The interplay among clast size, vesicularity, postfragmentation expansion, and clast breakage: An example from the 1.8 ka Taupo eruption
Published:February 07, 2019
- PDF LinkChapter PDF
S.J. Mitchell, S. Biass, B.F. Houghton*, A. Anderson, E. Bonny, B.H. Walker, B.G. Mintz, N.R. Turner, D. Frank, R.J. Carey, M.D. Rosenberg, 2019. "The interplay among clast size, vesicularity, postfragmentation expansion, and clast breakage: An example from the 1.8 ka Taupo eruption", Field Volcanology: A Tribute to the Distinguished Career of Don Swanson, Michael P. Poland, Michael O. Garcia, Victor E. Camp, Anita Grunder
Download citation file:
*Corresponding author: firstname.lastname@example.org.
Field studies of tephra-fall deposits traditionally use the density of juvenile pyroclasts to determine vesicularity of the host magma at the point of fragmentation. A range of pyroclast sizes between 16 and 32 mm has commonly been chosen for this purpose. Larger pyroclasts outside this range may undergo postfragmentation vesiculation due to slow cooling of the interior of the clasts, while smaller pyroclasts may be too small to represent accurately the distribution of the largest vesicles. The assumption of this method, of course, is that the 16–32 mm size range is representative of the fragmented magma. We explore, in detail, variations in density over a size range of 4–128 mm from Unit 2 pyroclasts of the 1.8 ka Taupo eruption and make inferences about the roles of postfragmentation vesiculation and secondary breakage of pyroclasts. We find (1) there is a clear threshold for onset of postfragmentation vesiculation at >32 mm, and (2) there are broken small pieces of the largest pyroclasts in the sample that artificially skew the density distribution for smaller size fractions. We constrain uncertainty associated with vesicularity measurements and offer best-practice recommendations in the hope of improving consistency of field sampling and laboratory processing of pyroclast populations for vesicularity studies.
Powerful sustained explosive volcanic eruptions erupt large volumes of tephra, ash, lapilli, and blocks/bombs into the atmosphere, with the potential to cause disruption to aviation (Guffanti and Tupper, 2015), damage to buildings, agriculture infrastructure, and the environment (Jenkins et al., 2014; Ayris and Delmelle, 2012; Wilson et al., 2014, 2011), and respiratory issues for both humans and livestock (Baxter and Horwell, 2015; Craig et al., 2016). Study of the microtextures of the ejected particles is a key tool with which to quantify the rate of ascent through the volcanic conduit, which determines the intensity of such eruptions. In particular, the size, shape, and number of vesicles and crystals preserve records of the ascent histories of the parent magma (Gardner et al., 1996; Kaminski and Jaupart, 1998; Mueller et al., 2008; Houghton et al., 2010; Rust and Cashman, 2011; Cashman and Scheu, 2015). Methods of clast sampling and analysis may bias microtextural analysis. Based on bulk deposit volumes and characteristic pyroclast sizes, we estimate that such Plinian eruptions eject between 1015 and 1020 clasts, so selection of a small number of representative clasts to quantify microtextures is an issue. A key factor is that the eruptions also disperse particles on length scales of hundreds of kilometers, and, during this transport, the population of tephra particles (pyroclasts) is sorted by both size and density. In order to ensure that microtextural studies are being applied to clast size fractions that are truly representative of the magma at fragmentation, it is critical to know how density, and thus vesicularity, varies with particle size.
There have been several different explanations for ranges of vesicularity or VG/VL (the ratio of gas volume to melt volume in a pyroclast [after Gardner et al., 1996]) in Plinian pumices. Gardner et al. (1996) found no significant change in VG/VL with clast size for 2 to 8 cm pumices, both between pumices from 13 Plinian eruptions and in samples collected over narrow stratigraphic intervals in each eruption. They suggested that the range of VG/VL values in a single sample reflects onset of connection and permeability at different times. Rust and Cashman (2011) extended this to propose that variations in vesicularity and permeability in silicic magma at the time of fragmentation drive the production of pyroclast subpopulations of different sizes.
