Classification of Pyroclastic Deposits
CLASSIFICATION OF PYROCLASTIC DEPOSITS OF THE LESSER ANTILLES
Although a general agreement exists on the classification of modern pyroclastic rocks, regional differences often add significant complications, so that specific classification schemes may not be readily adaptable to all areas. For example, in the Lesser Antilles the explosive and gravitational collapse of domes produce pyroclastic flows characterized by the presence of variable amounts of blocks, up to 6 m in size, enclosed in a matrix of fine-grained, pulverized ash-sized material. Such deposits are termed block and ash flows. In contrast, in the Cenozoic rift lava fields of Saudi Arabia (Camp and Roobol, 1991; Roobol and Camp, 1991a and b) morphologically similar trachyte lava domes are surrounded by aprons of pyroclastic flows and surges. These deposits however are almost entirely lacking in ash (Merle and Smith, 1998), thus making the term block and ash flows inappropriate. Factors that may explain these differences could be entirely environmental, water-logged stratovolcanoes in the Lesser Antilles versus the paucity of groundwater in Saudi Arabia, or inherent in the composition of the magma (andesite/dacite versus trachyte/comendite).
The formation and role of ash in the classification of pyroclastic deposits is an important subject in view of the differences between deposits such as those discussed above. Fine-grained material in pyroclastic rocks is composed of juvenile clasts, lithic clasts (derived from the erosion of the volcanic vent), and crystals. Original vitric-crystal ratios are preserved within large clasts in pyroclastic flow deposits, so that large pumice blocks can be crushed and original vitric-crystal ratios measured. Similar determinations on the ashy matrix of the same deposit will then reveal any depletion in fine-grained vitric ash. Such determinations have shown that Lesser Antillean pyroclastic flows can show vitric losses of up to 44 weight percent (Wright et al., 1984; Roobol et al., 1987; Smith and Roobol, 1990). These losses have been attributed to two mechanisms. The first of these is caused by the occurrence of low-lying oceanic volcanoes having high-level water tables as a consequence of heavy rainfall. These water tables cause hydrovolcanic activity with the production of phreatic/phreatomagmatic eruptions. The second is due to ingestion of lush dense tropical vegetation which fills the valleys draining the West Indian volcanoes, down which many of the pyroclastic flows travel (Wright et al., 1984; Smith and Roobol, 1990). The resulting water vapor causes expansion of the pyroclastic flows with increasing elutriation of the lightest vesiculated juvenile component which is lost in the rising ash cloud and dispersed to form thin laminated ash falls.
Fisher (1991) has emphasized that to call a deposit “fines depleted” requires proof that the original eruptive mass contained fines that then became depleted during transportation and deposition. There are two lines of evidence to show that the pyroclastic deposits of Lesser Antillean volcanoes are commonly fines depleted. The first is, as outlined above, that the vitric-crystal ratios of juvenile clasts in the pyroclastic deposits are higher compared to their fine-grained ashy matrices. A second indication that vitric-fines depletion occurs during transportation is demonstrated by the fact that ash hurricane deposits at locations progressively more distal to Mt. Pelée and nearer to the capital of Fort-de-France show a systematic increasing crystal enrichment (Smith et al., 1999).
For the Lesser Antilles a pyroclastic classification developed for the deposits on one island can however be applied to other islands in the same arc, but may be inadequate to address the details of specific deposits from different geological settings. The classification used here was specifically developed to understand the pyroclastic deposits of the Lesser Antilles and should be used in other geological settings with caution. A classification of pyroclastic deposits for the Lesser Antilles developed from that proposed by Smith and Roobol (1990) is presented in the accompanying Table. The classification emphasizes the relationship between eruptive style, the wide variety of deposit types that can result from each style, and the lithological variations found on Lesser Antillean volcanoes.
