruption of Taupo volcano (New Zealand), which spread over some 20 000 km² to distances of 80 km from source, was emplaced within 400 s – a scant 6.5 minutes! Caldera-collapse-generated PDCs can be emplaced as turbulent surges that overrun rugged topography and leave thin veneers (less than 1 m thick) that drape pre-existing topography (e.g. Wilson 1985; Scott et al. 1996; Hildreth and Fierstein 2012). They can also form dense pyroclastic flows that follow low topography and thickly bury valleys. Valley deposits from many caldera-collapse PDCs are commonly tens to a few hundreds of metres thick (Bacon 1983; Self et al. 1984; Wilson 1985; Self 1992; Fisher et al. 1993; Scott et al. 1996; Christiansen 2001; Maeno and Taniguchi 2007; Cas et al. 2011; Chesner 2012; Takarada and Hoshizumi 2020), but may be as little as a few metres thick (e.g. Hildreth and Fierstein 2012).

The initiation mechanisms of PDCs associated with caldera collapses are no different than described previously for smaller volume eruptions, they merely happen on a much grander scale. Caldera-collapse PDCs have been inferred to result from high mass-feeding rates from quasi-continuous collapse of parts of sustained eruption columns, by repeated brief collapses of entire eruption columns, and by intermittent partial collapses of eruption columns much as occur during smaller eruptions (e.g. Scott et al. 1996; Cas et al. 2011; Valentine and Cole 2021). Intracaldera fill, however, can affect the style of PDC emplacement (Valentine and Cole 2021). Large amounts of PDC sediment can become trapped within developing calderas. As a result, eruptions must penetrate the accumulating deposits. This behaviour, termed a gargling eruption (Wilson and Hildreth 1997; Valentine and Cole 2021), increases grain size along the margins of the eruption column and extracts momentum from the central jet. As a result, PDCs are generated from a column that might otherwise rise buoyantly and produce fall deposits (Valentine and Cole 2021). Furthermore, material along the edges of the eruption column collapses from a range of heights producing transient pulsing behaviour. The thickness and grain-size composition of the intracaldera fill strongly influence the behaviour of the erupting jet even if mass flow rate is constant, which can affect PDC behaviour and deposit characteristics (Valentine and Cole 2021).

Explosions from, or collapses of, actively growing lava domes can generate PDCs. Collapses can involve either minor segments of the dome (e.g. Mellors et al. 1988) or nearly an entire dome (Sparks et al. 2002) and they can be gravitationally or explosively driven. Upon collapse, failed dome rock rapidly disintegrates and fragments into flowing mixtures of coarse rock and finer-grained particles (Figs 9f & 12e). Commonly known as block-and-ash flows (BAFs), these types of PDCs form transient, unsteady currents. They occur most often during active extrusions, particularly if extrusion rates are high (several to a few tens of cubic metres per second; e.g. Komorowski et al. 2013; Pallister et al. 2013, 2019), domes are perched on steep slopes, there is substantial seismicity that can trigger dome failure, or if domes are overpressured and explosions occur during extrusion (e.g. Voight and Elsworth 2000). The style of dome growth – endogenous or exogenous – as well as magma composition, fluid pressure, dome volume, and mechanisms of collapse influence the style and size of collapse and formation of a subsequent PDC (e.g. Sato et al. 1992; Woods et al. 2002; Harnett et al. 2019). In general, dome-collapse PDCs are small in volume (10⁴–10⁶ m³), travel short distances (a few to several kilometres), and are spatially focused. However, some dome-collapse PDCs, such as those at Soufrière Hills volcano (Montserrat, British Overseas Territory) in 1997, can be large (10⁷ m³) and cover many square kilometres (Calder et al. 1999; Sparks et al. 2002). In most instances, drainages likely to be affected by these types of PDCs can be anticipated based on location of the lava dome and volcano topography, but overlying surges can detach from the coarser parent flow and affect a broader footprint (Fisher 1995). Dome-collapse PDCs, although small compared to column-collapse PDCs, can nevertheless substantially impact landscapes close to volcanoes (e.g. Sato et al. 1992; Yamamoto et al. 1993; Gardner et al. 1994; Calder et al. 1999; Voight et al. 2000b; Carn et al. 2004; Thouret et al. 2010; Vallance et al. 2010; Major and Lara 2013; Reyes-Dávila et al. 2016; Pallister et al. 2019; Darmawan et al. 2020). Repeated collapses produce abundant sediment that can later be remobilized as lahars and transported farther downstream. Collapses of older lava domes that are no longer active typically generate flows akin to cold rockfalls that are more restricted in volume and travel shorter distances than those that occur when a dome is actively growing.

If extrusion rates are sufficiently high, cooling rates sufficiently low, viscosity sufficiently low, and slopes sufficiently steep, actively extruding silicic lava domes can transition into slowly advancing lava flows. These dome-transition lava flows can extend several kilometres. On occasion, the front or sides of those advancing flows can fail, exposing gas-rich lava that can rapidly vesiculate, explode, and generate a PDC. Although less common than those from dome collapses, PDCs generated in this manner are similar to dome-collapse BAFs. Like those spawned by dome collapses, these PDCs can travel several kilometres at speeds of a few tens of metres per second (e.g. Rose et al. 1976; Harris et al. 2002; Saucedo et al. 2002; Pallister et al. 2019). However, because these PDCs form by collapse of lava flows, those lava flows displace the source of the PDC from the volcanic vent, possibly up to several kilometres. If the lava is sufficiently pressurized, the collapse of the front or side of the lava flow can spawn a directed-explosion PDC that can have severe societal and ecological consequences (e.g. Pallister et al. 2019).

Near-surface interactions of magma and groundwater can produce energetic explosions that generate PDCs. Rapid conversion of thermal to mechanical energy and generation of shock waves within a conduit cause extensive fragmentation and pulverization of both magma and wall rock (Zimanowski et al. 2015). This rapid conversion of energy can generate pyroclastic surges (known as base surges) rich in fine ash. Examples of phreatomagmatic eruptions producing PDCs include eruptions of Ontake volcano (Japan) (Kaneko et al. 2016), Kuchinoerabujima volcano (Japan) (Geshi and Itoh 2018), Aso volcano (Japan) (Miyabuchi et al. 2018), Ukinrek Maars (USA) (Self et al. 1980), the Oruanui eruption of Taupo volcano (Wilson 2001), and the Table Rock complex (USA) (Brand and Clarke 2012).

Deposition by PDCs is conceptually bounded by two end members: en masse deposition in which the flow stops abruptly and progressive deposition in which sediment at the flow base gradually accumulates during flow passage (e.g. Branney and Kokelaar 1992, 2002; Brown and Andrews 2015; Dufek et al. 2015; Lube et al. 2020). Through en masse deposition, a deposit represents the structure and texture of the entire dense underflow of the current at the instant of deposition. In contrast, progressive deposition reflects the time varying characteristics of the base of the current as it passes a location, which reflects only a limited part of the structure and texture of the entire flow. In this concept, deposition is a sustained but time-varying process with deposit characteristics determined by the nature of a flow-boundary zone – the region that encompasses the basal part of the current and the uppermost part of the aggrading deposit (Branney and Kokelaar 1992, 2002) (Fig. 14).

The nature and character of the PDC influences the nature of particle support, transport, and deposition at the flow-boundary zone (Branney and Kokelaar 1992, 2002). If a current is fully dilute with little intergranular particle support, then the flow-boundary zone is dominated by direct particle fallout from higher in the current and a strong shear gradient is developed near the deposit surface. Strong shear stress at the current–deposit interface facilitates tractional transport and leads to a stratified and moderately to well-sorted deposit. In contrast, in a highly segregated current with a dense basal underflow, the flow-boundary zone is dominated by particle–particle interactions modulated by interstitial fluid pressure. As a result, this flow-boundary zone is dominated by granular-flow processes commonly resulting in non-stratified deposits that can exhibit variable sedimentary textures with respect to clast composition and orientation, fluidization structures, and particle grading. As clasts settle through this concentration-stratified flow-boundary zone, they are subject to selective filtering (Branney and Kokelaar 2002). For example, dense lithic particles may pass downward and be deposited, whereas less-dense pumice particles may be unable to penetrate the flow-boundary zone locally and are transported farther downcurrent and deposited in clast-supported lobes and levees (e.g. Brown and Andrews 2015). Selective filtering of the flow-boundary zone can vary in space and time. PDCs are, however, more complex and gradational than these two conceptual end members, and thus PDC deposition is more complex. Variations in source emissions that produce unsteady source behaviour, longitudinal variations in topography, internal current stratification, flow-boundary zones, interactions with the substrate, and local accelerations and decelerations result in a complex interplay of forces among particles and gases, particle concentration, particle sedimentation, and ultimately the texture of PDC deposits (e.g. Druitt 1998; Branney and Kokelaar 2002; Brown and Andrews 2015; Dufek et al. 2015; Lube et al. 2020). Even single currents can produce deposits having broad temporal and spatial diversity.

Particle settling within and deposition by a PDC ultimately affects its behaviour and runout distance. PDCs are driven largely by their density difference with respect to the atmosphere (e.g. Branney and Kokelaar 2002; Dufek et al. 2015; Lube et al. 2020). Eventually, there is sufficient particle settling and deposition such that the density of the current becomes less than that of the atmosphere; consequently, it lifts and ceases forward flow. This buoyancy-induced change in current behaviour can occur abruptly, as seen by sharp changes in distal effects on vegetation and observed by eyewitnesses (e.g. Lipman and Mullineaux 1981, pl. 1; Major et al. 2013; Waitt 2015).

Topography can strongly influence PDC behaviour. The dense, basal part of a PDC is commonly focused in valleys and topographic lows whereas the overlying turbulent part of a PDC can overrun ridges and divides. Slope gradient influences the turbulence of a current and affects its erosivity and depositional regime (e.g. Brand et al. 2016). Topographic ruggedness can affect flow direction and local flow detachment, causing a single current to exhibit spatially variable directionality (e.g. Lipman and Mullineaux 1981, pl. 1; Fisher 1990).

PDC deposit characteristics are influenced by initiation mechanism, the nature of particle transport by the current, and particle composition. Single-pulse events consist of both dense and dilute flow phases that can leave superposed deposits that are gradational, or they can leave singular deposits if the dilute-flow phase outruns or detaches from the dense-flow phase. Sediment concentrations in PDCs can fluctuate with distance as the current entrains, deposits, and elutriates sediment. As a result, deposits can have textures that range from non-stratified, non-graded, and poorly sorted, to well sorted, cross-bedded and cross-stratified. Pumice and lithic particles can exhibit different size and sorting characteristics both spatially and within an outcrop owing to their different densities. Co-PDC fall deposits are a unique depositional phase that represents fall from a plume that originated from a flow phase. PDCs can be stratified both vertically and laterally; thus, deposit textures and compositions can vary not only along a flow path but also at a single site during passage of the current, leaving varied morphologic and sedimentologic signatures.

Despite insufficient understanding of the details of particle transport and depositional processes by PDCs, broad and simplified relations can be established between deposit properties and flow character. From a lithofacies perspective, PDC deposits are broadly categorized as one of two types: ignimbrites and block-and-ash flows (e.g. Wilson 1986; Branney and Kokelaar 2002). An ignimbrite is defined as the deposit of a PDC rich in pumice and pumiceous ash with dominant particle compositions having densities less than 1000 kg m⁻³. In an ignimbrite, clasts of pumice and subordinate lithic fragments are supported in an ashy matrix of vitric shards and crystal fragments (Fig. 15). Some lithic fragments within ignimbrites may show a distinctive stretched-crack texture known as breadcrusting, indicative of magmatic fragments having skins that cooled rapidly but having interiors that were still hot and expansive. In contrast, a block-and-ash flow is typically dominated by dense, juvenile lithic fragments that are poorly to moderately vesicular and supported in a non-pumiceous ashy matrix (Fig. 16). They can also contain prismatically jointed clasts indicative of high fragment temperature.