Alternatively, Thomas et al. (1994) and Kaminski and Jaupart (1997) proposed that the range in vesicularity within samples of many pumices from single stratigraphic levels is due to varying amounts of postfragmentation expansion, which, in part, reflects the fact that clasts follow different paths in the eruption plume. We explore here the extent to which postfragmentation expansion is a function of clast size and whether collecting populations of small pumices (e.g., 16–32 mm) can minimize this influence on the range of vesicularity values.
Density analysis of large populations of juvenile pyroclasts is the first step to selecting smaller subsets of clasts with which to quantify vesicle textures using two-dimensional (2-D) and threedimensional (3-D) reconstructions (Shea et al., 2010). Density measurements are commonly conducted on sealed lapilli within the size range of 16–32 mm (Houghton and Wilson, 1989). This limited size range was deliberately chosen to avoid issues in:
(1) larger clasts (lapilli >32 mm and bombs), which may have undergone secondary, postfragmentation vesiculation of slowly cooled clast interiors (Benage et al., 2014; Polacci et al., 2001; Kaminski and Jaupart, 1997; Hort and Gardner, 2000) that can obscure the state of vesicularity at the point of fragmentation; and
(2) clasts < 16 mm, which are more difficult to measure accurately, and, more importantly, because, below some critical size limit, clast size and bubble diameter converge, resulting in a decrease in measured vesicularity with decreasing clast size (Houghton and Wilson, 1989).
However, an implicit assumption in measuring clasts only within the 16–32 mm lapilli size range is that this subpopulation, sampled at a single site, should be representative of the fragmented magma as a whole. There are several possible reasons why this may not be the case. First, if magma is of variable vesicularity at the time of fragmentation, then domains of the magma of a given vesicularity may be fragmented either more coarsely or more finely than those of other vesicularity. For example, it has been suggested that ash particles are produced from relatively impermeable melt domains characterized by populations of small bubbles, and lapilli and bombs form preferably from domains with coarser vesicularity and higher permeability (Cashman and Scheu, 2015; Rust and Cashman, 2011). Second, clasts that are 16–32 mm diameter may be the result of secondary breakage of much larger primary pyroclasts in the eruption plume or on landing that were affected by postfragmentation expansion and vesiculation. Finally, for large plumes, the sedimenting clast population may reflect both density as well as size fractionation in the umbrella cloud.
This study explores the validity of these assumptions, in the context of a very “typical” Plinian phase of moderate volume and intensity.
1.8 ka Taupo Eruption
The 1.8 ka Taupo caldera eruption in New Zealand (Wilson and Walker, 1985) is an ideal case for sampling and study of a diverse range of rhyolitic eruption products, taking advantage of fine temporal resolution and good preservation of the deposits. The products comprise seven eruptive units from at least three vent localities (Fig. 1A) within the modern Lake Taupo (Smith and Houghton, 1995). They include both dry (Plinian) and phreatomagmatic (phreatoplinian) tephra falls (Houghton et al., 2003) and both weakly and highly energetic ignimbrites during the latter stages of the eruption (Wilson and Walker, 1985). Unit 2, also known as the Hatepe Tephra, represents a sustained, moderately powerful Plinian phase of the eruption lasting ~7 h (Walker, 1981). The deposits are moderately well sorted and relatively uniform in grain size and vesicularity at each site (Fig. 1C), suggesting relatively steady, sustained eruption of a chemically and physically homogeneous source magma (Walker 1981). The pumice population has high bubble number densities, between 6 and 13 × 105 mm–3 (Houghton et al., 2010), and contains less than 5% crystals, making Unit 2 ideal for microtextural study.
By performing this detailed study of pyroclast density distributions from Unit 2, we seek to constrain issues and uncertainty associated with the common practices of clast sampling and density/vesicularity measurement from only a single narrow size range of pyroclasts.
