SEM-based methods for the analysis of basaltic ash from weak explosive activity at Etna in 2006 and the 2007 eruptive crisis at Stromboli

Physics and Chemistry of the Earth 45–46 (2012) 113–127

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth journal homepage: www.elsevier.com/locate/pce

SEM-based methods for the analysis of basaltic ash from weak explosive activity at Etna in 2006 and the 2007 eruptive crisis at Stromboli Nicole C. Lautze a,b,⇑, Jacopo Taddeucci a, Daniele Andronico c, Chiara Cannata d, Lauretta Tornetta b, Piergiorgio Scarlato a, Bruce Houghton b, Maria Deborah Lo Castro c a

Istituto Nazionale di Geofisica e Vulcanologia, Roma, Italy University of Hawaii, Manoa, Department of Geology and Geophysics, United States Istituto Nazionale di Geofisica e Vulcanologia, Catania, Italy d Univerista della Calabria, Italy b c

a r t i c l e

i n f o

Article history: Received 1 October 2010 Received in revised form 22 January 2011 Accepted 2 February 2011 Available online 5 March 2011 Keywords: Ash Morphology Chemistry Etna Stromboli Volcano

a b s t r a c t We present results from a semi-automated field-emission scanning electron microscope investigation of basaltic ash from a variety of eruptive processes that occurred at Mount Etna volcano in 2006 and at Stromboli volcano in 2007. From a methodological perspective, the proposed techniques provide relatively fast (about 4 h per sample) information on the size distribution, morphology, and surface chemistry of several hundred ash particles. Particle morphology is characterized by compactness and elongation parameters, and surface chemistry data are shown using ternary plots of the relative abundance of several key elements. The obtained size distributions match well those obtained by an independent technique. The surface chemistry data efficiently characterize the chemical composition, type and abundance of crystals, and dominant alteration phases in the ash samples. From a volcanological perspective, the analyzed samples cover a wide spectrum of relatively minor ash-forming eruptive activity, including weak Hawaiian fountaining at Etna, and lava-sea water interaction, weak Strombolian explosions, vent clearing activity, and a paroxysm during the 2007 eruptive crisis at Stromboli. This study outlines subtle chemical and morphological differences in the ash deposited at different locations during the Etna event, and variable alteration patterns in the surface chemistry of the Stromboli samples specific to each eruptive activity. Overall, we show this method to be effective in quantifying the main features of volcanic ash particles from the relatively weak – and yet frequent – explosive activity occurring at basaltic volcanoes. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction The emission of volcanic ash is a common occurrence and can have far reaching effects – as clearly exemplified in April 2010 with the travel crisis in Europe caused by activity at Eyjafjallajokull volcano in Iceland. Globally, eruptions including those of much lower intensity than at Eyjafjallajokull eject > 1,000,000 m3 of ash into our atmosphere on a monthly basis (Simkin and Siebert, 2000), often affecting water supplies, crops, air and road traffic and climate on a local to international scale. Though a nuisance to humans, its small particle size, low terminal velocity and widespread distribution makes ash the safest and simplest volcanic particle to collect in real-time. Within the last decade, studies have linked variations in the componentry of ash emitted during explosive eruptions to ⇑ Corresponding author. Address: University of Hawaii, Manoa, Department of Geology and Geophysics, 1680 East West Rd., POST 606A, Honolulu, HI 96822, United States. E-mail address: [email protected] (N.C. Lautze). 1474-7065/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.pce.2011.02.001

changing eruptive styles (Andronico et al., 2005, 2009a, 2009b; Taddeucci et al., 2002), and used the ejection of juvenile ash as a precursor to larger explosive eruptions (Cashman and Hoblitt, 2004; Watanabe et al., 1999). Yet still, there is a paucity of published studies focused on characterizing the textural properties of ash (relative to lapilli and bombs), probably because its small size makes ash inherently difficult to analyze. The studies that do exist highlight a longstanding gap between grain-size/componentry analysis (e.g. Cas and Wright, 1987; De Rosa, 1999; Fisher and Schmincke, 1984), and scanning electron microscope (SEM) studies (e.g. the pioneering works of Heiken and Wohletz, 1985; Sheridan and Marshall, 1983; Wohletz, 1987). The former analyses are generally conducted using sieves, automated particle analyzers, and/or binocular microscopes. These studies provide fast, statistically robust information on a large number of particles, but lack detailed shape and morphological information. SEM studies in general provide detailed information, but on only a small number of particles within a sample.