PDC-deposit thicknesses and textures may show relations with topography. High-concentration flow phases tend to follow topographic lows, are focused along valleys, and can leave deposits that are a few to many metres thick. Deposits from high-concentration flows can form individual lobate deposits (Fig. 15a) or fans of debris at bases of lava domes or volcanoes (Fig. 15b). Valley-fill deposits thin laterally and terminate against valley walls or transition to deposits from a more dilute flow phase (e.g. Scott et al. 1996; Hildreth and Fierstein 2012; Brown and Andrews 2015). Thicknesses of deposits from high-concentration PDCs can reflect progressive aggradation of sediment during flow passage and are not necessarily representative of the depth of flow that passed. Progressive aggradation is particularly evident where chemical compositions of juvenile components change vertically within a deposit (e.g. Bacon 1983; Branney and Kokelaar 2002; Hildreth and Fierstein 2012) (Fig. 15c). Low-concentration PDCs are less constrained by topography and can flow well beyond the boundaries of channels and other topographic lows. Indeed, low-concentration currents can pass across rugged terrain surmounting multiple ridges and valleys (e.g. Fisher 1990). Deposits of these currents are thin (centimetres to tens of centimetres thick), and they can become thinner and finer with distance from source. Thicknesses of PDC deposits resulting from caldera collapse can vary from tens of centimetres to hundreds of metres, and stratigraphic sequences can exhibit time transgressions of currents having varied texture and thickness within a single large event. Particularly energetic PDCs that sweep across the landscape can leave deposits that are thin relative to deposit volume and coverage area (e.g. Hoblitt et al. 1981; Wilson 1986; Fisher et al. 1987). Non-stratified deposits related to concentrated parts of the current are commonly more prevalent and thicker along valley floors whereas thinner but stratified deposits from the same current may be found at higher elevations along valley walls and ridges (e.g. Scott et al. 1996; Charbonnier and Gertisser 2008).

Slope angle can influence longitudinal variations in deposit textures within a single current. Brand et al. (2016) showed that non-blast-origin PDCs from the 18 May 1980 eruption of Mount St Helens produced non-stratified, poorly sorted deposits on low-gradient slopes and generally stratified to cross-stratified deposits on steeper slopes leading from the volcano's crater, indicating that currents were more turbulent with less internal stratification on steeper-gradient slopes.

Because of the variety of initiation mechanisms, particle support mechanisms, and particle compositions, PDC deposits can exhibit a variety of textural characteristics and sedimentary structures (e.g. Branney and Kokelaar 2002; Brown and Andrews 2015). Ignimbrites are usually coarse grained and consist of angular to subangular lithic clasts and subangular to rounded pumice clasts, with clasts as large as metres in diameter, within an ashy matrix, whereas BAFs are dominated by angular to subangular lithic fragments up to metres in diameter in an ashy matrix (Figs 15 & 16). Both types of deposits are typically non-stratified and poorly sorted, although they can show textural variety (e.g. Wilson 1986; Branney and Kokelaar 2002; Hildreth and Fierstein 2012; Major et al. 2013; Brown and Andrews 2015; Douillet et al. 2015; Brand et al. 2016; Báez et al. 2020; Gillies et al. 2020).

Varying clast density within PDCs affects particle grading and sorting. Within ignimbrites, pumice and lithic clasts may be strongly segregated and exhibit different styles of grading (for example lithic clasts may be normally graded whereas pumice clasts may be inversely graded). However, clast segregation may be suppressed and particles non-graded. An extreme form of particle segregation in PDCs is recorded by lithic-rich, co-ignimbrite lag deposits (e.g. Druitt and Bacon 1986). Those deposits typically are coarse grained, fines depleted, lithic-and-crystal enriched, and they can extend several kilometres from a volcano. They are distinguished from fall or other lithic-rich deposits through evidence of lateral flow and association with ignimbrite deposits. They represent exceptional lithic-particle segregation and sedimentation within proximal reaches of a column-collapse PDC and are commonly associated with caldera-collapse events. Overall, the grading of lithic and pumice particles reflects relations between size and density (e.g. Wilson 1986; Choux and Druitt 2002). Within BAFs, lithic fragments may exhibit some grading, but they are most commonly non-sorted and non-graded.

Vesicularity of juvenile lithic particles can vary within PDC deposits. Highly vesicular particles within ignimbrites are the result of magmatic explosions or column collapse, whereas denser juvenile particles are more likely the result of BAFs induced by dome collapse or reflect inclusion of fragments of largely degassed magma. Other lithics may include conduit material entrained during ejection or eroded during flow.

Particle composition and texture have been used to infer volcanic processes. For example, some ignimbrite deposits associated with caldera collapse show sudden onset of abundant lithic fragments within vertical sections of deposits, especially in proximal deposits. This sudden onset of lithic material, especially of non-juvenile lithics, reflects abrupt entrainment of conduit wall rock within the eruption column. Its appearance is generally inferred to represent onset of ring-vent development and caldera collapse (e.g. Bacon 1983; Druitt and Bacon 1986; Scott et al. 1996). PDC deposits rich in vesicular, cauliflower-head-shaped clasts have been inferred to have formed by a boiling-over mechanism at Cotopaxi volcano (Ecuador) (Hall and Mothes 2008), Tungurahua volcano (Ecuador) (Hall et al. 2013; Rader et al. 2015), and Citlaltépetl volcano (Mexico) (Carrasco-Núñez and Rose 1995) among others, but may not be diagnostic of such an origin.

PDC deposits can display a variety of sedimentary structures. Ignimbrite and BAF deposits are commonly massively textured but may show diffuse stratification, indicative of unsteady flow behaviour or amalgamation of multiple pulses of flow (e.g. Branney and Kokelaar 2002; Brand et al. 2014). Clast compositions and basal boundaries may show evidence of substrate erosion, such as clasts and other debris clearly entrained from the substrate. Deposits may exhibit internal channel erosion and other cross-cutting relations indicative of self-channelization during sustained flow (e.g. Brand et al. 2014) or of time breaks between flows, soft-sediment deformation structures indicative of sediment fluidization, and fines-depleted pipe structures related to degassing (e.g. Branney and Kokelaar 2002; Douillet et al. 2015; Valentine et al. 2021). Lithic-rich BAF deposits (Fig. 16) may exhibit textures and structures similar to other breccias, and can be difficult to distinguish. However, BAF deposits contain clues to their origin. Uniform remanent magnetism of originally hot clasts, slightly reddish coloration owing to iron oxidation from percolating gases, an ashy or a less dense and more friable matrix owing to elutriation of some fines by gases and ingested air, evidence of pipes along which fine material has been removed, and inclusion of prismatically jointed clasts can signal that the deposit is from a PDC (e.g. Crandell 1987). Lahar deposits may sometimes contain pipe or dish structures related to escape of water (e.g. Scott et al. 1995), which can mimic gas-pipe structures in PDC deposits. But lahars may show small voids in the matrix related to air bubbles trapped within water-saturated sediment. Such voids are not present in PDC deposits. In some instances, deposit origin may be difficult to identify until or unless detailed field mapping is conducted. Fisher and Schmincke (1984) provide a table of deposit characteristics (table 11-3) that may help distinguish breccia-deposit origin.

Deposits from pyroclastic surges, in which transport and deposition are strongly influenced by turbulence, are commonly moderately sorted to well-sorted, distinctly stratified to cross-stratified and may show evidence of bedform development (e.g. Waitt 1981; Sigurdsson et al. 1984; Scott et al. 1996; Sulpizio et al. 2007; Brown and Andrews 2015; Brand et al. 2016) (Fig. 17). However, they can also be non-stratified if deposited rapidly (e.g. Fisher et al. 1987). These deposits typically form subordinate facies within ignimbrite and BAF successions, although some successions may be dominated by this type of deposit. Phreatomagmatic eruptions, in which near-surface magma interacts with water, can produce very energetic surges. Because of the interaction with water, the magma and conduit wall rock can become highly fragmented, and that fragmentation can produce abundant fine ash. Surge deposits from phreatomagmatic eruptions are commonly highly enriched in fine ash whereas deposits of surges caused by other mechanisms are commonly poorer in fine ash owing to elutriation caused by ingestion of air and magmatic gases.

Deposits of particularly high-energy PDCs, such as at Mount St Helens in 1980 and Lamington volcano in 1951, and especially those that result from directed explosions, exhibit characteristics of deposits resulting from flow, surge, and fall (Hoblitt et al. 1981; Fisher et al. 1987; Belousov et al. 2007). Like flow deposits, they can contain poorly sorted units that are thicker in topographically low areas and taper against valley walls, can contain juvenile rock fragments having variable density and vesicularity, and those fragments commonly constitute 50% or more of the particle composition. Like surge deposits, they can have units that are moderately sorted to well sorted and bedded to cross-bedded. Overall, the deposits are thin compared to the area they cover, typically are thinner on ridges and thicker in topographically low areas at any given distance from source, and are less sensitive to topography than pyroclastic flows in that they can be spread broadly over rugged topography. Like fall deposits, they become thinner and finer grained with distance from source. Topographic grain can affect deposit character depending on whether the current moves parallel or perpendicular to topographic trends (Fisher 1990).

High-energy PDC deposits are normally graded overall, but are commonly subdivided into a bipartite or tripartite stratigraphic layering with layers exhibiting variable grading, sorting, and stratification (e.g. Hoblitt et al. 1981; Waitt 1981; Fisher et al. 1987; Belousov et al. 2007, 2020; Major et al. 2013) (Fig. 18). Typically, they consist of a basal layer of friable, poorly sorted, non-graded to normally graded angular gravel and coarse sand (blocks, lapilli, and ash) admixed with soil and organic debris. This layer commonly has an erosive contact and may be smeared across the ground surface. It typically grades into a layer of moderately sorted, fines-depleted angular gravel and sand (lapilli and ash), which may be non-stratified to indistinctly bedded. These two layers may be stratigraphically distinct, or the basal layer may form the lower part of a single stratigraphic subunit. Commonly, the fines-depleted unit grades into, or transitions abruptly into, a less friable, poorly sorted layer rich in fine material and commonly consisting of a normally graded sand (ash), which may be non-stratified to laminated or distinctly bedded with evidence of translational bedforms. Commonly, but not always, this depositional sequence is capped with a normally graded sandy silt (fine to extremely fine ash) rich in accretionary lapilli. Contacts and stratigraphic relations indicate that these diverse layers were deposited rapidly by a single event.

PDCs can deposit sediment under a variety of temperatures. Those deposited at low temperatures (<550°C) are commonly loose to weakly compacted unless altered by hydrothermal activity or vapour-phase mineralization, and deposits have textural characteristics essentially identical to those at the time of deposition (Wilson 1986). They may show varying shades of pink and red coloration related to deposition of iron oxides from percolating gases (Wilson 1986). In contrast, high-temperature PDC deposits may be fully or partly indurated, a textural characteristic known as welding (e.g. Smith 1960; Wilson 1986). Welding results from cohesion and deformation of particles at temperatures above the glass-transition threshold, first at local contacts between particles and ultimately through complete fusion of particles by elimination of pore space. Deformed pumice particles known as fiamme are common in welded ignimbrites (e.g. Wilson 1986). The degree of welding in high-temperature PDC deposits is largely a function of chemical composition (which affects viscosity of glassy particles), temperature, and deposit thickness, which affects contact pressure at any point within the deposit and rate of cooling (Smith 1960; Wilson 1986; Brown and Andrews 2015). As a result, welding tends to be most intense within, but near the base of, an ignimbrite deposit. Exceptionally intense welding can lead to glassy deposits that are difficult to distinguish from lava flows. On steep slopes, intensely welded ignimbrites can continue to deform and flow, forming what are known as rheomorphic ignimbrites (Wilson 1986; Brown and Andrews 2015). In addition to fusing of particles by welding, vapour transfer through deposits can lead to crystallization within pore spaces and densification and induration of deposits. Textures such as columnar jointing, primarily attributed to cooling of welded deposits, can also develop in high-temperature, vapour-phase-altered, non-welded PDC deposits (Wilson 1986; Wright et al. 2011).

A lahar is a mobile, saturated mixture of water and sediment that flows swiftly along a channel that drains a volcano or a volcanically impacted landscape (Fig. 19). Unlike a water flood, a lahar is a coherent mixture of water and sediment that can transport abundant rocky and organic debris. As a result of its mass, momentum, and sediment competence, it poses a serious threat to people and property, and typically deposits a sheet or lobe of muddy sand and gravel, which ranges from many centimetres to many metres thick (e.g. Thouret et al. 2020).