The sample was collected in situ from a single proximal outcrop (Fig. 1A) that preserves intact the stratigraphy from Unit 1 through Unit 4 of the Taupo 1.8 ka eruption, where Unit 2 is ~3.5 m thick. The sample was collected from only the upper 0.5 m of the section, and so it is likely to represent approximately the final hour of eruption and deposition (Wilson and Walker, 1985). The sample was sieved in the field at 0.5 phi (Φ) intervals between –5Φ and –2Φ (>32–4 mm diameter). Two-hundred pumices were collected for each size fraction from –4.5Φ to –2Φ (4–22 mm), but only 115 clasts were collected for greater than –5Φ (>32 mm diameter), limited by restrictions of the total processed volume (400 kg).
All sampled clasts were brought back to University of Hawai‘i at Mānoa, where density analysis was carried out for approximately half of each collected size fraction using the Houghton and Wilson (1989) method, which recommends a sample size of 100 clasts for a representative distribution. The clasts were set out in size order and numbered, and then only odd-numbered clasts were used for density measurement, and even-numbered clasts were archived for use with other techniques in the future. Only 53 clasts were measured for the –5Φ size fraction (32–45 mm) due to the initial smaller sample size. For comparison, we also measured densities on 97 of the bigger than –5Φ clasts from the entire excavated sample, which had been measured for a separate grain-size exercise, namely, 14–6.5Φ (90–128 mm), 49–6Φ (64–90 mm), and 33–5.5Φ (45–64 mm) clasts. In total, 756 clasts were measured.
Following Houghton and Wilson (1989), all clasts were cleaned, washed, and dried at 110 °C for 48 h to remove trapped moisture and any coatings of ash. They were then sprayed with a hydrophobic-silicon coating of negligible weight and volume, dried for 24 h, and then weighed dry. Clasts were then submerged in water and weighed under a ballast of known weight and volume to determine “wet” weights of the clasts. Density was then calculated by Archimedes principle,
Shape Parameterization and Secondary Fragmentation during Transport
Prior to density analysis, a qualitative preservation index representing the extent of postfragmentation breakage was assigned to individual clasts based on whether each face reflected viscous/fluidal or brittle breakage. Smooth, gently rounded, irregular faces with uniform microvesicularity were inferred to be original fragmentation surfaces, whereas clasts with some planar sharp-edged, broken faces, and/or larger vesicles defining “frothy” interiors were interpreted as broken in transport (Fig. 2). Every clast was independently assigned a preservation index by three people, and then assignments were discussed before a final value was assigned. Clasts were grouped into bins of 0%–20%, 20%–40%, 40%–60%, 60%–80%, and 80%–100% original surface area (Figs. 2A, 2B, and 2C). In total, 716 clasts were quantified from –5Φ to –3.5Φ, i.e., 11–32 mm; the –6.5Φ to –5.5Φ clast sets (45–128 mm) were too small for a representative distribution, and clasts below –3.5Φ (<11 mm) were too small to assess the surface textures accurately. Each clast from –6.5Φ to –2.0Φ (4–128 mm) was also classified as either dominantly macrovesicular (i.e., containing abundant millimeter-sized vesicles) or dominantly microvesicular (Figs. 2D and 2E).
Table 1 summarizes the statistical parameters calculated for our populations. Results of the density analysis (and the corresponding vesicularity values) are first shown as box-and-whisker plots in Figure 3, from which a conspicuous shift of average density appears between ≥–5Φ and ≤–4.5Φ (at 32 mm clast diameter). This is seen in Table 1, where densities for diameters ≥–5Φ and ≤–4.5Φ cluster around 600 and 700 kg m–3, respectively (i.e., corresponding to vesicularities of 74% and 71%, respectively). Density distributions for clasts ≥–5Φ (≥32 mm) are dominantly characterized by a negative skewness, denoting a shift of the distribution toward lower densities, and this trend is inverted for clasts ≤–4.5Φ (<32 mm). Although distributions are slightly skewed, the similarity between mean and median density values and the distribution of the percentiles (Table 1) suggest that density distributions were initially symmetrical.