Pelean-style eruptions are well documented for the type example of the 1902-05 eruption of Mt. Pelée (LaCroix 1904; Roobol and Smith, 1975a; Fisher et al., 1980; Smith and Roobol, 1990), which was repeated in the 1929-32 eruption (Perret 1937; Roobol and Smith, 1975a; Smith and Roobol, 1990). The juvenile component is andesite, and the activity commences with phreatic/phreatomagmatic explosions followed by dome growth which is continually interrupted by both explosive pyroclastic emissions through and around the dome, and gravitational collapse of the dome. The pyroclastic flows of block and ash type typically emerge to expand rapidly into a flow-head rich in dense blocks ripped from the dome, which by air ingestion rapidly expands to form an overriding ash-rich cloud. Following the flow-head is the body of the flow containing smaller blocks in a fine vesicular ash/gas suspension. The final part is the tail of the flow which is ash rich. The most striking deposit produced by Pelean-style eruptions is a coarse, unsorted block and ash flow deposit which occurs as valley-fills (high-aspect ratio or HAR deposits). The body and tail of the flow composed of progressively smaller clasts and ash may break away from the main part of the flow by surmounting intervening topographic barriers and cover wider sectors of the volcano’s flanks to form ash-rich, fine-grained block and ash flow deposit (low-aspect-ratio or LAR type). The overriding ash cloud can cover even wider sectors of the volcano to form dense andesite surge deposits (ash cloud surges) (Sparks and Walker, 1973; Fisher, 1979) with variable thickness and internal dune and anti-dune structures. Such deposits are characteristically rich in carbonized plant remains. The finest ash, which rises to the highest elevations, is deposited as thin laminated ash and dust fall layers, composed dominantly of pulverized rock. These are deposited over wide sectors of the volcano’s flanks.
St. Vincent-style eruptions
The type St. Vincent-style eruption was that of Soufriere, St. Vincent in 1902, when explosion columns of vesiculated basaltic andesite rose from an open crater. Partial collapse of the columns produced pyroclastic flows which followed the valleys to the coast to produce distinctive scoria and ash flow deposits. These are characterized by blue-black basalt/basaltic andesite clasts up to 50 cm diameter showing cauliflower-textured surfaces. The cores of the clasts and the smaller angular clasts (formed from the break up of larger clasts) are vesicular with large rounded vesicles. They are contained in a matrix of black-purple ash. The pyroclastic flow deposits are associated with scoriaceous surge deposits showing variable thickness and internal dune bedding. Also associated with these deposits are well-sorted stratified scoriaceous lapilli and ash fall beds with internal layering composed of lapilli and ash of vesiculated scoria. These arise directly from pyroclastic eruptions from an open crater. The pyroclastic flow and fall deposits from the type eruption were first described by Hay (1959). A description of the eruption was given by Anderson and Flett (1903) and was discussed by Roobol and Smith (1975a).
Plinian-style eruptions have not occurred during the period of European settlement of the Lesser Antilles. They were experienced by the pre-Columbian Indians however, as locations are known where village sites on the flanks of Mt. Pelée, Martinique are overlain by deposits from this style of activity (Roobol et al., 1976) This type of eruption comprises a series of sustained explosions from an open crater that produces an eruption column up to 50 km high (Walker, 1981). Column collapse generates pumiceous pyroclastic flows and surges (Fisher, 1979). The former, as with other flows, can be divided into a flow-head, flow-body and flow-tail (Wilson and Walker, 1982).
Pumice and ash flow deposits (ignimbrites) occur as valley fills (HARI type). They are composed of white, highly-vesicular blocks (to 1m) and lapilli set in a matrix of white pumiceous ash. Individual deposits may be up to 50 m thick, contain abundant carbonized wood, gas segregation pipes and are rarely welded. Ash hurricane deposits are ash-rich pumiceous flow deposits of variable thickness up to 3 m. Individual beds thin over topographic highs and lack dune bedding structures characteristic of surge deposits. Ash hurricane deposits are widely dispersed and may be termed low aspect ratio ignimbrites or LARIs (Wright et al., 2016). Pumiceous surge deposits are composed of pumiceous lapilli, crystals and ash with variable thickness and internal dune bedding. They commonly contain carbonized wood. On Mt. Pelée these surge deposits are closely associated with the central vent and are not found at any distance from it, suggesting that they originated from the partial collapse of the primary eruption column (Smith and Roobol, 1990). Plinian-style fall deposits are pumiceous lapilli and ash fall layers which are well sorted, internally layered, and mantle topography on all flanks of the volcano. Individual layers are thought to originate from successive explosions.