A lahar spans a spectrum of sediment–water flows having varying sediment concentrations; thus, no single sediment concentration defines a lahar. Instead, a lahar is a flowing sediment–water mixture that is different from normal sediment-laden streamflow (Smith and Fritz 1989; Vallance and Iverson 2015). Lahar characteristics can evolve in time and over travel distance. Two basic styles of lahar are commonly identified: a debris-flow lahar, here called a type-1 lahar, contains roughly equal proportions of sediment and water and looks very much like flowing, wet cement. This type of flow is poorly sorted and transports sediment ranging in size from clay to boulders. A hyperconcentrated-flow lahar, here called a type-2 lahar, is more dilute and composed of more water than sediment (Pierson 2005). The proportion of water and sediment can vary broadly, and as a result a type 2 lahar takes on a more liquid-like appearance than does a type-1 lahar. A type-2 lahar is broadly transitional between sediment-laden streamflow and a type-1 lahar. A type-2 lahar is composed dominantly of moderately to poorly sorted sand but can transport larger particles as bedload.

A lahar can form in many ways, and it can occur both during and after an eruption (Fig. 20). A lahar commonly forms in one of five principal ways: (1) by scour and melt of snow and glacier ice during passage of a PDC to form a mixture of water and sediment; (2) by liquefaction and direct transformation of a volcanic debris avalanche; (3) by an explosive eruption or other mechanism that releases a crater lake to form a flood that erodes sediment; (4) by other floods that erode sediment (such as glacier outburst floods or when groundwater is released during an eruption or intrusion); and (5) by rainfall that erodes and mobilizes fresh volcanic ash (tephra) or triggers a shallow landslide from the flank of a volcano. On rare occasions, a lahar can be triggered by a phreatic eruption that directly ejects a mixture of water and sediment (e.g. Sasaki et al. 2016). A primary lahar is directly associated with co-eruptive processes, such as snow and ice melt by PDC or explosive ejection of a crater lake. A secondary lahar occurs after primary volcanic sediment has been deposited. It can occur during or after an eruption or during a quiescent period between eruptions. A secondary lahar results from remobilization of volcanic sediment by heavy rainfall, landslide, lake breach, or water released from a glacier, and it can occur days, weeks, months, or even centuries after initial sediment deposition. More rarely, a lahar can be initiated by an earthquake that generates a failure of a segment of a volcano or many shallow landslides that liquefy and coalesce into a larger flowing mass (e.g. Scott et al. 2001; Worni et al. 2012). Close association with an eruption, but not direct association with eruptive processes, can blur distinction between primary and secondary lahars and make interpretation of the geological record challenging. For example, the North Fork Toutle River lahar caused by the 1980 eruption of Mount St Helens was not associated directly with the eruption, but rather formed over a span of hours as the large debris avalanche locally dewatered. Thus, the original source sediment for this lahar was temporarily stored and subsequently remobilized. Yet, in the geological record, it will largely be interpreted as a primary lahar owing to its close association with the eruption. Broader sedimentologic and geomorphic context and relative position and timing in the stratigraphic record must be scrutinized when deciphering primary from secondary origins of lahars.

Initiation mechanism affects the nature, size, and composition of a lahar. A lahar formed during an explosive eruption by snow-and-ice melt, by release of a crater lake, by failure of wet, weak rock from a volcano, or by sudden release of water from a large lake dammed by volcanic debris commonly is large, fast, and very destructive far downstream from a volcano. Lahars formed by rainfall erosion of volcanic ash or by release of modest volumes of subglacial water are likely to be smaller and travel shorter distances, but they may occur more frequently. These smaller but more frequent lahars can gradually fill river channels close to volcanoes with thick amounts of sediment. A lahar formed when groundwater is released during an eruption or intrusion can vary in size depending upon the volume of water released. A secondary lahar formed by transformation of a shallow landslide commonly is small (typically to a few hundreds of thousands of cubic metres) compared to an eruption-triggered lahar and is largely restricted to an area close to a volcano.

Pyroclastic density currents commonly trigger lahars at glaciated volcanoes. These hot currents sweep across snow- and ice-covered slopes, scour and mix with snow and glacier ice, and produce watery floods or slurries. As these floods sweep downhill, they erode additional sediment from the volcano's flanks and surrounding river channels, grow in volume, and become sediment-rich lahars. Notable lahars of this type have formed during many eruptions (e.g. Wolf 1878; Gorshkov 1959; Murai 1960; Gorshkov and Dubik 1970; Pierson 1985; Eppler 1987; Fairchild 1987; Scott 1988a; Major and Newhall 1989; Pierson et al. 1990; Dorava and Meyer 1994; Branney and Gilbert 1995; Mothes et al. 1998; Vallance et al. 2010; Waythomas et al. 2013; Waythomas 2015).

Lahars caused by PDCs scouring and melting snow and ice range widely in size (Pierson 1985; Pierson et al. 1990; Vallance et al. 2010; Waythomas et al. 2013). Some are relatively small and localized (Vallance et al. 2010) whereas others can be as large as hundreds of millions of cubic metres (e.g. Worni et al. 2012; Waythomas et al. 2013). At Mount St Helens in 1980, PDC-generated lahars had volumes of about 15×10⁶ m³ and travelled many tens of kilometres down valley (Fairchild and Wigmosta 1983; Pierson 1985; Scott 1988a). At Nevado del Ruiz (Colombia) in 1985, small PDCs swiftly melted snow and ice on the volcano's summit and produced abundant meltwater (Naranjo et al. 1986). The resulting floods eroded sediment from several steep, narrowly confined valleys, and transformed to lahars ranging in volume from 1–40×10⁶ m³ (Pierson et al. 1990). Those lahars descended nearly 5000 m in elevation from the summit and travelled more than 100 km (Pierson et al. 1990). During the 2009 eruption of Redoubt volcano (USA), PDC-triggered lahars had volumes of 10⁷–10⁸ m³ (Waythomas et al. 2013). Field studies at Cotopaxi volcano revealed that the Holocene Chillos Valley Lahar, possibly formed in part by PDC melt of the volcano's icecap, had a volume of c. 3.8 km³, flowed more than 300 km from the volcano, and had local depths of 80–160 m (Mothes et al. 1998).

A volcanic debris avalanche can produce a lahar directly if the avalanche is particularly wet, clay rich (many weight-percent clay), and transforms from a slide to a flow as it sweeps downslope. Lahars caused by transformations of debris avalanches are not known to be common, but, where documented, they have been large, commonly on the order of 0.1–1 km³ (MacPhail 1973; Carrasco-Núñez et al. 1993; Vallance and Scott 1997; Capra and Macías 2000; Detienne et al. 2017). A classic example of this type of lahar is the Osceola Mudflow from Mount Rainier (USA), which formed during an eruption 5600 years ago. High on the volcano's flank, weak, water-saturated rock collapsed. That material transformed to a deep, valley-filling lahar within a few kilometres of its source. This lahar, nearly 4 km³ in volume, funnelled into and filled mountain valleys 75–150 m deep, travelled more than 100 km to Puget Sound (Washington, USA), and continued some 15 km underwater (Vallance and Scott 1997).

Most large debris avalanches from collapses of volcanoes do not transform immediately, if at all, to lahars. For example, the 2.5 km³ debris avalanche that initiated the 18 May 1980 eruption of Mount St Helens slid off the volcano and came to a complete stop within 30 km. Then, over the next hours, water-saturated parts of its deposit liquefied and began to flow (Janda et al. 1981; Fairchild and Wigmosta 1983). Several small flows coalesced to produce the 140×10⁶ m³ North Fork Toutle River lahar. That lahar was the largest of the 1980 eruption, flowing some 100 km from the volcano (Janda et al. 1981; Schuster 1983).

Heavy rainfall and snowmelt can trigger relatively small (10³–10⁵ m³) landslides at volcanoes. Such meteorologically triggered landslides are generally shallow slope-failures (a few tens of metres thick). Unlike large debris avalanches, these smaller landslides can transform to lahars that range from a few thousands to a few hundreds of thousands of cubic metres in volume. For example, intense rainfalls at San Vicente volcano (El Salvador), Mount Hood (USA), and Izu Oshima volcano (Japan) triggered shallow landslides that evolved into lahars (Gallino and Pierson 1985; Major et al. 2004; Miyabuchi et al. 2015; Smith et al. 2015).

Floods resulting from mechanisms other than by scour of snow and ice during an eruption can also entrain sediment and transform into lahars. These floods can form by release of water from a summit crater lake, breaching of a valley-margin lake dammed by volcanic sediment, release of abundant groundwater from a volcano, or sudden release of stored water from a volcano glacier (e.g. Suryo and Clarke 1985; Scott 1988b; Umbal and Rodolfo 1996; Cronin et al. 1997a; Capra and Macías 2002; Rodolfo and Umbal 2008; Massey et al. 2010; Worni et al. 2012; Gudmundsson 2015; Pagneaux et al. 2015; Johnson et al. 2018). Depending on the volume of water released and the rate at which it is released, these lahars can vary in size, speed, and travel distance. A 2008 phreatomagmatic eruption of Huila volcano (Colombia) triggered an estimated 300×10⁶ m³ lahar (Worni et al. 2012; Pulgarín et al. 2015). Subsequent inspection showed the glacier on the volcano's west flank to be heavily fractured, but the exact source of the large amount of water needed to produce this lahar is not entirely clear (Worni et al. 2012; Pulgarín et al. 2015).

The largest known lahar at Mount St Helens occurred about 2500 years ago when a large lake dammed by volcanic debris breached its blockage. Sudden release of lake water produced a series of flood surges that entrained channel sediment through bed incision and lateral erosion over many kilometres of valley. The largest lahar had a volume of about 1 km³ (Scott 1988b) and inundated now-urbanized areas 80–100 km downstream with a coarse-sediment-rich flow to depths of 5–10 m (Chan 2008). The largest lahar at Mount St Helens since its great 1980 eruption happened when a temporary meltwater lake formed and spilled from the crater. In 1982, an explosion from a growing lava dome sprayed hot rock across the volcano's crater walls, which melted snowpack and formed a transient lake (Waitt et al. 1983). Released lake water produced a flood that eroded sediment and transformed into a 15×10⁶ m³ lahar that flowed at least 80 km downstream (Pierson 1999).

Release of water from existing crater lakes can occur during an eruption. Kelut volcano (Indonesia) is notorious for expelling water from its crater lake during eruptions and generating large floods that scour sediment from the volcano's flanks and form lahars. Such lahars occurred in 1919 and 1966 during explosive eruptions through its crater lake (Suryo and Clarke 1985). Drainage tunnels now limit the volume of the crater lake, reducing the amount of lake water available to form lahars. A month-long series of explosions through the crater lake at Ruapehu volcano (New Zealand) in 1995 emptied the lake, causing some 26 lahars having a cumulative volume of 10⁶ m³ at a distance 56 km from source (Cronin et al. 1997b).

Crater lakes can also release water during periods of inactivity. During dormant periods between eruptions, the summit crater lake at Ruapehu volcano has breached its volcanic sediment dam (formed by heavy ash fall during eruptions) and produced notable lahars. The most recent instance in 2007 released 1.3×10⁶ m³ of lake water (Procter et al. 2010). The consequent flood mobilized roughly 3×10⁶ m³ of sediment from the initial 5 km of downstream channel, forming a nearly 4.5×10⁶ m³ lahar. As this lahar travelled farther downstream, it alternately eroded and deposited sediment along the channel, which maintained a nearly 3×10⁶ m³ lahar for more than 60 km (Massey et al. 2010; Procter et al. 2010).

A sudden release of water stored within or beneath glaciers or released from moraine-dammed lakes can also produce flood surges that spawn lahars. Glacier-outburst floods can occur during eruptions when subglacial lava flows, pyroclastic eruptions, or increased heat flux melt glacier ice. But they can also occur during non-eruptive times, such as during spells of hot weather or heavy rainfall when subglacial and intraglacial storage cavities link, pressurize, and release. In Iceland, for example, lahars – or at least sediment-laden floods – can form from large glacier outburst floods (jökulhlaups), which are common when eruptions occur beneath massive glacial ice caps that overlie many of its volcanoes (Gudmundsson 2015). Lahars caused by glacier-outburst floods also occur on much smaller scales. In the Cascade Range (USA), small (typically 10⁴ m³) lahars formed by outburst floods are common. For example, outburst-flood-triggered lahars occur frequently at Mount Rainier and Mount Shasta, and have also been witnessed at Glacier Peak and Mount Hood (Richardson 1968; Crandell 1971; Walder and Driedger 1994, 1995; Blodgett et al. 1996). Most of these lahars travel only a few kilometres. Other interactions between glaciers and streams can also trigger small lahars. At Mount Rainier, for example, a meltwater stream along the margin of a glacier spilled through a notch in a moraine, eroded and entrained sediment, and transformed into a small lahar (Vallance et al. 2002). Glacial-lake-outburst floods occur when a moraine-dammed lake is released through dam failure or displacement of lake water over the dam or when a summit meltwater lake is displaced by collapsing ice and debris. Similar to breachings by lakes dammed by volcanic debris, these breakout floods can cause lahars of varying size depending on volume and rate of water released (e.g. Coombs et al. 2006; George et al. 2019).

