On the emplacement of ignimbrite in shallow-marine environments

https://doi.org/10.1016/S0377-0273(99)00116-XGet rights and content

Abstract

The emplacement of primary pyroclastic flow deposits on the sea floor can occur (1) by nonturbulent, dense flows that conserve their heat under water, or (2) because the flow (either turbulent or nonturbulent) temporarily pushes back the shoreline. The feasibility of the second mechanism has been investigated theoretically and experimentally using a simple analogue system. The experiments involved sustained (turbulent) saline currents of different densities flowing into a sectorial tank of ambient fluid that was either lighter or denser than the current. The currents in all cases displaced the entire layer of ambient fluid up to a distance (R), irrespective of whether they were denser or lighter. R is analogous to the shoreline displacement distance up to which hot ignimbrite can be emplaced by a pyroclastic flow entering the sea. By varying flow rate, density contrast, and depth of the ambient fluid layer, a simple equation for R is found that is in good agreement with the experimental data. The effect of seawater boiling on shoreline displacement was also investigated by experiments using currents of carbonate solution that produced gas when flowing into a layer of ambient acid. The distance R was not reduced in these experiments. The equations show that pyroclastic flows of about 10 km3 or more are capable of pushing back the sea at least a couple of kilometres and possibly more in areas of extensive shallow water, enabling hot ignimbrite to be laid down on the seabed at depths corresponding to a few tens of metres. Shoreline displacement is thus a feasible mechanism for the emplacement of welded ignimbrite in shallow-marine settings. Emplacement of subaerially erupted ignimbrite below storm wave base is not ruled out by the calculations, but requires a combination of high discharge rate, sustained eruption, low density contrast between flow and water, steep bathymetric gradient, and small island volcano. Application of the model to the 1883 Krakatau eruption suggests that seawater displacement may explain the occurrence of primary ignimbrite on the seabed 10 km northwest of the volcano.

Introduction

Pyroclastic flows are a common product of volcanoes in marine settings (Cas and Wright, 1987). However, exactly what happens when a pyroclastic flow enters the sea is poorly understood. Small-volume block-and-ash flows formed by dome collapse are known to travel as highly concentrated avalanches that are probably denser than water. Lacroix (1904) observed examples of such flows entering the sea at Montagne Pelée in 1902, and their submarine equivalents broke cables nearly 20 km offshore. At Montserrat in 1996, the basal avalanches of some dome-collapse flows entered the sea, generating a zone of intense boiling, while the overriding ash cloud passed over the surface (Cole et al., 1998).

The case of the voluminous pyroclastic flows that generate ignimbrite is more controversial Fisher, 1984, Cas and Wright, 1987, Cas and Wright, 1991. The occurrence of welded ignimbrites in marine sedimentary successions has led to the assertion that pyroclastic flows can enter the sea and travel at high temperatures for considerable distances under water (e.g., Francis and Howells, 1973, Sparks et al., 1980). However, the evidence has been reviewed critically by Cas and Wright (1991) using strict criteria for water depth and high-temperature emplacement (welding, columnar jointing, elutriation pipes, thermal remnant magnetism). They concluded that most ignimbrites in ancient marine successions were emplaced in shoreline or shallow-water settings. The best examples include the Capel Curig Tuffs Howells et al., 1979, Fritz and Howells, 1991, Orton, 1991, McArthur et al., 1998, the Pitts Head Tuff (Wright and Coward, 1977), the Upper Donzurobo Formation Kato et al., 1971, Yamazaki et al., 1973, the Koura Formation (Kano, 1990), and the Shirahama group (Tamura et al., 1991).

In these cases, the sedimentary facies associations suggest water depths no more than a few tens of metres when the ignimbrite was laid down. A modern example is the 1883 Krakatau ignimbrite, which is known from measurements of thermal remnant magnetism to have been emplaced at about 500°C in 40 m of water 10 km from the caldera centre Mandeville et al., 1994, Mandeville et al., 1996. It is therefore well established that ignimbrite can in some circumstances deposit and even weld in water depths up to a few tens of metres, although the mechanism is not well understood.