This type of activity has also not yet been witnessed in the Lesser Antilles and has only been identified by comparison with deposits described from the 1783 eruption of Mt. Asama in Japan (Aramaki, 1956, 1957). During this eruption, the activity was described as a pyroclastic flow that “boiled over” the rim of the crater without forming an explosive eruption column. Deposits of this type are not common in the Lesser Antilles, but have been described from Saba (Roobol and Smith, 1980b), the Quill, St. Eustatius (Roobol and Smith, 1980b; Roobol et al., 1997), the Soufriere Hills, Montserrat (Smith et al., 1985; Roobol and Smith, 1998), Dominica (Roobol and Smith, unpub. data), and Mt. Pelée, Martinique (Smith and Roobol, 1990). They are of interest as they may mark transitional intervals between long periods of Pelean activity and long periods of Plinian activity as has been indicated by the stratigraphy of Mt. Pelée.
The deposits are typically unsorted pyroclastic flow type containing subangular clasts up to 30 cm in a fine-grained ash-rich matrix, that are termed semi-vesicular andesite block and ash flow deposits (Roobol and Smith, 1980a and b; Wright et al., 1980; Smith and Roobol, 1990), and were first described in the Lesser Antilles from the islands of Martinique and Saba (Roobol and Smith, 1980b). The deposits appear intermediate in character between block and ash flow and pumice and ash flow deposits. The larger clasts show outer vitric selvages devoid of vesicles and the cores contain 30-50 % small round vesicles up to 1.5 mm in diameter. In one example on Mt. Pelée, a semi-vesicular andesite block and ash flow deposit, has a thickness of 11 m and is welded with columnar joints. Asama-style eruptions can also produce semi-vesicular andesite surge and lapilli and ash deposits, although these are much less common than the flow deposits.
The eruption of the Soufriere Hills volcano on Montserrat, 1995 to present, has drawn attention to a type of block and ash flow deposit which is intermediate in character between block and ash flow deposits produced by Pelean-style activity (dominated by juvenile blocks of dense andesite) and the semi-vesicular andesite block and ash flow deposits produced by Asama-style activity (dominated by juvenile clasts of semi-vesicular andesite). These are block and ash flow deposits with abundant juvenile blocks of both semi-vesicular andesite and dense andesite types. In these, the semi-vesicular clasts are concentrated in the upper one-third of each deposit. Such deposits occur also in stratigraphic sections of the same volcano where they have been dated between 22,000 to 20,000 years B.P. (Roobol and Smith, 1998; Smith et al., 2007).
Since European settlement of the Lesser Antilles the most common style of eruptive activity has been phreatic or phreatomagmatic eruptions. Historic examples include fine ashes composed entirely of hydrothermally altered dome rocks produced during the 1976-77 eruption of La Soufriere, Guadeloupe (Le Guern et al., 1980), and thin ashes with accretionary lapilli produced during the 1979 eruption of Soufriere, St. Vincent when rising basaltic andesite magma interacted with water in a crater lake (Brazier et al., 1982; Shepherd and Sigurdsson, 1982). Fine-grained ashes with accretionary lapilli and entombed gas cavities were also characteristic of the initial stages of both the 1902 eruption of Mt. Pelée, Martinique (Smith and Roobol, 1990) and the current eruption of the Soufriere Hills, Montserrat. In both of these cases rising magma is thought to have interacted with abundant groundwater, which in the case of Mt. Pelée appears to have been expelled onto the surface (Roobol and Smith 1975a; Smith and Roobol, 1990). Prehistoric deposits of a similar nature have been reported from Martinique (Roobol and Smith 1976b; Smith and Roobol, 1990), Montserrat (Smith et al., 1985), and St. Kitts (Roobol et al., 1985).