Heavy rainfall on freshly deposited volcanic ashfall can promote surface runoff that can lead to flash-flood-like events that transform into lahars by eroding and entraining sediment from both hillsides and river channels (e.g. Waldron 1967; Barclay et al. 2007; de Bélizal et al. 2013; Pierson and Major 2014). These types of lahars occur frequently and can persist for many years, especially if the landscape is repeatedly recharged with ashfall or deposits from PDCs during prolonged eruptions (e.g. Waldron 1967; Barclay et al. 2007). Although individual lahars may not be large, their frequency can have significant cumulative effects downstream; cumulated deposits can overwhelm channels and cause rivers to change course (e.g. Crittenden and Rodolfo 2002; Pierson et al. 2013). Furthermore, rainfall lahars can occur well after eruptions end if rainfall intensities, durations, and sediment supply are sufficient (e.g. Crittenden and Rodolfo 2002; Capra et al. 2018; Tsunetaka et al. 2021).

Rainfall-triggered lahars are distributed across many environmental settings, from tropical- to high-latitude volcanoes. The most devastating rainfall-generated lahars in modern times occurred at Mount Pinatubo (Philippines) during and following its 1991 eruption. There, multiple drainages around the volcano were affected for many years (Pierson et al. 1996; Umbal 1997; Crittenden and Rodolfo 2002). Notable rainfall lahars, many in just the past few decades, have been documented in varied environmental settings including at tropical volcanoes (Waldron 1967; Rodolfo and Arguden 1991; Thouret et al. 1998; Harris et al. 2006; Barclay et al. 2007; Paguican et al. 2009; Capra et al. 2010; Doyle et al. 2011; de Bélizal et al. 2013; Vázquez et al. 2014; Dibyosaputro et al. 2015; Cando-Jácome and Martínez-Graña 2019), low-latitude, high-altitude volcanoes (Jones et al. 2015), mid-latitude volcanoes (Hodgson and Manville 1999; Miyabuchi 1999; Pierson et al. 2013; Miyabuchi et al. 2015; Kataoka et al. 2018; Hayes et al. 2019; Mosbrucker et al. 2019; Baumann et al. 2020), and even heavily glacier-clad, high-latitude volcanoes (Jensen et al. 2013).

Conceptual understanding of lahar deposition has evolved over the past several decades. In the late 1960s and 1970s, Johnson (1970) proposed that non-volcanic debris flows could be modelled as Bingham fluids having yield strength. As long as applied stresses exceeded the inherent yield strength of the mixture, it remained in motion. When stresses dropped below the inherent yield strength of the mixture (for example when flows reached shallow gradients or spilled across flood plains) the flow stopped en masse, essentially ‘freezing’ in place. In this view of deposition, the deposit reflects the characteristics of the flow at a given instant in time. In the 1980s, field examinations of lahar deposits in the aftermath of eruptions of Mount St Helens and studies of ancient lahar deposits began to document vertical variations in deposit textures that were incompatible with a model of flow instantaneously freezing in place (e.g. Pierson and Scott 1985; Vallance and Scott 1997). In addition, large-scale flume experiments with debris flows in the 1990s showed that complex deposition by waves of flow could produce deposits that appeared to have textural characteristics identical to those attributed to en masse instantaneous deposition (e.g. Major 1997). These observations and experiments, coupled with emerging ideas regarding deposition by PDCs (Branney and Kokelaar 1992, 2002), spurred the hypothesis that lahars can deposit sediment progressively from flow-front to tail. Thus, deposits can reflect progressive changes of flow character over time and not simply a snapshot of flow character at an instant in time (e.g. Vallance and Iverson 2015; Fig. 21).

Many lahars evolve by varying their sediment load through erosion or deposition of sediment. Although nearly all lahars are the result of sediment erosion, evidence of the sources of sediment and how those sources evolve along flow paths may be subtle or obvious depending on the textural and lithological characteristics of the entrained sediment. For example, a lahar that forms on the flank of volcano will commonly contain large proportions of angular to subangular clasts composed of volcano lithologies. But as it travels downstream, a lahar can erode and entrain channel sediment that is dominantly subrounded to rounded, composed of lithologies not found on the volcano, and include intact stratigraphic sections of bank sediment. As a result, the amounts of rounded and exotic particles in its deposit can provide information on the degree of sediment entrainment along the flow path and offer clues to its origin and transport behaviour (e.g. Scott 1988a, b; Vallance and Scott 1997; Thouret et al. 1998; Capra et al. 2002; Scott et al. 2005). Aside from entraining sediment, a lahar can mix with streamflow, deposit sediment, and evolve toward a less sediment-rich flow with distance, ultimately transforming into a type-2 lahar or sediment-laden flood (Fig. 22). A flow that contains less than a few percent clay-sized or clay mineral particles is more likely to transform into a type-2 lahar or sediment-laden water flow with distance and can do so comparatively quickly. In contrast, a flow that contains a greater amount of clay-sized and clay mineral particles commonly maintains its textural character as a type-1 lahar over many tens of kilometres of travel distance.

A lahar deposit can exhibit a variety of characteristics related not only to flow initiation mechanism but also to the ability of the flow to interact with the channel and streamflow and evolve along its flow path. As a lahar moves downstream, it can entrain and deposit sediment, sometimes repeatedly, allowing its volume and composition to fluctuate with distance (e.g. Scott 1988a; Pierson et al. 1990; Hodgson and Manville 1999; Capra et al. 2010; Procter et al. 2010; Doyle et al. 2011). Furthermore, flow duration at a given point along a channel can vary from swift passage as a flash flood to prolonged flow at maximum stage, and characteristics of the passing flow can vary from head to tail (Vázquez et al. 2014; Vallance and Iverson 2015). Consequently, a lahar can leave a deposit that has morphologic and sedimentologic characteristics that vary both in space and time.

Deposit thickness can vary widely along the flow path, is greatly influenced by flow initiation process and topography, and commonly is thin in relation to flow depth. Flood-plain deposit thickness can vary from centimetres to a few metres. The channel deposit, however, results largely from progressive accumulation during flow passage and its thickness can vary considerably, from a few to many metres. Initiation process can greatly affect deposit thickness. For example, a flow that results from transformation of a debris avalanche or from a break-out flood from a large lake, which can entrain immense amounts of sediment, can form a deep flow that leaves a thick (many metres) deposit. In contrast, rainfall-runoff generated lahars are shallow and although cumulative deposit thicknesses from multiple flows can be substantial, individual deposits are thin (metre-scale). Topography, however, commonly exerts a greater control on deposit thickness than does flow genesis. Along unchannelled fans at the bases of volcanoes, or beyond the mouths of canyons, flows can spread widely and leave thin deposits. In more confined reaches, thick deposits can accumulate on channel floors, flood plains, and terraces. Metre-thick deposits of type-1 lahars on flood plains and terraces many tens of kilometres from volcanoes attest to great sizes and mobilities of many lahars.

Deposit thickness, however, provides an incomplete picture of the nature of a flow. Flow depths of moderate to large flows can be much greater than deposit thickness (Fig. 23). Deposits of smaller flows may also be misleading indicators of original flow depth. As noted, deposits can accumulate progressively, yet appear to be deposited en masse. Caution should be exercised when inferring relations between deposit thickness, flow depth, and the nature of the depositional process.

Textures of lahar deposits can vary considerably depending on sediment concentrations and compositions of lahar flows. Nevertheless, there are many common and distinctive traits. A type-1 lahar deposit is commonly non-stratified, poorly sorted, and consists of particles ranging in size from clay to boulders (Fig. 24). Clasts larger than about 2 mm are commonly supported within a matrix of finer particles, especially in flood-plain deposits. Locally, coarse clasts may be in clast-to-clast contact, especially at deposit margins. Channel deposits of a type-1 lahar are typically coarser grained than flood-plain deposits and may exhibit abundant clast-contact texture that may look similar to gravel-rich flood deposits. Clast angularity can provide clues to the dominant transport process; gravel-rich flood deposits will be composed mainly of rounded particles whereas those resulting from lahars will likely be composed of angular or a mix of angular and rounded particles. However, if the dominant source of sediment for a lahar is channel sediment (such as for a lahar triggered by a lake breakout along valley margins), then channel and flood-plain deposits can be dominated by rounded particles, and other evidence is needed to decipher deposit origin.

Clasts within a type-1 lahar deposit can exhibit wide variation in size, grading, and dispersal (Fig. 24). Some deposits may consist of a relatively narrow range of sizes, with the coarsest clasts only a few to several centimetres in diameter, whereas others may contain clasts many tens of centimetres to a metre or more in diameter. Coarse clasts may be normally graded, fining upward, or inversely graded, coarsening upward. Deposits can exhibit inverse grading of clasts near the base, little grading in the middle section, and normal grading above. In contrast, the finer (<2 mm) matrix particles commonly exhibit little grading and are distributed uniformly through the deposit. Internal sorting and grain migration allow large clasts to accumulate along flow fronts and margins producing bouldery snouts and levees (Johnson et al. 2012; Vallance and Iverson 2015). Vertical grading of clasts can reflect longitudinal variations in flow composition (see Fig. 21). Thus, vertical variations in clast grading may be influenced more by progressive aggradation than by internal sorting processes.

Clast shapes and compositions are indicative of predominant sediment sources. Angular to subangular clasts composed of volcano lithologies are indicative of lahars forming at a volcano or of flows that entrained sediment from older primary volcanic deposits. In contrast, clasts that are dominantly rounded to subrounded and/or composed of mixtures of volcano lithologies and exotic lithologies are indicative that stream alluvium was an important sediment source.

Clasts in type-1 lahar deposits can also consist of large intact fragments of older deposits. Such clasts, commonly referred to as fragile megaclasts, may be composed of fragments of sediment entrained from a single deposit, or of stratigraphic sections entrained largely through bank erosion (Fig. 25). Fragile megaclasts, in conjunction with other deposit characteristics, have been inferred as evidence that the lahar entrained substantial channel sediment along its flow path (Scott 1988b; Major and Scott 1988). In some instances, entrained fragments of debris-avalanche sediment provide possible evidence for lake blockage by a debris-avalanche deposit (Fig. 25a, d).

In contrast to the compositions and textures of a type-1 lahar deposit, a type-2 lahar deposit is composed dominantly of moderately sorted sand. Because type-2 lahars span a range of sediment concentrations, their deposits can exhibit a variety of textures. They can consist of non-stratified to laminated and stratified sand (Fig. 26) or of moderate to poorly sorted mixtures of sand and small gravels. A key feature of type-2 lahars is that they transport large amounts of sand both in suspension and as bed load (Pierson and Scott 1985; Cronin et al. 1999; Pierson 2005). The coarse sediment in transport is commonly deposited rapidly and progressively whereas the fines (silt and clay) remain in suspension and are transported farther downstream. Consequently, resulting deposits consist mainly of moderately sorted sand (Pierson and Scott 1985; Cronin et al. 1999; Pierson 2005; Pierson et al. 2013; Wilcox et al. 2014). Deposits may show some vertical grading of sand-sized particles. Dispersed lithic or pumice clasts may be sporadically embedded within type-2 lahar deposits, or they can be abundant, producing fines-depleted gravel lenses. Pebble-sized clasts largely reflect particles transported as bed load.

Deposit textures can evolve down valley if a lahar undergoes distal transformation. If a lahar progressively entrains sediment, it can evolve from a low-concentration flood or type-2 lahar to a type-1 lahar. In contrast, a lahar that mixes with streamflow can drop sediment and transform from a type-1 to type-2 lahar and ultimately to sediment-laden streamflow. As a result, deposit characteristics can change longitudinally from non-stratified, poorly sorted gravelly sand that contains dispersed large clasts to crudely stratified, moderately sorted sand lacking coarse clasts (e.g. Pierson and Scott 1985), and ultimately to well-stratified, cross-bedded, well-sorted sand indicative of alluvial transport. Deposits may show inconsistent longitudinal variations in texture if flows undergo episodic erosion and deposition along transport paths (e.g. Procter et al. 2010).