114

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

Recent progress has been made toward automating methods to obtain grain size and particle morphology data for ash samples (Dellino and La Volpe, 1996; Dellino and Liotino, 2002; Ersoy, 2010; Ersoy et al., 2007; Kueppers et al., 2006; Maria and Carey, 2002; Riley et al., 2003). However the attainment of automated, quantitative data for the surface chemistry of volcanic particles until recently remained limited to volcanic aerosols (Martin et al., 2008), despite the wide application and usefulness of surface chemistry data, e.g. in aiding the establishment of component classes (Cioni et al., 2008), in tephrostratigraphy (Sulpizio et al., 2008), in unraveling eruption plume chemistry (Delmelle et al., 2007), which has further implications for the health and environmental aspects of ash transport and deposition (Horwell et al., 2003). Italy’s Istituto Nazionale di Geofisica e Vulcanologia (INGV) is in charge of monitoring active and quiescent volcanoes nationally. Recently, scientists developed ash collection and analysis tools specifically suitable for eruption monitoring, mainly at Etna and Stromboli volcanoes (Andronico et al., 2009a; Taddeucci et al., 2002). Such techniques enable largely automated, relatively quick, and quantitative classification of the morphoscopy and surface chemistry of a large number (hundreds) of ash particles using a SEM equipped with particle analysis software (Taddeucci et al., 2007). Here we test the effectiveness and robustness of such methods on multiple ash samples from two study cases: an explosive, Strombolian to Hawaiian, episode at Etna on 24 November 2006 (Andronico et al., 2009b), and the February–April 2007 eruptive crisis of Stromboli (Barberi et al., 2009; Mori and Burton, 2009; Calvari et al., 2010). The latter includes ash from lava-sea water interaction, the 15 March paroxysm, and weaker Strombolian explosions at the summit craters. Our data show that the methods are efficient in characterizing and distinguishing ash particles collected at different locations (Etna case), and from different sources (Stromboli case), and also give insight into the particle source and eruptive dynamics at both volcanoes. In particular, the different sources of ash at Stromboli have distinctive alteration signatures, while the Etna samples highlight subtle differences in grain size and componentry. This and future studies focused on ash at basaltic volcanoes are seminal given that the fragmentation processes leading to ash from Hawaiian and Strombolian style explosions are unclear (Patrick et al., 2007). 2. Methods Sampling strategies employed at each volcano are described below. All ash particles were analyzed in the state they were collected, i.e. without washing or chemical treatment, and without any sorting by grain size. The Etna samples consisted of only several hundred grains, so the complete sample was analyzed. The Stromboli samples each consisted of thousands of ash particles, from which a subset of 500–1000 was randomly selected by scooping a fraction of ash from the container that housed the entire ‘randomly-sorted’ sample. Ash grains were mounted on a 3 mm piece of double sided carbon tape adhered to a metal stub, then loaded into the SEM. The JEOL JSM 6500F Field Emission (Schottky-type) scanning emission microscope (FE-SEM) at the INGV-Roma was used to obtain data regarding texture (vesicularity), morphoscopy, and surface chemistry of ash grains, with the latter two parameters obtained using the Particle Analyzer tool of the JED 2200 software package. In general, the complete analysis of one sample took 3– 5 h on the SEM. 2.1. High resolution imaging In comparison to conventional SEMs, FE-SEMs offer a more stable electron source and a smaller beam capable of higher spatial

