The influence of conduit processes on changes in style of basaltic Plinian eruptions: Tarawera 1886 and Etna 122 BC

Abstract

Basaltic volcanism is most typically thought to produce effusion of lava, with the most explosive manifestations ranging from mild Strombolian activity to more energetic fire fountain eruptions. However, some basaltic eruptions are now recognized as extremely violent, i.e., generating widespread phreatomagmatic, subplinian and Plinian fall deposits. We focus here on the influence of conduit processes, especially partial open-system degassing, in triggering abrupt changes in style and intensity that occurred during two examples of basaltic Plinian volcanism. We use the 1886 eruption of Tarawera, New Zealand, the youngest known basaltic Plinian eruption and the only one for which there are detailed written eyewitness accounts, and the well-documented 122 BC eruption of Mount Etna, Italy, and present new grain size and vesicularity data from the proximal deposits. These data show that even during extremely powerful basaltic eruptions, conduit processes play a critical role in modifying the form of the eruptions. Even with very high discharge, and presumably ascent, rates, partial open-system behaviour of basaltic melts becomes a critical factor that leads to development of domains of largely stagnant and outgassed melt that restricts the effective radius of the conduit. The exact path taken in the waning stages of the eruptions varied, in response to factors which included conduit geometry, efficiency and extent of outgassing and availability of ground water, but a relatively abrupt cessation to sustained high-intensity discharge was an inevitable consequence of the degassing processes.

Introduction

Basaltic volcanism is a common and spectacular natural phenomenon. Many studies portray basaltic volcanism as overwhelmingly effusive, and indeed, most basaltic eruptions are weakly explosive at best. Basaltic explosive eruptions are commonly Strombolian or Hawaiian in style, ranging from mild explosions to more energetic fire-fountains impacting areas up to a few km across. However, some moderate to large explosive eruptions at several basaltic volcanoes were extremely violent in character, including large phreatomagmatic, subplinian and Plinian types (Fig. 1). Examples include the scoria fall deposits of the prehistorical Fontana Lapilli and San Judas Formation from Masaya Caldera (Nicaragua) Williams, 1983, Bice, 1985, the 122 BC eruption of Etna (Coltelli et al., 1998), the 1790 AD Keanakakoi eruption of Kilauea (Hawaii) McPhie et al., 1990, Dzurisin et al., 1995, Mastin, 1997 and the 1886 Tarawera eruption (New Zealand) (Walker et al., 1984).

Basaltic Plinian volcanism deserves detailed study because it is the most poorly understood type of basaltic activity, and also the most dangerous. Highly explosive basaltic eruptions are scarce in the geologic record. The study of such eruptions is still in its infancy; the main known examples have been recognized only within the last 20 years (e.g., Williams, 1983, Walker et al., 1984, McPhie et al., 1990, Dzurisin et al., 1995, Mastin, 1997, Coltelli et al., 1998, Coltelli et al., 2000). Highly explosive basaltic eruptions are all the more hazardous because the rapid ascent rates of basaltic magma means that the warning time between onset of unrest and eruption may be as short as a few hours. In addition, because such eruptions are atypical of most volcanism at basaltic centers, their precursors may be ignored or misunderstood until too late. The increased awareness of the potentially highly explosive character of basaltic magmas also raises important issues for hazard assessment of basaltic volcanoes. For example, volcano tourism is bringing increasing numbers of visitors onto many basaltic volcanoes, including those featured in this study.

Large explosive eruptions are highly destructive events. The intensity, duration and style of such eruptions are determined within (i) the region of magma storage, which provides the overpressure driving an eruption, and (ii) the volcanic conduit, where magma decompresses, loses volatiles and sometimes partially crystallizes or interacts with wall rock and groundwater. Eruptive style or intensity may be driven by (i) flow behavior in the conduit (e.g., Jaupart, 1998, Papale et al., 1998, Denlinger and Hoblitt, 1999) which depends on the magma rheology as determined by the concentrations of dissolved volatiles, bubbles and crystals (e.g., Pinkerton and Stevenson, 1992, Hess and Dingwell, 1996, Manga et al., 1998, Stevenson et al., 1998), or (ii) external environmental factors, such as width of the conduit or influx of external water (Barberi et al. 1989).