In the case of phreatic activity the resulting lithic lapilli and ash fall deposit comprises stratified ash composed of rusty altered lithic clasts lacking any juvenile component (Smith and Roobol, 1990). Where fine-grained, the beds may be internally laminated, weakly lithified, contain accretionary lapilli and entrapped air bubbles (Roobol et al., 1985). In the case of phreatomagmatic eruptions, lapilli and ash fall deposits are very similar in appearance to the lithic fall deposits but contain, in addition, a juvenile component usually of vitric-crystal material. Lithic surge deposits (phreatic / phreatomagmatic) have been described by Baker (1985) from St. Kitts, by Shepherd and Sigurdsson (1982) from the 1979 eruption of Soufriere, St. Vincent, and from the older deposits from the Soufriere Hills volcano on Montserrat (Smith et al., 1985). Clast-supported lithic block flow deposits resulted from the 1976 phreatic eruption of La Soufriere, Guadeloupe and were described by Sheridan (1980). A general description of the structures of phreatomagmatic ashes resulting from water-magma interaction can be found in Lorenz (1974).
The above examples all involve the interaction of groundwater or surface water with rising magma, however other forms of water-magma interaction are also possible. For example the interaction of magma and sea water. An excellent example of such interaction is the hornblende-bearing alkali basalt volcano of Kick’em Jenny in the Grenadines. This volcano, which is at a minimum depth of 182 m below sea level, has undergone ten episodes of activity since 1939, several of which have been accompanied by boiling of the sea. Similar reports of boiling of the sea have been reported offshore of Carbet on the west coast of Martinique (Smith and Roobol, 1990).
Volcaniclastic deposits lacking primary pyroclastic structures
The lower subaerial flanks of most of the Lesser Antillean volcanoes contain a significant proportion (up to 50% of the stratigraphic column) of deposits that have been fluviatile reworked and have lost their primary pyroclastic characteristics. For example, in the older stratigraphy of Mt. Pelée, revealed on the Atlantic coast of Martinique, the deposits are dominated by thick successions of yellow pumiceous fluviatile strata with alternations of thick successions of gray coarse dense andesite boulder deposits. The younger stratigraphy of Mt. Pelée, well exposed on the Caribbean coast, comprises thick successions of pumiceous deposits resulting from Plinian activity which alternate with thick successions of dense andesite block and ash flow deposits produced by Pelean eruptions. It is therefore reasonable to conclude that the pattern of alternating Plinian and Pelean activity dominated the entire history of the volcano. Smith and Roobol (1990) have classified these fluviatile reworked deposits to be epiclastic even though their components are entirely pyroclastic. This has been questioned by Fisher (1991) who prefers that the term epiclastic be reserved for particles whose primary formation is due to the erosion of any type of pre-existing rock. In the terminology of Fisher (1991), the fluviatile reworked deposits should be termed fluviatile reworked pyroclastic rocks, such as fluviatile reworked ash, or fluviatile reworked block deposits rather than fluviatile sandstone and conglomerate as used by Smith and Roobol (1990).
Another problem of terminology arises with the definition of mudflows and lahars. Lahars are essentially mudflows generated during the course of an eruption. Their composition may range from entirely reworked pyroclastic material (e.g. when rain falls on unconsolidated ash) to a multilithologic volcaniclastic deposit caused by a volcanic-related earthquake triggering the collapse of a water-saturated slope. Mudflows form in a similar manner but are not directly associated with a volcanic eruption. Both types of deposit contain non-carbonized wood or hollows where the wood once existed but has since rotted away. They also lack gas escape structures. This serves to separate them from pyroclastic flow deposits, but they cannot be confidently distinguished from one another in the stratigraphic record. Lahars can in fact only be confidently identified by direct observation during the course of an eruption. Smith and Roobol (1990) included mudflow/lahar deposits under epiclastic rocks which was also questioned by Fisher (1991). However as we have not yet confidently identified a lahar in the stratigraphic record, we prefer using the combined term mudflow/lahar, and draw attention here to the fact that in general they probably should not be grouped with epiclastic rocks sensu stricto, although a detailed study of the lithic components of this type of deposit might establish a range from wholly pyroclastic to wholly epiclastic.