Most pyroclastic flow deposits in deep-marine environments are inferred to have been emplaced as water-saturated mass flows, possibly in some cases the cold, lateral equivalents of subaerial ignimbrites Yamada, 1973, Niem, 1977, Carey and Sigurdsson, 1980, Wright and Mutti, 1981, Cas, 1983, Kano et al., 1989. There are fewer occurrences of ignimbrites emplaced hot below wave base Schneider et al., 1992, Kano et al., 1994, Fritz and Stillman, 1996, White and McPhie, 1997. Some of these may have been laid down close to source during subaqueous eruptions, as demonstrated for the Lower Rhyolitic Tuff (Howells et al., 1986) and at Mineral King Caldera (Kokelaar and Busby, 1992).

Other developments have further complicated this debate. Large pyroclastic flows have been widely assumed to travel as highly concentrated flows, like their smaller-scale equivalents (e.g., Sparks, 1976, Wilson, 1985). However, there is good evidence that some flows such as those of the Koya Ignimbrite (Ui, 1973), the Campanian Ignimbrite (Fisher et al., 1993), and the Kos Plateau Tuff Keller, 1970, Allen, 1998 travelled many kilometres or tens of kilometres over the sea, showing that parts of the flows were significantly less dense than water. Parts of the Krakatau pyroclastic flows also travelled across water, laying down hot deposits on nearby islands (Carey et al., 1996). If many large pyroclastic flows are light enough to travel across water, a mechanism needs to be found to explain how hot ignimbrite comes to be deposited on the seabed. In this paper, we use the term pyroclastic flow in a general sense for any pyroclastic density current that lays down ignimbrite, no matter what the particle concentration.

Fig. 1 shows four possible situations when a pyroclastic flow enters the sea:

(1) The flow is denser than water and travels beneath it (ρc>ρwρa). The bulk densities of block-and-ash or scoria flow deposits typically lie in the range 1500 to 2000 kgm−3, so these flows, if poorly expanded during transport, are most likely to behave in this manner. Pyroclastic flows that travel in a non-turbulent mode and that are protected by a steam carapace Francis and Howells, 1973, Sparks et al., 1980, Kokelaar and Busby, 1992 are most likely to travel furthest before transforming into debris flows and turbidity currents by water ingestion.

(2) The flow is slightly lighter than water and forms a surface current (ρw>ρcρa). Many ignimbrite deposits have bulk densities only slightly greater than 1000 kgm−3 (e.g., Mandeville et al., 1996); thus, the parent flows are likely in many cases to be less dense than water, as supported by the known ability of some such flows to cross open sea.

(3) The flow is much lighter than water and mainly passes over it (ρwρc>ρa). This corresponds to flows that are highly expanded or to the upper parts of density-stratified flows.

(4) Irrespective of flow density, ingestion and vaporisation of water is so rapid that destruction at the flow front exceeds supply rate and the flow is destroyed explosively at the shoreline Carey and Sigurdsson, 1980, Whitham, 1989.

In cases 1 to 3, there is a region in which the flow totally displaces the sea, irrespective of whether the flow is lighter or denser than water. Provided that the eruption is sustained long enough, it is possible for the flow to lay down hot ignimbrite on this region of the seabed. Once the eruption ceases, the sea will then return to cover the ignimbrite. There are therefore two mechanisms by which hot ignimbrite can be emplaced below sea level: (1) by dense laminar flows that travel under the sea without significant mixing, and (2) by flows (laminar or turbulent) that simply push back the shoreline and deposit ignimbrite on the seabed. We do not address the first mechanism any further here, but focus on the ability of pyroclastic flows to displace the sea. We refer to this mechanism as shoreline displacement.

In this paper we constrain the extent of shoreline displacement theoretically and experimentally with simple analogue systems. The ability to push back the sea is a function of discharge rate, flow density, and sea floor bathymetry. We assume that the pyroclastic flow spreads radially and is sustained over a long time period. Our model is only applicable to sustained eruptions in which there is enough time for the hot pyroclastic flows to flush the seawater and cold, water-saturated flow front out of the displacement zone. We also assume that the flow behaves as a homogeneous, turbulent fluid. This is perhaps most appropriate for fast-moving flows, although even small pyroclastic flows at Mount St. Helens have been successfully modelled as turbulent flows (Levine and Kieffer, 1991). Experiments were carried out both with simple saline currents and with saline currents that produced gas by mixing with the ambient fluid. The latter were designed to place constraints on the effects of water ingestion and steam production on the extent of shoreline displacement.