Although a microvesicular morphology is always dominant (Fig. 4), dominantly macrovesicular pumices represent 27% of the −5Φ (32–45 mm) population versus 8% for the −4.5Φ (22–32 mm) population (Table 1). Macrovesicular pumices are characterized by lower densities, especially for diameters ≤−5Φ (≤32 mm). From Figure 4, it appears that microvesicular morphologies are absent in clasts with densities <400 kg m−3. Similarly, few macrovesicular clasts have densities >1000 kg m−3.
We interpret the coarsening of the vesicle population at −5Φ (32 mm) as marking the onset of secondary, postfragmentation vesiculation.
Clast Density and Secondary Fragmentation
Figure 5 summarizes the relationships among diameter, density, and preservation index and reveals an increasing proportion of broken clasts with increasing diameter irrespective of clast density. For instance, in the −3.5Φ (11–16 mm) population, only 20% of clasts were quantified as <60% original surface area, in comparison with 50% in the −5Φ (32–45 mm) population; these smaller clasts tended to have smoother, less irregular and rough faces, as seen in Figure 2C.
For a given clast size, densities remained relatively constant with changing preservation indices, as shown by the overlap of interquartile ranges (i.e., 25th–75th percentile interval; box in Fig. 5). This suggests that the broken clasts within the smaller size fractions are not necessarily always from only cores of larger clasts that have experienced secondary vesiculation. One exception can be observed in the case of −4.5Φ (22–32 mm), where the 0%–20% preservation bin shows an asymmetrical distribution skewed toward lower densities. This observation should, however, be considered in the context of the small subpopulation (i.e., 5 clasts).
The density comparison of rapidly quenched exteriors and broken frothy interiors, as illustrated in Figure 2, shows that the low-density clasts in the –5Φ and larger subsamples (32– 128 mm) are largely a function of the existence of frothy, dominantly macrovesicular interiors inferred to be postfragmentation in origin.
The clasts within the –3.5Φ fraction (11–16 mm) have a slightly lower mean density than neighboring bins (Fig. 3A), and this is the result of a larger number of less dense, broken clasts with 60%–100% original surface (Fig. 5). This signal potentially records generation of a preferred clast size by the breakage process, but no theoretical nor experimental data yet exist to support this idea. A few outlying clasts (very high or very low density) seen on the tail ends of the histograms (Figs. 3A and 5) could account for the shifting of the –3.5Φ (11–16 mm) and –5Φ (32– 45 mm) distributions.
In general, it appears that density distributions for clasts <–5Φ (<32 mm) shift to slightly higher densities (Fig. 3A) when we consider only those clasts with >60% or >80% original surface area, i.e., of the dominant subset for that clast size. At –5Φ (32–45 mm), the density remains relatively constant whether considering all clasts or just those that are unbroken (>80% original), and the mean density remains constant from –5Φ to –6.5Φ (32–128 mm; Fig. 3A). This has implications for how we consider the selection of a representative subset of clasts appropriate for density analysis.
DISCUSSION AND CONCLUSIONS
Threshold for Significant Postfragmentation Vesiculation
The data suggest that, at Taupo, there is a clearly defined threshold such that clasts finer than –4.5Φ (<22 mm) can be considered as “instantaneously” quenched and as such representative of the physical state of the magma close to fragmentation. The interpretation of any coarser clasts is more enigmatic and must take into account postfragmentation processes of continued vesiculation and coalescence in the shallow conduit and the eruption plume (Thomas et al., 1994; Kaminski and Jaupart, 1997).