Deposit textures, clast shapes and compositions, and the characteristics of megaclasts provide an array of information regarding lahar initiation and transport processes. The amounts of rounded clasts and exotic lithologies can be used to distinguish a lahar that entrained sediment at or near a volcano from one that largely scavenged sediment from channels and entrained bed and bank sediment farther along its flow path. A lahar that eroded sediment at or near a volcano contains abundant angular debris of volcanic lithologies, whereas one caused by a flood that entrained abundant channel sediment commonly contains dominantly rounded alluvium composed of a mix of volcano and exotic lithologies. It may also contain megaclasts of alluvium. Although a lahar that forms at or near a volcano can contain entrained stream gravels, the amount of such clasts typically is subordinate. Such a criterion for distinguishing initiation mechanism is generalized, however. For example, at Mount St Helens in 1980, a large debris-avalanche deposit blocked the outlet to a large lake at the foot of the volcano. Had that lake breached and released a large flood surge that ultimately transformed into a lahar, that surge would have entrained sediment from the nearly 30 km-long debris-avalanche deposit – a deposit composed predominantly of angular volcanic debris. The distal lahar deposit would therefore be composed largely of angular to subangular volcanic clasts. This hypothetical texture is considerably different from the lahar deposit related to an ancient breakout of the same lake, a breakout that also followed blockage by a debris-avalanche deposit (Scott 1988b). That ancestral lake-breakout lahar deposit is dominated by rounded volcanic clasts that represent stream alluvium (Fig. 24c). The abundance of entrained stream alluvium and paucity of angular clasts indicates that the predominant sediment source for that ancient lahar was the stream channel beyond the debris-avalanche blockage. That predominant sediment source suggests that the ancient debris-avalanche(s) deposit that blocked the lake was not as extensive down valley as the 1980 debris-avalanche deposit.

Amounts and types of clay within lahar deposits have been used to distinguish possible origins of lahars. A lahar that forms at a volcano as a result of a large slope failure is more likely to contain greater amounts of clay-sized and clay-mineral material than one formed by another mechanism. A lahar containing a few to several percent clay-sized and clay-mineral material commonly contains volcanic debris that may be hydrothermally altered or debris from which the finest particles had not been sorted. Abundant hydrothermally altered clay minerals may provide evidence for a debris-avalanche origin of a lahar (Scott et al. 1995). In contrast, a lahar formed from sediment entrainment by runoff and flood erosion typically contains little clay, which indicates that the source sediment contained little clay. Channel-sediment entrainment by flood surges, transformations of small surficial landslides, and remobilization of tephra fall by rainfall runoff are some of the most likely causes of such lahars. Clay-poor lahars commonly transform toward dilute sediment-laden floods as they move down valley (e.g. Zehfuss et al. 2003). Although some clay-poor lahars can maintain their type-1 flow character for many tens of kilometres, others transform to type-2 lahars or sediment-laden floods over distances of a few to a few tens of kilometres (e.g. Scott 1988a).

In summary, multiple lines of evidence, including clast compositions, shapes, percentages of volcanic v. non-volcanic lithologies, morphologic and stratigraphic position, and sedimentary textures such as grain size, sorting, grading and bedforms must all be considered when assessing potential initiation mechanisms.

The most widespread volcaniclastic deposits result from tephra fall – the rain of particles from volcanic plumes that drift downwind (Fig. 27). These particles consist of angular fragments of rock, pumice, crystals, and glass. The size distribution of particles within tephra-fall deposits, and local deposit thicknesses, can range broadly as functions of mass eruption rate, distance from a volcano, particle aggregation, particle composition and density, wind speeds and directions, and atmospheric dynamics at various altitudes.

Fall deposits provide information about volcanic eruptions and eruptive processes. They record eruptions and thus help delineate eruptive histories, record changes in magmatic compositions within and between eruptive periods, and serve as crucial regional time-stratigraphic marker beds (Mullineaux 1986). Grain characteristics (size, shape, and density) and textural compositions provide insights into fragmentation and collisional processes operating within volcanic conduits (e.g. Dufek et al. 2012) as well as processes operating within volcanic plumes (e.g. Van Eaton et al. 2015). Fall-deposit preservation, and its reliability as a time-stratigraphic marker and eruption record, is affected by depositional environment. Nevertheless, preservation of primary fall can be substantial, especially within tens of kilometres of a volcano. Although tephra fall can induce significant changes to a landscape's hydrogeomorphic regime and result in substantial erosion (see the ‘Posteruption sediment erosion, transport, and deposition’ section), recent studies have estimated that some 50–80% of primary fall can remain uneroded within proximal watersheds for at least decades or longer (Collins and Dunne 1986, 2019; Manville and Wilson 2004; Pierson et al. 2013). Even thin tephra-fall deposits (less than 30 cm thick) can be well preserved at great distances downwind for centuries or longer (Blong et al. 2017). But tephra preservation becomes less consistent the greater the distance from a volcano and under variable surface covers (e.g. Cutler et al. 2018). Tephra preservation is common in peatlands and lakes (e.g. Hardardóttir et al. 2001; Kuehn and Negrini 2010; Moreno et al. 2015; Jensen et al. 2021), but whereas peatlands typically preserve primary tephra (and cryptotephra) fall, deposits in lakes reflect a combination of primary fall and reworked tephra washed in from erosion of the surrounding catchment (e.g. Hardardóttir et al. 2001; Watson et al. 2016). Tephra falls are also preserved in continental glaciers, notably in the Greenland Ice Sheet and Antarctica, and provide not only time-stratigraphic horizons within the ice but also facilitate correlations of various climatic archives (Abbott and Davies 2012; Jensen et al. 2021; Narcisi and Petit 2021).

Tephra fall is the outcome of competing influences of volcanic plume behaviour, particle characteristics, atmospheric dynamics, and the nature of volcanic processes. Volcanic plumes fall into three major classes – those driven by volatile exsolution and fragmentation of magma, those resulting from near-surface interactions of magma and water, and those resulting from material lofting from PDCs (Bonadonna et al. 2015; Carey and Bursik 2015). Plume character represents interactions between eruption (e.g. plume rise velocity) and wind characteristics (Bonadonna et al. 2015). When plume rise velocity exceeds horizontal wind velocity, a strong vertical column develops and feeds an umbrella cloud that advects downwind at the level of neutral buoyancy (Figs 27a, c & 28a). As a result, downwind fall deposits are broadly distributed in a wide, elongate pattern, perhaps with some upwind deposition. In contrast, weak plumes form when horizontal wind velocities exceed plume rise velocities (Figs 27b & 28b). Weak plumes commonly develop narrow, elongate fall deposits with little upwind deposition.

Sedimentation from volcanic plumes is affected by particle settling, aggregation, and plume dynamics (Bursik et al. 1992; Bonadonna et al. 2015). In general, tephra fall results from sedimentation along the margins of a rising turbulent volcanic plume and settling from a horizontally drifting cloud. It can also result from co-PDC ashclouds. Particle settling is related to particle size, shape, and density. Relations between particle settling velocities and plume vertical velocities cause particle segregation within a plume. Larger and denser particles separate quickly from the plume and fall near the vent whereas smaller and less dense particles are carried higher in the plume and farther downwind. Strong density differences between pumice and lithic particles greatly affect their fall behaviour. Particle aggregation affects fall of fine ash. Moisture within a volcanic plume causes fine ash to cluster in dense, coherent aggregates known as accretionary lapilli. Electrostatic forces can draw fine ash together into porous, fragile dry aggregates. Aggregation increases the effective size of particles within a volcanic plume and causes fine ash to fall out much more rapidly than it would otherwise. Aggregation of fine ash can be effective at removing these particles from a plume. During the 2009 eruption of Redoubt volcano, Van Eaton et al. (2015) estimated that more than 95% of deposited fine ash fell as aggregates and less than 5% remained in the plume as single particles. Eruption dynamics, such as mass eruption rate, eruption duration, total grain-size distribution within the plume, downwind plume dynamics such as gravitational instabilities, and atmospheric dynamics affected by local topography also affect tephra fall (e.g. Bonadonna et al. 2015; Watt et al. 2015).

Fall deposits blanket the landscape and typically fine and thin exponentially with distance from source (Pyle 1989; Houghton and Carey 2015), an observation consistent with predictions from tephra sedimentation models (Bursik et al. 1992; Bonadonna et al. 2015). Relations between fall-deposit thickness and distribution, grain size, and grain-size distribution are commonly used to estimate mass eruption rates, plume heights, and eruption volumes (e.g. Carey and Sparks 1986; Pyle 1989; Fierstein and Nathenson 1992; Bonadonna and Costa 2013). However, these relations are affected by variations in eruption source parameters, aggregation of fine particles, wind speeds, variations in plume conditions, interactions among sediment sources (e.g. fallout from central volcanic plumes co-mingled with fallout from a co-PDC plume), and deposit preservation (Sigurdsson and Carey 1989; Sparks et al. 1992; Eychenne et al. 2012; Engwell et al. 2013; Bonadonna et al. 2015; Houghton and Carey 2015; Van Eaton et al. 2015).

Characteristics of fall deposits represent competing influences of volcanic plume behaviour, wind and atmospheric characteristics, and the nature of volcanic processes. As a result, fall deposits can exhibit a variety of bedding and textural characteristics (Fig. 29). Unlike PDC deposits, fall deposits are generally well sorted and drape the pre-existing topography uniformly (Fig. 29b, f), except where locally thickened or thinned by erosion. Topography exerts little influence on deposit distribution (except where it influences atmospheric turbulence and fallout from a volcanic plume; Watt et al. 2015). Depending on proximity to source, the nature of the source, eruption vigour, and wind and atmospheric characteristics, fall deposits can range from dominantly coarse lapilli, blocks, and bombs to micron-sized ash. Fall deposits can also range from non-stratified to finely stratified and exhibit normal to inverse size grading (e.g. Houghton and Carey 2015). They can also be diffusely stratified or sharply bedded. Fall deposits are distinguished from pyroclastic surge beds by a lack of internal directional bedding such as cross-stratification or bedforms (Fig. 29).

Accretionary lapilli are formed by wet aggregation of fine ash within a volcanic plume. Water content and residence time of aggregates within a plume affect the size and character of accretionary lapilli (Bonadonna et al. 2015; Van Eaton et al. 2015). The modal size of aggregated particles in accretionary lapilli is typically around 0.03–0.06 mm (4φ–5φ) (Bonadonna et al. 2015), whereas accretionary lapilli are commonly a few millimetres in diameter. Accretionary lapilli (Fig. 29g) are typically spherical, internally massive to weakly layered, and display concentric layering of ash particles (e.g. Bonadonna et al. 2015; Brown and Andrews 2015; Van Eaton et al. 2015). Because they are fragile and often have an ice-particle nucleus, they can break apart upon impact and are thus poorly preserved overall. Dry aggregates formed by electrostatic forces are typically not preserved in deposits; they are observed in rare instances when ash is collected while actively falling (e.g. Sorem 1982; Taddeucci et al. 2011) or inferred from grain-size character of the deposit (e.g. Carey and Sigurdsson 1982). Deviations from exponential deposit thinning with distance from source and anomalous abundance of fine ash in deposits relative to nominal particle-settling velocities may be related to particle aggregation (e.g. Brazier et al. 1983).

Unlike PDC deposits, tephra-fall deposits are generally cool at the time of deposition. As a result, tephra-fall deposits are usually not welded. Those that are welded are most commonly the result of deposition of spatter from basaltic and intermediate-composition eruptions (Houghton and Carey 2015). In those instances, welding and agglutination provide an indication of proximity to the source vent.