resolution at a lower acceleration voltage. The nominal resolution of the FE-SEM at INGV-Roma is 1.5 and 3 nm at 15 and 1 kV voltage acceleration, respectively. Magnifications up to 100,000 were used to visualize sub-micron size features of the ash. The carbon coating was visible at higher magnifications such that use of a noble metal coating was necessary. 2.2. Morphoscopy Morphoscopic data were obtained by first setting the brightness and contrast of the FE-SEM so that ash particles appeared bright against the dark background of the sample holder. The Particle Analyzer software was then programmed to automatically acquire backscattered electron (BSE) images over the entire sample holder. As a compromise between image resolution and acquisition time, images were acquired at 512  384 pixels, and at a single magnification chosen such that it was low enough to include the largest particles in a single image but high enough to measure the smallest particles. Within each image, the area, perimeter, rectangularity, elongation, circularity, compactness, Feret (maximum) diameter and Heywood (equivalent) diameter of each particle was automatically calculated – generally for a total of 500–1000 particles. In this study we report results for the parameters of compactness, elongation, and Heywood diameter. Compactness was calculated as the ratio between the area of the minimum rectangle circumscribed by the particle and the particle area, with a non-dimensional value between 0 and 1 (1 being a square). Irregular and rounded particle edges decrease the compactness parameter. Elongation was calculated as the square of the longest particle segment divided by the particle area (Dellino and La Volpe, 1996), such that the more elongate the particle, the higher the elongation parameter. Heywood diameter was calculated from the particle area as the diameter of a circle with the same area. The morphoscopic analyses were determined to be orientation invariant through repeat analysis on rotated particles. As a test for the effectiveness of FE-SEM based grain size distribution analysis we also performed, on duplicate samples, grain size analyses by an automated, digital image processing-based particle size analyzer (CAMSIZERÒ, see http://www.retsch-technology.com/rt/ products/digital-image-processing/camsizer/function-features/ for more information) at INGV-Sezione di Catania. 2.3. Surface chemistry Surface chemistry analyses were conducted using the energy dispersion system (EDS) of the FE-SEM. Within the Particle Analyzer software and using images obtained in the morphoscopic analyses, a rectangular area was defined on >100 randomly chosen particles. An X-ray spectrum from the surface of each particle within the rectangle was then automatically acquired and converted into a standardless quantitative chemical analysis for specified elements. Since the X-ray spectrum was obtained over a large portion of the particle surface, the corresponding analysis is an average of the phases cropping out in the scanned area, possibly including pristine glass, crystals, and alteration phases in variable proportions. This technique provides ‘‘bulk’’ information on the glass composition, crystallinity and degree of secondary alteration of the particle (Taddeucci et al., 2007) and was used to effectively characterize the chemical composition of magnetic particulate matter in the city of Rome (Sagnotti et al., 2009). The analytical error associated with spectrum quantification is within 10% of the measured value (5% for oxides with abundances >10% mass tot.), as deduced by repeated analyses and a comparison with electron microprobe analyses performed on basaltic glasses (see Appendix A). A larger and less-explored source of error is the effect of particle roughness and orientation relative to the

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

115

EDS detector. This source of analytical variability was investigated by performing repeated analyses on the same clast, and by rotating the particle with the respect to the EDS detector. These results showed that particle rotation has a relatively minor effect, on the order of 15% of the measured value, which is incorporated in the error bar for each plot (see Appendix A). The automated whole grain morphoscopic and EDS analyses can provide guidance to user-operated analysis, such as imaging specific grains or features at higher magnification, and/or ‘‘spot’’ chemical analysis, e.g. to confirm the composition of crystals or alteration phases. 3. Etna case study Mt. Etna (Sicily, Italy) is well-known for its frequent episodes of effusive and explosive volcanic activity, which occur at any 1 of 4 summit craters located between 3000 and 3300 m a.s.l., or along two rift zones that mark its flanks. More than 140 explosive events have occurred since 1995 (Alparone et al., 2003, 2007; Andronico et al., 2009a; Coltelli et al., 1998, 2000), with many of these causing the closure of Catania or Reggio Calabria airports and disrupting other infrastructure. Routine monitoring of Etna is conducted by scientists within the INGV-Sezione di Catania. 3.1. Activity and sampling The 24 November 2006 explosive event was among the most intense of 18 explosive episodes at Etna’s Southeast Crater (SEC) that occurred between 30 August and 15 December 2006. This 3.5 months of activity was characterized by an increasing intensity of transient explosive events, each of which formed a weak ash plume (e.g. Fig. 1), and lava effusion from a fissure at 2800 m a.s.l. (below and east of the base of the SEC; Andronico et al., 2009a, 2009b). Andronico et al., 2009b describes the 24 November explosive event in detail. The eruptive onset was preceded by an increase in seismic tremor, and recorded by a thermal camera within the INGV monitoring network – weather conditions were excellent. Explosion style was characterized as strong Strombolian to Hawaiian. The onset of eruptive activity occurred at 6:00 GMT and the to-