Consequences of Selecting Clasts That Have Undergone Postfragmentation Vesiculation
There are some limited published quantitative data showing the effect of postfragmentation expansion of the textures of pumice at Taupo (Houghton et al., 2010). Two of the measured Unit 2 clasts have densities of 560 kg m–3 and 350 kg m–3 (vesicularities of 76% and 85%) and show microvesicular and coarsely vesicular textures, respectively (Figs. 6A and 6B). The clasts represent the modal density (A) and low-density tail (B) of a population of 498 clasts in the 16–32 mm size range (Fig. 7, left). Clast B is identical to those clasts in this study that we infer underwent postfragmentation expansion vesiculation. The images in Figure 6 show that its higher vesicularity reflects a general coarsening of the vesicle population, i.e., greater abundances of relatively large bubbles across all magnifications, with respect to clast A. Figure 7 (right) shows that clasts A and B have similar bubble number densities, indicating that postfragmentation expansion of clast B was accomplished largely by expansion of existing bubbles, as suggested by Toramaru (1990) and Thomas et al. (1994). Unit 2 is the product of moderate-intensity Plinian phase (Wilson and Walker, 1985). Pyroclast pairs from other phases of the eruption with higher eruption rates (Units 5 and 6), and hence greater thermal insulation in either a higher eruption plume (Unit 5) or a moving pyroclastic density current (Unit 6), show even more marked contrasts (Fig. 7, right), as do Plinian phases from several other major eruptions (e.g., Kaminski and Jaupart, 1997). Kaminski and Jaupart (1997) interpreted this to be due to the great potential for bubbles to expand at high rates of decompression.
Vesicle sizes and number densities are measured principally to establish syneruptive conditions at the time of fragmentation (e.g., Gardner et al., 1996; Houghton et al., 2010; Rust and Cashman, 2011). If clasts that have undergone postfragmentation vesiculation were chosen for quantitative measurements of vesicularity, it would appear that the most significant consequence will be in overestimating vesicle sizes (and the role of free growth in the melt prior to fragmentation), rather than in vesicle number densities.
Role of Breakage of Clasts during Transport and upon Deposition
Breakage is seen to be more likely in larger clasts, and this has the effect of “adding” particles into the smaller size bins that may be either microvesicular and denser or macrovesicular and lighter in density, depending on whether they were derived from the core or rim of a larger clast affected by postfragmentation expansion. This suggests the statistically most representative results will be obtained from 16 to 32 mm clasts, which preserve most of their original surfaces. Breakage of clasts in the plume and on landing will also affect the apparent transport and sedimentation patterns from Plinian plumes.
Recommendations for Sampling
Several criteria should be considered in field studies sampling tephra-fall deposits for vesicularity work. We suggest that sampling for density and vesicularity studies should focus on the size population of 16–32 mm (–4Φ) as recommended by Houghton and Wilson (1989), but sample numbers should be expanded to 200 clasts to allow for the discarding of broken, originally large clasts and macrovesicular outliers for statistically viable componentry analysis. While recommending use of –4Φ (16–32 mm) clasts to represent vesicularity at magmatic fragmentation, we recognize that the 32 mm threshold identified within this study may not be appropriate for every pyroclastic deposit. Ideally, each study would first identify the size threshold for secondary vesiculation to be certain that –4Φ (16–32 mm) will be truly representative. Future studies should be prepared to discount broken fragments of larger clasts that may skew the truly representative distribution of primary 16–32 mm clasts, so a basic evaluation of componentry and preservation index is essential. By accounting for the influence of the processes of secondary fragmentation and continued vesiculation in erupted clasts, more appropriate density ranges and vesicle size distributions can be used in conduit ascent and fragmentation models for improved understanding of explosive eruptions such as the 1.8 ka Taupo Plinian eruption. Equally well, subsets of clasts larger than –5Φ can be selected and targeted for future work to understand the nature of postfragmentation vesiculation and expansion, e.g., the relative roles of nucleation of new bubbles versus expansion of the synfragmentation population.
This research was supported by National Science Foundation grant EAR-0739060 and subcontracts from the New Zealand Foundation for Research, Science and Technology via GNS Science. We acknowledge significant contributions from Thomas Giachetti, an anonymous reviewer, and editor Anita Grunder.