Fragmentation of bubbly silicic melts due to syn-eruptive, closed-system degassing during rapid magma ascent is common and is fairly well constrained by studies of size distributions in vesicle populations Cashman and Mangan, 1994, Mangan et al., 1993. However, varying degrees of open-system degassing may occur if time and bubble interconnectedness permit the separation of gas from the melt (Westrich et al. 1988). Highly explosive activity is favored by high viscosity of the magma, which retards bubble growth and allows bubble pressure to build (McBirney, 1973). This is because fragmentation occurs when the pressure within the gas bubbles exceeds the tensile strength of the melt (McBirney and Murase, 1970), and explains why eruptions of rhyolitic/andesitic magma are likely to be more explosive than basaltic eruptions. As an example, bubble growth rates of 3.2×10⁻⁴, 9×10⁻⁴ and 1.2–5.7×10⁻⁷ cm s⁻¹ were determined for lava effusion from Kilauea, Hawaii, for fire-fountaining episodes at Kilauea, and for the May 1980 eruption of Mount St. Helens, respectively Mangan et al., 1993, Mangan and Cashman, 1996, Klug and Cashman, 1994. In fluid basaltic melts, some special circumstances are required to limit open-system behaviour and promote Plinian eruption. The growth of anhydrous microlite phases as a result of rising liquidus temperatures accompanying decreasing volatile content of the melt Geschwind and Rutherford, 1995, Wolf and Eichelberger, 1997, and enhanced interaction of the rising magma with groundwater and wall rock are processes that could feed back to change the rheology of the magma.

Fragmentation in magmatic systems once was assumed to occur as soon as a melt reaches a vesicularity threshold (commonly taken as 75%). However, a simple fragmentation threshold of 75% vesicularity does not explain the range in density commonly observed in the products of “dry” (magmatic) explosions Houghton and Wilson, 1989, Thomas et al., 1994, Gardner et al., 1996, Klug and Cashman, 1996, Cashman et al., 1998. Numerical and experimental simulations suggest, as an alternative, that magma fragments in brittle fashion when the time scale of deformation is shorter than the viscous relaxation time Dingwell and Webb, 1990, Alidibirov and Dingwell, 1996, Dingwell, 1998, Mader, 1998. If volatile loss is the only process occurring in the melt, the viscosity (and thus the relaxation time) increases only gradually. However, viscosity increases rapidly if (i) degassing causes abundant crystallization of microlites, or (ii) significant magma–water interaction has occurred before fragmentation.

A few models have been recently proposed to explain the high explosivity of some basaltic magmas. For example, volcanologists agree that the violence of the 1790 eruption at Kilauea volcano was caused mainly by magma–water interaction, with some debate on the circumstances that allowed this water to mix with magma (e.g., McPhie et al., 1990, Mastin, 1997). At Masaya (Nicaragua), Williams (1983) suggests that the high temperature and density of basaltic magma resulted in a higher eruption column than would have resulted from a comparably sized eruption of silicic magma. Walker et al. (1984) use interaction with a geothermal system to explain the intensity of the 1886 Tarawera eruption, however, there is a total absence of hydrothermally altered wall rock clasts in the Plinian products of the eruption. The Coltelli et al. (1998) model for the 122 BC Etna eruption suggests that a sudden decompression event allowed the magma to develop rapidly into a microvesicular foam. One possible cause of decompression is unloading by a landslide, but there is no evidence for contemporaneous slope failure. The model proposes that the injection of >1 km³ of very buoyant magma into the volcanic edifice caused non-elastic deformation on the eastern flank. The resulting sudden decompression enabled a large part of the magma to become foam and accelerate upward, producing the Plinian eruption.

Less well explained for both Tarawera and Etna are the rapid transitions that occurred into phreatomagmatic explosivity of markedly reduced intensity, which we will explore in this paper.

The eruptions we have studied are large explosive eruptions that were sustained and intense events yet both show ill-explained and abrupt changes in eruptive style with dramatic reductions in intensity. We have very elegant models for the ‘steady-state’ behavior of sustained eruption plumes Sparks, 1986, Sparks et al., 1991, Sparks et al., 1992, Sparks et al., 1997, Bursik et al., 1992, Koyaguchi, 1994, Ernst, 1996, Woods, 1998 yet know very little about what causes unidirectional or pulsatory changes in their eruptive style and intensity. This lack of knowledge is a first-order question for physical volcanologists as it reduces our ability to anticipate the patterns of behavior during future damaging large eruptions.