Fig. 2. Digital elevation map showing location of the eight Etna samples, represented by their number. Red line shows extent of ash fallout from the 24 November 06 event. Blue areas represent cities, as labeled. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

tal eruption duration was 10 h, with nearly continuous ash fallout occurring over nearly 12 h. Shifting wind directions from the SE to the S between 11:00 and 12:00 GMT caused rotation of the plume and broad ash fallout. The column height has been estimated at 1.3 km above Etna’s summit (Fig. 1; Andronico et al., 2009b). A total of 30 ash samples were collected from different sites by Andronico and Lo Castro between a distance of 1 and 80 km from source, with analysis of eight of the samples reported here (Fig. 2). Samples were collected in the order numbered – 5, 6, 8,

Fig. 1. Photo of ash plume from Etna 24 November 2006, taken from south, with village of Ragalna in foreground. Plume height is 1.3 km (Andronico et al., 2009b).

116

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

10, 12 and 20 were collected between the late morning and evening of the 24th, and samples 22 and 25 on the morning of the 25th. Considering the time of sampling, samples 5, 6, 8, and 10 are associated with approximately the first half of activity, and samples 12 and 20 almost the whole duration. Given the shift in wind direction, samples 22 and 25 were deposited during the last half of the eruption. 3.2. Data and observations 3.2.1. General observations & componentry For the Etna study, SEM data was combined with componentry analysis performed using a binocular microscope. The componentry

analysis (carried out by A. Cristaldi) involved the definition of classes and assignment of at least 500 ash particles per sample to the class (Andronico et al., 2007). All grains in each sample were characterized on the basis of appearance under binocular microscope, including color, shape, and texture. Four component classes were identified in each sample: sideromelane, tachylite, crystals, and wall-rock lithics. Consistent with previous Etna ash studies (Andronico et al., 2009c; Taddeucci et al., 2004), sideromelane is described as vesicular, glassy, and of fluidal form, and tachylite as slightly to nonvesicular, having a micro-crystalline groundmass, and blocky in form. Crystals include plagioclase, pyroxene, and olivine. Lithics are altered, mostly reddish particles. Sideromelane and tachylite

22

10

12

20

5

6

25

8

Fig. 3. SEM image of one grain from each Etna sample, roughly organized as proximal (top) to distal (bottom). Grains from samples 12, 20, 25, and 6 were classified as sideromelane, and 22, 10, 5, and 8 as tachylite. Scale bar in upper left of each image is 100 lm.

117

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

Fig. 4. Histogram showing results of componentry analysis for each Etna sample, roughly arranged as proximal (left) to distal (right). Lithics dominate in all samples, however note the decrease in crystals and increase in juvenile content (tachylite and sideromelane) with increasing distance from vent (modified from Andronico et al., 2007).

% of Clasts

5

25 20

12

% of Clasts

% of Clasts

30

15 10 5 0

0

50

100

150

200

250

300

350

Diameter (µm) 40 35

% of Clasts

3.2.2. Grain size & morphoscopy Fig. 5 shows the results of the automated FE-SEM analysis of equivalent diameter, compactness, and elongation for the eight Etna samples. To assess repeatability, the equivalent diameter and compactness analyses were performed on two different sets of particles from Etna sample 8, labeled as 8 and 8bis. The variability between the two measurements of sample 8 is significantly smaller than the variability between different samples. This indicates that differences in the morphoscopic parameters from sample to sample exist, and lends confidence in the technique to discern them. Results of the size analysis show that particles with diameters between a few to 350 lm were measured, with sample modes between 50 (sample 10) and 220 (sample 22) microns. Sample 22, the most proximal, has the coarsest mode, and sample 8, the most distal, has among the smallest mode diameter (80 lm), however there is no clear trend in decreasing size with distance in the intermediate samples, probably because of the complex dispersal pattern with time. Most samples show a unimodal distribution, with variable degree of sample sorting. Samples 20, 8, and 5 show a relatively tight distribution (better sorted), while samples 25, 22, and 10 show wider distributions, indicative of a general trend of better sorting with distance. The secondary bimodal peak of some samples (12, 6, 5, 20) could be due to transport of small, weakly adhering particles on the surface of the coarser ash grains. Sample 10 has perhaps the ‘oddest’ distribution, with a mode at 50 lm. Fig. 5