Fall-deposit textures provide an array of information about depositional process and eruption characteristics. For example, basic differences in fall-deposit compositions – such as pumiceous fall v. lithic- and crystal-rich fall – have been used to discriminate fall resulting from magmatic explosions v. fall resulting from phreatic explosions or co-PDC ash clouds (e.g. Scott and McGimsey 1994). Average maximum sizes of pumice and lithic fragments are commonly used to estimate plume height (e.g. Carey and Sparks 1986). The presence of fine-scale stratification within fall deposits is used to distinguish fallout from non-sustained eruptions having mass eruption rates that wax and wane v. those from sustained eruptions with constant mass eruption rate, which produce more uniformly textured deposits (Houghton and Carey 2015). Sharp bedding contacts and abrupt changes in grain size have been inferred by some as representing pulsating eruptive behaviour and by others as indicative of column collapses that have generated PDCs, with deposition of fine ash by dilute currents or by co-PDC ashfall (e.g. Paladio-Melosantos et al. 1996; Houghton and Carey 2015). Thin, fine-ash partings have been used to infer pauses during an eruption, because fine ash can remain suspended for days; its presence may indicate a pause during which time it settles (Houghton and Carey 2015). Varying amounts of very fine ash within proximal fall deposits may lend insights into dry v. wet eruptions – with large amounts indicative of wet eruptions thought to have greater fragmentation efficiency (Houghton and Carey 2015). Deposit sorting may also provide insights on dry v. wet eruptions. Because wet eruptions contain larger amounts of very fine ash, their proximal fall and surge deposits are more poorly sorted (commonly σ_φ>2) compared to those of dry eruptions (commonly σ_φ∼1–1.5) (e.g. Houghton and Carey 2015). Sorting characteristics and bimodality of grain-size populations in fall deposits have been used to infer that fall deposits reflect synchronous deposition from different volcanic processes (Eychenne et al. 2012). Relations among median grain size, sorting, and deposit area have been proposed as ways of distinguishing deposits of different volcanic processes (Walker 1971) as well as distinguishing styles of eruption (Walker 1973). Although there is broad correlation among median particle size, sorting, deposit footprint characteristics and depositional process, subtle complexities preclude discrimination of process solely by these characteristics (Houghton and Carey 2015). Overall, a key question to be answered in the field is whether a deposit results from a flow or fall process. Uniform topographic draping, nearly universal particle angularity, generally well to moderate sorting, and a clear lack of palaeoflow indicators are common hallmarks of fall deposits.

The volcanic processes discussed can broadly modify landscapes and disrupt normal hydrogeomorphic functioning (Fig. 30). Much of the following discussion is modified from Pierson and Major (2014). Explosive eruptions affect the hydrological functioning of watersheds in three basic ways: (1) they damage or remove vegetation, which decreases (or eliminates) interception and evaporation of precipitation; (2) volcaniclastic deposits on hillsides commonly reduce surface infiltration, which increases overland flow; and (3) large injections of valley-floor sediment alter hydraulic properties of river channels and enable efficient transport of water and sediment (Pierson and Major 2014). Alterations to the hydrogeomorphic functioning of landscapes by explosive eruptions thus affect the routes and rates of precipitation runoff, which in turn affect erosion and sediment transport.

Volcanic processes can remove, damage, or bury vast tracts of vegetation. Trees can be toppled or defoliated, and understorey can be damaged or buried (e.g. Dale et al. 2005; Ayris and Delmelle 2012; Swanson et al. 2013, 2016). As a result, more precipitation in the form of rainfall or snowfall reaches the ground surface and, in the case of rainfall, impacts the surface with greater force. Tree canopies intercept, on average, 10–40% of incoming precipitation (Reid and Lewis 2009; Carlyle-Moses and Gash 2011). Furthermore, loss of or damage to vegetation reduces transpiration of soil moisture, which alters subsurface storage and flow of water.

Tephra-fall and PDC deposits draping hillsides commonly alter the rate at which precipitation infiltrates the ground surface (e.g. Teramoto et al. 2006; Jones et al. 2017; Tarasenko et al. 2019). The finer the surface of the deposited sediment, the greater the loss of surface infiltration (Pierson and Major 2014; Fig. 31). As a result, more of the precipitation that reaches the ground surface is partitioned into overland surface flow as opposed to shallow subsurface flow. This change in partitioning of runoff promotes hillside erosion and allows more water to reach river channels faster. These changes in hydrogeomorphic functioning promote the occurrence of larger floods and lahars after an eruption (e.g. Todesco and Todini 2004; Favalli et al. 2006; Major and Mark 2006; Alexander et al. 2010; also see the ‘Rainfall-runoff erosion of volcanic ash and other pyroclastic sediment’ section).

Volcaniclastic deposits affect river channels in various ways. Extensive channel deposition by debris avalanches and large PDCs smothers valleys and disrupts channel networks (Fig. 30). In those instances, unchannelled runoff must concentrate and carve new channels to re-integrate channel networks. Complete integration of channel networks can take years to accomplish (e.g. Daag and van Westen 1996; Simon 1999). Early posteruption channels are commonly straighter, wider, and steeper than their pre-disturbance counterparts (Meyer and Martinson 1989; Gran and Montgomery 2005). In contrast, lahars commonly do not fully bury river channels. Rather, they strip river corridors of vegetation, straighten channels, and pave channel beds with large loads of sand making them hydraulically smoother (Janda et al. 1984; Pierson and Major 2014). Such changes enhance the efficiency with which rivers can transport exceptional posteruption sediment loads.

Sediment reworked after an explosive eruption comes from two basic sources: (1) hillsides where sheet and rill erosion as well as shallow landslides mobilize tephra-fall and PDC deposits, and (2) channels where debris-avalanche, PDC, and lahar deposits are reworked as channels reestablish or react to these perturbations. The sediment eroded from these two sources is delivered from watersheds at different rates and persists for different durations (Major et al. 2000; Gran et al. 2011). In addition to freshly deposited sediment, older hillside and channel sediments can be remobilized providing additional sediment supply (e.g. Waldron 1967; Pierson et al. 1990, 1996; Swanson and Major 2005; Korup et al. 2019).

Owing to loss and damage of vegetation and to changes in the hydrological regime, sheet and rill erosion are the dominant processes that mobilize hillside volcaniclastic sediment (Segerstrom 1950; Waldron 1967; Kadomura et al. 1983; Swanson et al. 1983; Chinen 1986; Collins and Dunne 1986; Takeshita 1987; Leavesley et al. 1989; Shimokawa et al. 1989, 1996; Yamakoshi et al. 2002; Waythomas et al. 2010; Pierson et al. 2013). Sediment erosion from hillsides is typically acute and rapid, but once a rill network is established, rates of erosion diminish swiftly. Diminished erosion occurs once rills incise into coarser, more permeable layers or reach resistant substrates (Collins and Dunne 1986). Once hillside incision ceases, rill networks evolve toward fewer active rills (Swanson et al. 1983; Collins and Dunne 2019). Biogenic and cryogenic processes along with wind deflation can coarsen the surface and improve infiltration, which reduces or eliminates surface runoff (e.g. Yamakoshi and Suwa 2000; Major and Yamakoshi 2005). As a result, initially high sediment yields owing to hillside erosion can decrease sharply within just a couple of years, even in the absence of revegetation (Chinen 1986; Collins and Dunne 1986, 2019). However, at some volcanoes, such as Sakurajima (Japan) and Santiaguito/Santa Maria (Guatemala), persistent recharge of tephra-fall deposits by frequent eruptions can maintain high rates of hillside erosion and sediment yield for long durations (e.g. Shimokawa et al. 1989; Harris et al. 2006).

Rapid reduction of hillside erosion permits large volumes of proximal tephra fall to remain in place. Approximately 80–90% of the tephra fall deposited by eruptions of Usu volcano (Japan) in 1977–78 and Mount St Helens in 1980 is estimated to remain in place decades after deposition (Kadomura et al. 1983; Chinen 1986; Collins and Dunne 1986, 2019; Smith and Swanson 1987). As much as 50% or more of tephra fall deposited in proximal areas following eruptions of Irazú volcano (Costa Rica) (1963–65) and Chaitén volcano (Chile) (2008) appeared to remain in place (Waldron 1967; Pierson et al. 2013).

Although rill erosion of tephra-fall deposits diminishes rapidly, shallow landslides can emerge as an effective process delivering sediment to channels. Some landslides occur shortly after an eruption (Swanson et al. 1983; Smith and Swanson 1987) as an immediate response to loss of vegetation, deposition of fresh sediment, and increased precipitation throughfall. Others, however, are delayed and can mobilize not only fresh tephra fall, but also older tephra-fall deposits (e.g. Swanson and Major 2005; Korup et al. 2019). The efficacy of landslides to mobilize tephra-fall deposits depends on a number of factors, including recurrence and duration of intense precipitation, compositions and structure of hillside strata, and a competition between the timescales of root decay of trees killed by an eruption, reducing strength within deposits, and regrowth of new vegetation sufficient to anchor tephra-fall deposits (Swanson and Major 2005; Korup et al. 2019).

Erosion of valley-filling deposits causes greater-magnitude and more prolonged posteruption sediment delivery than does erosion of hillside tephra or thin valley deposits. This difference in erosion response occurs because rivers can deeply incise and greatly widen channels as they adjust to posteruption water and sediment fluxes. Sediment supply commonly exceeds river transport capacity, and thus sediment shifts about causing channel instabilities that perpetuate channel widening and lateral channel migration. In general, posteruption channels experience complex sequences of incision, aggradation, and widening (e.g. Meyer and Martinson 1989; Daag and van Westen 1996; Simon and Thorne 1996; Gran and Montgomery 2005; Ulloa et al. 2016; Major et al. 2019).

Initial erosion of valley-filling deposits can be rapid and dramatic. Following the great eruptions of Mount St Helens (1980) and Mount Pinatubo (1991), channels carved into debris-avalanche and PDC deposits were incised tens of metres and widened hundreds of metres within a year (Meyer and Martinson 1989; Daag and van Westen 1996; Simon 1999; Major et al. 2019). Following the 1982 eruption of El Chichón (Mexico), valleys buried thickly by PDC deposits were incised up to 20 m, most of that incision occurring within months after the eruption (Inbar et al. 2001). But channels need not be thickly buried to experience substantial hydrogeomorphic responses. Even channels relatively thinly paved by lahar deposits can incise by metres and widen by many tens of metres owing to channel instabilities associated with altered water and sediment discharges (e.g. Meyer and Martinson 1989; Lavigne 2004).

Although channel erosion is most dramatic within the first few years after an eruption, persistent channel instabilities and geomorphic adjustments can prolong notable channel erosion for years or decades. Such prolonged channel adjustment, which Gran et al. (2011) referred to as phase II adjustment, or the phase that follows initially rapid hillside and channel erosion (Fig. 32), is the result of persistent mining of channel bed and bank sediment. Persistent mining of bank sediment is a result of incision having created tall banks susceptible to undercutting and mass failure along channels that are highly mobile (Gran 2012; Major et al. 2018). Indeed, major sediment sources along rivers in the Cascade Range (USA) are Holocene terraces composed of glacial and lahar sediment subject to small, frequent mass movements induced by inexorable lateral erosion (Scott and Collins 2021).

Posteruption erosion of volcanically disturbed landscapes increases sediment transport. Because of induced hydrogeomorphic changes, sediment-transport processes and channel geometries evolve and adjust to convey the supplied sediment load. Commonly, sediment concentrations increase over a broad range of discharges (Dinehart 1998), lahars and high-sediment-transporting floods become more prevalent, and channels become smoother, steeper, and straighter to convey these sediment loads (Pierson and Major 2014).

Local climate affects the sediment transporting processes. In tropical and subtropical climates, secondary lahars become more frequent and may continue for many years (Waldron 1967; Rodolfo 1989; Umbal 1997; Suwa and Yamakoshi 1999; Lavigne et al. 2000; Lavigne and Suwa 2004; Gran and Montgomery 2005; Harris et al. 2006; Barclay et al. 2007; Cinque and Robustelli 2009). In temperate, non-tropical climates, early lahars may occur, but fluvial processes subsequently dominate (Major 2004; Manville et al. 2009b; Pierson et al. 2011, 2013; Major et al. 2016).

The proportion of volcanic sediment output from a watershed v. that stored within a watershed can be highly variable. As noted above, large amounts of deposited tephra fall commonly remain in hillside storage. Abundant valley sediment can also remain in storage if the area of deposition greatly exceeds the fluvial footprint of channel systems. Large amounts of reworked volcaniclastic sediment commonly move among storage areas within watersheds rather than being exported rapidly (e.g. Pierson et al. 1992). The proportion of volcanic sediment output from a watershed v. that moved into storage reflects differences in volcanic process, the amount of sediment input to watersheds, watershed size, hydrology, and the distances and average gradients between the source volcano and watershed outlet (Davies et al. 1977; Pierson et al. 1992; Manville et al. 2009b; Pierson and Major 2014).