Section snippets

Pyroclastic flows denser than seawater

We consider first the case of a pyroclastic flow denser than seawater (case 1; Fig. 1). A gravity current on a horizontal surface is driven by the difference in hydrostatic pressure between the current and the ambient fluid in the far field. The Bernouilli equation states that, for the inviscid turbulent case, the difference in hydrostatic pressure must equal the dynamic pressure Benjamin, 1968, Simpson, 1987:c−ρw)gh=12ρwU2so that the velocity of the current is:U=2c−ρw)ρwghwhere ρc and ρw

Saline currents without gas production

Two sets of experiments were carried out (Fig. 3). The first involved light currents of fresh water in denser salt water. We used a sectorial Plexiglass tank with a 2.8-m-long triangular base, an apex angle of 5.9°, and a height of 30 cm (Fig. 3). A reservoir with a removable gate was added near the apex of the tank. The tank was filled with salt water up to a height between 1.2 and 9 cm. The reservoir was then fed by a constant flux of coloured fresh water. Flow rates varied between 61 and 755

Discussion

We have presented equations to describe the entry of a pyroclastic flow into the sea and have tested those equations using a simple laboratory system. The equations and experiments suggest that, given steady-state conditions, the pressure exerted by a pyroclastic flow might be capable of temporarily displacing the sea, enabling hot ignimbrite to be deposited on the seabed.

The experiments with gas production show that buoyancy generation in the flow does not necessarily affect the displacement

Acknowledgements

Thoughtful reviews by Steven Carey and Brian Dade led to improvements to the manuscript. Financial support from the French Centre de Recherche Volcanologique (CRV) is acknowledged.

References (53)

  • A.G. Whitham

    The behaviour of subaerially produced pyroclastic flows in a subaqueous environment: evidence from the Roseau eruption. Dominica, West Indies

    Mar. Geol.

    (1989)
  • Allen, S.R., 1998. Volcanology of the Kos Plateau Tuff, Greece: The Product of an Explosive Eruption in an Archipelago....
  • T.B. Benjamin

    Gravity currents and related phenomena

    J. Fluid Mech.

    (1968)
  • S.N. Carey et al.

    Pyroclastic flows and surges over water: an example from 1883 Krakatau eruption

    Bull. Volcanol.

    (1996)
  • R.A.F. Cas

    Submarine ‘crystal tuffs’: their origin using a Lower Devonian example from southeastern Australia

    Geol. Mag.

    (1983)
  • R.A.F. Cas et al.

    Volcanic Successions. Modern and Ancient

    (1987)
  • R.A.F. Cas et al.

    Subaqueous pyroclastic flows and ignimbrites: an assessment

    Bull. Volcanol.

    (1991)
  • P. Cole et al.

    Pyroclastic flows generated by gravitational instability of the 1996–97 lava dome of Soufriere Hills Volcano, Montserrat

    Geophys. Res. Lett.

    (1998)
  • F. De Rooij et al.

    Saline and particle-driven interfacial intrusions

    J. Fluid Mech.

    (1999)
  • T.H. Druitt

    Pyroclastic density currents

  • R.V. Fisher

    Submarine volcaniclastic rocks

  • E.H. Francis et al.

    Transgressive welded ash-flow tuffs among the Ordovician sediments of NE Snowdownia, N. Wales

    J. Geol. Soc. London

    (1973)
  • M.F. Howells et al.

    The Capel Curig Volcanic Formation, Snowdonia, North Wales: variation in ash-flow tuffs related to emplacement environment

  • M.F. Howells et al.

    The submarine eruption and emplacement of the Lower Rhyolitic Tuff Formation (Ordovician), North Wales

    J. Geol. Soc. London

    (1986)
  • K. Kano et al.

    Deformation structures in shale bed indicate flow direction of overlying Miocene subaqueous pyroclastic flow

    Bull. Volcanol.

    (1989)
  • I. Kato et al.

    Subaqueous pyroclastic flow deposits in the Upper Donzurobo Formation, Nijosan district, Osaka, Japan

    J. Geol. Soc. Jpn.

    (1971)
  • Cited by (0)

    View full text