Despite differences in sediment output and storage among volcanic landscapes, posteruption sediment yields (mass output per unit watershed area) can be extraordinary. Erosion and transport of fresh volcanic sediment, as well as of older sediment from storage, can generate sediment yields that rival those of Earth's greatest sediment-transporting rivers. Common values range from 10³–10⁷ Mg km⁻² (Fig. 33). When converted to average rates of landscape denudation, erosion rates of volcanically disturbed landscapes typically exceed those of other landscapes by 3–4 orders of magnitude (see Pierson and Major 2014, supplemental table 1). Even though releases of reworked sediment from volcanically disturbed landscapes are largely transitory, they can still dominate geomorphic and sedimentologic functioning of landscapes over decadal- to century-scale timeframes – timeframes critical to human societies (Umbal 1997; Manville and Wilson 2004; Cinque and Robustelli 2009; Manville et al. 2009b; Gran et al. 2011; Pierson et al. 2011; Major et al. 2020).

Durations of exceptional sediment yield from volcanically disturbed landscapes vary with nature of disturbance. Delivery from disturbed channels is commonly greater and more prolonged than that from disturbed hillsides. Nevertheless, exceptional sediment delivery diminishes rapidly within a few years of disturbance (Chinen 1986; Simon 1999; Suwa and Yamakoshi 1999; Major et al. 2000, 2016, 2021; Lavigne 2004; Yamakoshi et al. 2005; Gran et al. 2011) (Fig. 34). Even so, ongoing adjustments of disturbed channels can maintain prolonged, elevated sediment delivery for decades (e.g. Major et al. 2021), and modelling projections suggest they can last for centuries (e.g. Meadows 2014).

Changes to the hydrogeomorphic regime of a volcanically disturbed landscape trigger geomorphic adjustments in fluvial systems. Extraordinary sediment mobility can induce net storage where sediment supply exceeds transport capacity. That accumulation of sediment induces channel aggradation (e.g. Smith 1987; Gran and Montgomery 2005; Kataoka et al. 2009; Manville et al. 2009b; Pierson et al. 2011, 2013; Zheng et al. 2014; Major et al. 2021), which affects channel pattern. When sediment supply wanes, the balance between supply and transport capacity adjusts and streams can re-incise channel beds (e.g. Gran and Montgomery 2005; Cinque and Robustelli 2009; Pierson et al. 2011; Major et al. 2019). Channels can undergo cycles of aggradation and incision as diffuse waves of sediment pass through fluvial systems (e.g. Janda et al. 1984; Tanarro et al. 2010). Channel aggradation can occur rapidly – hours to days – especially when sediment is transported by secondary lahars or sediment-laden floods (e.g. Punongbayan et al. 1996; Lavigne et al. 2000; Pierson et al. 2013; Fig. 35). Channel aggradation and degradation are not synchronous along the longitudinal profile; rather, they reflect interactions among channel morphology, streamflow, sediment supply, sediment size, and sediment transport (e.g. Lisle et al. 2001; Pierson and Major 2014; Major et al. 2021).

Aggradation of volcanic sediment induces changes in channel patterns. In particular, aggradational channels invariably evolve from single-thread to braided channel patterns (e.g. Davies et al. 1977; Kuenzi et al. 1979; Janda et al. 1984; Smith 1987; Manville et al. 2009a, b; Gran 2012; Ulloa et al. 2015) (Fig. 36). Accumulations of large woody debris, often transported by lahars and PDCs, can strongly influence sediment storage, channel patterns, and variations in sediment composition and geomorphic processes (Lisle 1995; Swanson et al. 2013; Ulloa et al. 2014, 2015; Umazano et al. 2014; Umazano and Melchor 2020).

Lahars and floods become more prevalent after an eruption. The duration and dominance of lahars is a function of climate, local hydrology, durations of eruptive activity, and the distributions and characteristics of deposited volcaniclastic sediment. Following many eruptions, initial phases of hydrogeomorphic responses are dominated by an increase in sediment transport by lahars (e.g. Waldron 1967; Smith 1987; Rodolfo and Arguden 1991; Umbal 1997; Lavigne et al. 2000; Perrotta et al. 2006; Barclay et al. 2007; Manville et al. 2009a). Following other eruptions, posteruption lahars are rare and sediment transport is dominated largely or exclusively by fluvial processes (e.g. Janda et al. 1984; Major 2004; Pierson et al. 2011, 2013). The initial posteruption phase of transport by lahars commonly evolves toward dominance by fluvial transport. Indeed, observations following modern eruptions and investigations of ancient stratigraphic sequences reveal abrupt to gradual transitions from mass-flow to fluvial-transport processes (e.g. Smith 1987; Gran and Montgomery 2005; Perrotta et al. 2006; Manville et al. 2009a; Pierson et al. 2013; Major et al. 2016).

Posteruption volcaniclastic sequences exhibit a variety of deposit textures and sedimentary structures. Lahar deposits exhibit the textural characteristics typical of type-1 and type-2 lahar flows, whereas fluvial, fluvial-lacustrine, and fluvial-deltaic deposits exhibit a broad range of characteristics and sequencing from sand-rich to gravel-rich, crudely stratified to cross-bedded, well- to moderately sorted, and with variable degrees of clast imbrication and evidence of bedforms (e.g. Davies et al. 1977; Kuenzi et al. 1979; Vessell and Davies 1981; Smith 1987; White and Riggs 2001; Friele et al. 2005; Kataoka et al. 2009; Manville et al. 2009b; Sohn et al. 2013). Pumice clasts within fluvial deposits are typically rounded. In distal settings, tens to hundreds of kilometres from volcanic sources, posteruptive sediment deposits can be many metres thick (e.g. Kataoka et al. 2009; Pierson et al. 2011; Pierson and Major 2014). In contrast, primary deposits at these distances commonly range from a few millimetres or centimetres (fall deposits) to a few metres thick (lahars). Within tens of kilometres, debris-avalanche deposits may be many metres to tens of metres thick; in rare instances they may be similarly thick at greater distances (e.g. Stoopes and Sheridan 1992). At distances of thousands of kilometres, only trace amounts of cryptotephra may be preserved (e.g. Jensen et al. 2021).

Magnitudes of posteruption aggradation can vary widely. The magnitude of aggradation is affected not only by sediment supply, but also by local channel morphology. As noted by Pierson and Major (2014), peak aggradation levels have ranged from a few metres to nearly 40 m in channel reaches up to 100 km from source, with aggradation on the order of 5–10 m common. Except where massive sediment inputs are involved, aggradation levels that exceed 20 m are typically limited to confined valleys (e.g. Pierson et al. 2011). Beyond about 50 km from source, channel aggradation is caused mainly by fluvial deposition; at lesser distances both fluvial and lahar processes contribute.

In the geological record, vertical and lateral sequences of volcaniclastic deposits commonly exhibit a transition from primary to secondary deposits, illustrating the initial sediment input from eruptions followed by consequent deposit reworking (Fig. 37). To capture the importance of volcanism-induced sedimentation in the geological record, Smith (1991) proposed that facies sequences in volcanic settings can be divided into two fundamental conditions: syneruptive periods and intereruptive periods. Syneruptive periods represent episodes that produce large volumes of volcaniclastic sediment. This sediment production is driven by frequent volcanism and, in response, by the occurrence of lahars and floods that result from alterations to hydrogeomorphic regimes. These geologically brief but intense periods of sediment production are characterized by a general lack of lithologic diversity, deposits rich in sand, and large lateral extents of lahar and flood deposits. In contrast, intereruptive periods represent longer times when volcanism has had little detectable impact on the landscape or on the character of fluvial systems. During intereruptive periods, sediment delivery is greatly diminished, normal streamflow processes (seasonal floods and infrequent large floods) dominate, deposits exhibit greater lithological diversity as contributions from multiple parts of the landscape are averaged over time, and they are comparatively gravel rich as predominantly bed-load-transported sediment is preserved. Intereruptive deposits are commonly thinner and more spatially restricted than syneruptive deposits because they are often confined to valleys incised into syneruptive deposits.

Volcanism affects sedimentation on a variety of scales in both space and time. Volcaniclastic sediment can mantle, modify, or create new topography at the landscape scale and alter the ‘normal’ hydrogeomorphic functioning of the landscape for years, decades, and sometimes millennia. Indeed, in some settings, volcanism may be sufficiently frequent to drive a landscape into a perpetual state of disequilibrium with regard to hydrogeomorphic functioning (repeated cycles of perturbation and response) for hundreds to thousands of years. As a result, volcanism can have an outsized impact on sedimentation in continental settings for extended periods of time. For example, Friele et al. (2005) showed that a disproportionate amount of sediment within the Lillooet River basin (Canada) was derived from Mount Meager volcano, a massif that constitutes only 2% of the area of the drainage basin. Manville and Wilson (2004) showed that the sedimentary response to the 530 km³ (DRE) 26.5 ka Oruanui eruption of Taupo volcano produced a massive downstream response, including sufficient aggradation along 180 km of a major river system to trigger avulsion into a different watershed. The scale of response was driven by the character of the eruption, but the duration of response, which lasted some 10–12 000 years, was influenced and prolonged by suppression of revegetation during harsh periglacial climate conditions that accompanied the Last Glacial Maximum. Smith (1991) showed that volcaniclastic sediments in river systems draining volcanic arcs dominate stratigraphic sequences during periods of active volcanism and can form distinctive facies sequences and geometries that may be used to lend insights into the relative importance of volcanism and tectonics on arc-basin sedimentation. Recognition of volcanism-induced sedimentation in the stratigraphic record and the relative influences of volcanism, tectonism, and climate on facies sequences and geometries are key challenges for sedimentary geologists and volcanologists (e.g. Fisher and Smith 1991).

At a local scale, volcaniclastic sedimentation not only affects the landscape, but also poses a variety of hazards to society. Ensembles of volcanic processes strongly influence the nature of volcaniclastic deposition and its societal and environmental impacts. Volcanoes can shed vast amounts of clastic debris that can fill and smooth topography across reaches extending a few kilometres to many tens of kilometres. Volcaniclastic fills can be highly complex, with deposits from numerous processes intercalated, eroded, reworked, and redeposited (Fig. 38). Proximal stores of deposits tend to be thick, relatively coarse grained, and composed of variable compositions and textures. They provide a rich archive of the eruptive histories and hazards of a volcano, but they can be challenging to accurately decipher. Some deposits may be buried and not exposed whereas others may represent isolated fragments of their initial extent. With increasing distance from a volcano, volcaniclastic deposits commonly are composed of thin, fine-grained tephra falls or thick sequences of lahar and fluvially reworked sediment. Deposits along river valleys may exhibit cycles of deposition and erosion, and deposits from certain volcanic processes, such as lahars, may be easier to decipher because they are less susceptible to being trapped within the jumble of proximal processes and deposits. However, the distal geological record is incomplete and is biased toward larger events capable of inundating flood plains and terraces. Distal environments are also the store of abundant remobilized sediment accumulation, and are thus further biased toward preservation of fluvial, fluvial-lacustrine, and fluvial-deltaic sediment. Furthermore, specific information regarding the nature and character of proximal primary deposits may be obscured in distal deposits. Nevertheless, some characteristics of distal deposits, such as inclusion of pumice pebbles and cobbles, may provide insights into generalized primary processes such as occurrence of PDCs (e.g. Kataoka 2005). Distal deposits, although providing a lens through which to view partial histories of some of the more significant eruptions and eruptive impacts, are clearly inadequate for understanding the full complexity and hazards posed by a volcano. In some instances, a distal deposit preserved in the geological record may be non-representative of the hazard posed. For example, during the 2009 eruption of Redoubt volcano explosions early in the eruption sequence triggered a large, ice-rich lahar that travelled more than 40 km, had local flow depths of 6–8 m, left deposits as much as 5 m thick, and threatened a critical oil-storage and transfer facility (Waythomas et al. 2013). However, once the ice melted, the record of that event was an inconspicuous silty sand no more than a few centimetres thick – a deposit that might be easily overlooked or misunderstood (Waythomas 2014). Although that deposit reflects the sediment load transported by the lahar, it is not representative of the hazard posed by the lahar or by eruptions from this ice-clad volcano. Similarly, PDC deposits from directed explosions and surges may not adequately reveal the hazards posed by those processes. Correct interpretation of volcaniclastic deposits and recognition of volcanic processes is imperative for understanding both eruptive histories and the hazards posed by volcanoes.

Interpretations of volcaniclastic deposits and the processes that produce them have advanced considerably in the past few decades. Since the eruption of Mount St Helens in 1980, volcanic debris avalanches have become recognized as a common, and oft-repeated, process at volcanoes worldwide. Their morphological and sedimentological characteristics are now well defined, their causative mechanisms much better understood, assessments of their frequency improved, and modelling of their flow behaviour is advancing (e.g. Roverato et al. 2021). The 1980 Mount St Helens eruption also clearly elucidated the catastrophic nature of directed volcanic blasts, an association with volcanic debris avalanches, and the unique nature of their deposits (Lipman and Mullineaux 1981; Belousov et al. 2007). Study of subsequent eruptions of Soufrière Hills volcano and reinvestigations of deposits from the 1956 eruption of Bezymianny volcano, the 1951 eruption of Lamington volcano, and the 1964 eruption of Shiveluch volcano (Russia) have further refined the sedimentological signature of directed-blast deposits (Bogoyavlenskaya et al. 1985; Sparks et al. 2002; Belousov et al. 2007, 2020). Continued advances in numerical modelling have convincingly demonstrated that although such spatially directed PDCs may be propelled initially by rapid expansion of volatiles and fragmentation of magma, they quickly collapse into high-energy gravity-driven currents (e.g. Esposti Ongaro 2012). Even small-scale directed explosions, such as occurred at Chaitén volcano in 2008, produce deposits with characteristic directed-explosion signatures. The 1980 Mount St Helens eruption along with the 1985 eruption of Nevado del Ruiz volcano and 1991 eruption of Mount Pinatubo sharpened understanding of the devastating nature of lahars, highlighted their causative mechanisms at snow-clad volcanoes, illustrated associations with debris avalanches, re-emphasized associations with rainfall, greatly improved our understanding of their propensity to evolve in space and time along their flow paths, and reinforced that even communities far from volcanoes can be vulnerable to devasting volcanic impacts (e.g. Janda et al. 1981; Pierson et al. 1990; Newhall and Punongbayan 1996; Vallance and Iverson 2015). Since those eruptions, physical understanding of the mechanics of multiphase mixtures has advanced considerably and sophisticated numerical models can now predict depth, velocity, and impact forces of lahars with substantial accuracy as well as the dynamics of explosive eruptions and characteristics of PDCs (e.g. George and Iverson 2014; Iverson and George 2014; Neri et al. 2022). The 1982 eruption of El Chichón volcano, the 1991 eruptions of Mount Pinatubo and Unzen volcanoes, and the 1995 and later eruptions of Soufrière Hills volcano further highlighted the impacts and hazards of PDCs – from column collapse, caldera collapse, and dome collapse (e.g. Sigurdsson et al. 1984; Scott et al. 1996; Calder et al. 1999; Miyabuchi 1999; Sparks et al. 2002; Carn et al. 2004). Modern technological observations of those eruptions and others – both optically and instrumentally – have helped to better link the characteristics of deposits with the nature of the volcanic eruptions. Eruptions of Soufrière Hills volcano, Merapi volcano (Indonesia), Colima volcano (Mexico), and Santiaguito/Santa Maria volcano (Guatemala) have highlighted the hazards and deposit characteristics of PDCs (block-and-ash flows) associated with protracted growth and collapse of lava domes as well as the propensity for the occurrence of secondary lahars resulting from rainfall runoff (e.g. Rose et al. 1976; Calder et al. 1999; Saucedo et al. 2002; Lavigne 2004; Harris et al. 2006; Barclay et al. 2007; Charbonnier and Gertisser 2008; Capra et al. 2018). Eruptions of Galunggung volcano (Indonesia) in 1982, Redoubt volcano in 1989, Mount Pinatubo in 1991, and Eyjafjallajökull volcano (Iceland) in 2010 highlighted the long-range hazards of volcanic plumes to aviation and the impacts of volcanic aerosols on climate (Robock 2000; Guffanti et al. 2010; Guffanti and Tupper 2015). Studies of fall deposits from several eruptions in the past decades have highlighted aggregation of extremely fine ash within volcanic plumes and have identified causative processes and refined modelling of volcanic plumes and tephra fall. Characteristics of fall deposits are now used to make robust estimates of eruption characteristics such as mass eruption rates and plume heights. Several eruptions of the past few decades have highlighted the hydrological changes that can follow tephra fall. These hydrological changes can lead to widespread erosion and formation of secondary lahars and floods. Of particular note is that hazardous mobilization of sediment can be triggered by unremarkable rainfalls and that such mobilization can happen swiftly after rainfall begins (Pierson et al. 2013). Greater recognition that releases of crater and valley-marginal lakes, both during eruptions and intereruption periods, can influence volcaniclastic sedimentation has sharpened appreciation for ancillary hazards associated with volcanoes (e.g. Scott 1988b; White et al. 1997; Manville et al. 2007; Capra et al. 2010; Manville 2010; Massey et al. 2010). Many of the advances in deposit and process interpretation have allowed refined understanding of past historical and prehistoric eruptions and a greater understanding of the histories and hazards of many volcanoes.

The past few decades have also garnered greater appreciation for the impacts and hazards associated with posteruption remobilization of volcaniclastic sediment. Indeed, we are beginning to appreciate that after eruptions end, some of society's most difficult challenges may just begin, especially for communities distant from volcanoes. Volcanically disturbed landscapes can generate some of the world's greatest sediment releases, and even though extraordinary releases diminish rapidly, elevated releases can endure for years to decades, and in rare cases millennia. Posteruption sediment redistribution can be one of the greatest and costliest challenges society must confront in volcanic regions; indeed, after some eruptions, posteruption sediment redistribution can cause greater social and economic harm than the direct impacts of the eruptions themselves.

Eruption processes, deposit textures and compositions, depositional environments, and climatic regimes affect deposit preservation. Eruptions can spawn ensembles of processes that produce volcaniclastic deposits on a variety of scales. Deposit volumes can range from as little as a few hundreds or thousands of cubic metres to as much as a few thousands of cubic kilometres during exceptionally rare super-eruptions. Areas affected by flowage deposits can range from a few to a few tens of thousands of square kilometres; fall deposits can affect greater areas. Deposit thicknesses can range from trace amounts to tens or hundreds of metres. Volcaniclastic deposits are also highly erodible. Deposits that are loosely textured and friable are more apt to be easily eroded than are deposits having denser textures or deposits that have been welded. Deposits that contain abundant lithic clasts may erode and leave armors of winnowed clasts that curtail further erosion, whereas loose, sandy, pumice-rich deposits may easily erode and ultimately preserve little of the original deposit. Deposit textures can also influence preservation. Tephra-fall deposits in proximal areas are commonly composed of coarse sediment (typically medium to coarse ash and lapilli) overlain by finer ash. This pavement of finer ash can severely alter the characteristics of rainfall and snowmelt runoff and lead to erosion of the tephra mantle (e.g. Chinen 1986; Collins and Dunne 1986; Németh and Cronin 2007; Ogawa et al. 2007; Pierson et al. 2013; Engel et al. 2021). However, after rills and gullies have eroded into the coarser underlying tephra fall, erosion can cease, leaving much of the original proximal tephra-fall deposit in place (e.g. Collins and Dunne 1986, 2019). These variations in deposit volume, area, thickness, and texture can greatly affect the preservation potential of volcaniclastic deposits.

Depositional environments and climate regimes also affect preservation potential. Proximal deposits from eruptions of glaciated volcanoes and from volcanoes in wet, tropical climates generally have lower preservation potential than do those in arid environments. In arid environments, wind erosion and aeolian transport can rework primary deposits, creating aeolian deposits that must be carefully distinguished from (surge) PDC deposits (e.g. Smith and Katzman 1991). Consequently, variations in preservation potential can severely skew not only our understanding of eruptive histories of volcanoes and their eruptive processes, but also the perceptions of hazards their eruptions pose. For example, at Ruapehu volcano Gillies et al. (2020) showed that deposits from small- to medium-volume PDCs have low preservation potential, particularly on the steep, glaciated flanks of the volcano. This poor preservation has created an incomplete eruptive record of the volcano. From the limited preservation of these PDC deposits, Gillies et al. (2020) concluded that they formed from column collapse and dome collapse or explosion events. Hence, Ruapehu volcano produces a broader spectrum of PDC styles and sizes than has previously been inferred. At Mount Hood (USA), the Polallie eruptive period occurred c. 12–15 ka when glaciers at the volcano were more extensive. Topographic positions of proximal deposits of PDCs and lahars from that eruptive period are determined largely by the extent of glacier ice in valleys at the time. Those topographic positions range from deposits primarily on ridgetops exposed when glaciers filled valleys, to deposits plastered on valley sides after glaciers had shrunk, and to valley floors beyond the limits of the glaciers (Crandell 1980). The proximal record of this eruptive period is thus poorer than its distal record. In distal settings, deposit preservation is affected by depositional environment, such as channel v. flood plain, as well as by deposit thickness, and in modern times by societal actions such as dredging and channel mining.

The degree of deposit preservation affects the types of questions one can address. Variations in preservation of proximal deposits affect understanding of eruptive histories and hazards. Although distal accumulations of remobilized volcaniclastic sediment generally have high preservation potential owing to their thickness and lateral extent, they typically disaggregate information about specific volcanic processes upstream. Thus, they preserve records of volcanism and allow general questions regarding periods of syneruption v. intereruption to be addressed but can limit understanding of the timing of events and specific volcanic processes active during periods of eruption.

Volcaniclastic sedimentation has an outsized geomorphic and sedimentologic impact on proximal drainage basins and river channels downstream. Deposits of volcaniclastic flows and falls can mantle, modify, or create new topography, and can adversely affect communities many tens to hundreds of kilometres downstream and downwind of volcanoes. Posteruption erosion and sediment remobilization can endure for decades or longer, sometimes causing more social and economic harm than the direct impacts of eruptions themselves. In the last four decades, especially since the 1980 eruption of Mount St Helens, studies of volcanic processes that generate volcaniclastic sediment have blossomed (e.g. Manville et al. 2009a). As a result, our understanding of deposit character and their linkages to initiation mechanisms, transport, and depositional processes have increased immensely. Major subaerial volcaniclastic processes, including debris avalanches, pyroclastic density currents (PDCs), lahars, and tephra fall, can produce deposits with widely ranging sedimentologic characteristics. Yet, those deposits have diagnostic characteristics that can point toward deposit provenance. Debris-avalanche deposits show contextual and diagnostic association with transport of pieces of a volcano. PDC deposits exhibit characteristics that point toward hot, dry flowage emplacement and initiation by column collapse, directed explosions, or failure of lava domes or lava flows. Lahar deposits show evidence of transportation as saturated mass flows and their sedimentological and morphological characteristics point toward initiation mechanism (e.g. snowmelt triggered, transformation from a debris avalanche, lake-breakout triggered). Tephra-fall deposits exhibit characteristics indicative of fall not flow, some of which can be used to assess plume height, mass eruption rate, and relative wet v. dry eruptions. Posteruption analyses of deposits from several modern eruptions, progress in our understanding of the physical behaviour of multiphase mixtures, greater insights on the physical interactions within PDCs and volcanic plumes, and improvements in physical and numerical modelling have vastly enhanced our understanding of volcanic processes, interpretations of eruptive histories, and the hazards posed by volcanic eruptions. Each study of deposits from a new eruption, reanalysis of a past eruption, experimental interrogation of a volcaniclastic process, and analysis of hydrogeomorphic response to volcanic disturbance of landscapes further contributes to and refines our store of knowledge.

In this chapter, I have highlighted the characteristics of major subaerial volcaniclastic deposits and the influence of initiation mechanisms and transport processes on the character, storage, and preservation of those deposits. This work summarizes and builds on an immense body of literature and highlights major advances that have occurred in the past few decades. This chapter provides context for the interpretation of volcaniclastic deposits, highlights limitations of those interpretations, and illustrates how the nature of volcaniclastic processes and their initiation mechanisms and transport behaviour can bias their preservation in the geological record.

My ability to contribute this review is based on a career of interactions with many colleagues worldwide. I am particularly indebted to field excursions and discussions with Tom Pierson, Jim Vallance, Richard Iverson, Kevin Scott, Willie Scott, Cynthia Gardner, Richard Waitt, Lee Siebert, Barry Voight, Fred Swanson, Patrick Pringle, and my late colleagues Dick Janda, Hiroshi Suwa, and Harry Glicken. I am also indebted to the many colleagues and associates who graciously granted me permission to use their photographs. Thoughtful critiques by Chris Harpel, Vern Manville, Larry Mastin, Y.K. Sohn, and Sebastian Watt improved the clarity and sharpened the presentation of this contribution.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

JM: conceptualization (lead), writing – original draft (lead), writing – review & editing (lead).

Funding to produce this review was provided by the United States Geological Survey, Volcano Science Center, Cascades Volcano Observatory.

All data generated or analysed during this study are included in this published article (and its supplementary information files).