CAMSIZER FE-SEM

35

30 25 20

22 10 12 20 5 6 25 8 8 bis

15 10 5 0 0.2 0.3 more irregular

0.4

0.5

0.6

0.7

0.8 0.9 more rectangular

Compactness

30 25

% of Clasts

are considered to represent magma that quenched at fragmentation (i.e. juvenile particles). Fig. 3 shows a SEM image of one grain from each sample, and is intended to highlight the contrast between sideromelane and tachylite that is notable in SEM images. The images are organized as roughly proximal to source (top left) to distal (bottom right). Grains from samples 22, 10, 5, and 8 were classified as tachylite, and from samples 12, 20, 6, and 25 as sideromelane. Fig. 4 is a histogram showing the results of the componentry analysis, with samples organized as roughly proximal (left) to distal (right). Lithics comprise over 50% and crystals under 5% of each sample. The sideromelane to tachylite ratio is roughly equal in most samples. Only in samples 5, 6 and arguably 8 is the sideromelane content higher than the tachylite content. Interestingly, the crystal content roughly decreases and juvenile content (tachylite and sideromelane) roughly increases with distance from source. Sample 22, collected closest to source, has the highest abundance of crystals and lithics, and the proportion of juvenile material generally increases with distance from source (to the right).

20 15 10 5 0 1 1.5 less elongate

2

2.5

Elongation

3

3.5 more elongate

Fig. 5. Results of morphological analysis for Etna samples. 8bis is plotted to show methodological error, as shows data from a different set of grains from sample 8. Legend is the same for all plots, and lists samples in order of most proximal (top) to most distal (bottom). The upper right inset shows the grain size distributions of samples 5 and 12 under FE-SEM (in red) and CAMSIZERÒ (in black).

118

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

(a)

(b)

(c)

(d)

Fig. 6. Ternary plots of surface chemistry data for all Etna samples. Also plotted are typical values of Etna’s bulk composition (B), melt (M) and crystals of olivine (Ol), clinopyroxene (Cpx), plagioclase (Plag), oxide (Ox), gypsum (Gyp), and halite (Ha; from Corsaro and Pompilio, 2004 and Metrich et al., 1993). Plots (a–c) show that in general ash has chemistry typical of Etna melt, with some influence of crystals. Plot (d) shows that all samples are variably altered by sulfur.

also shows that the grain size distributions obtained by FE-SEM and by CAMSIZERÒ are comparable, supporting the robustness of our methodology. In general, compactness measurements for all samples show a broadly uniform distribution that peaks between 0.6 and 0.7. Samples 20, 10, 6, 12 and 25, have relatively sharp peaks at 0.68, sample 5 peaks at 0.65, and samples 22 and 8 (the most proximal and distal, respectively) have broader distributions that peak at 0.60. This indicates no clear relationship between particle compactness and distance from source, although the broadest distribution of all samples is 22, by far the most proximal, which may be due to an abundance of altered grains and/or adhering particles. Alternatively, in the proximal region, we can expect particles that were shed from a wide range of heights on the plume margin and hence with contrasting size and shape characteristics. All samples have an elongation parameter that peaks between 1.5 and 2, indicating that most erupted particles are more equant than elongate – probably related to the high lithic content. Samples 22 and 8 contain the most elongate particles, consistent with the irregularity of these samples as noted in the compactness analysis, and also indicating no clear trend of elongation with distance from source.

mately 50 of >100 randomly selected EDS points/grains per sample are plotted to prevent saturation of points on the plots. These oxides were chosen to highlight chemical differences in the glass, the presence of crystals of plagioclase, pyroxene and oxides, and to show alteration by sulfur compounds. The composition of typical compositions of Etna glass and crystals (Corsaro and Pompilio, 2004; Métrich et al., 1993) are also plotted. Although the plots show significant overlap, compositional differences between the samples are apparent (samples 8 and 20 for example). Plots (a–c) show that most ash particles in all samples have the surface chemistry of Etna melt, with some influence of crystals. Plot (d) shows that all grains have been variably altered by sulfur (Etna glass typically has 0.3 wt.% sulfur; Métrich et al., 1993). There is no apparent correlation between samples classified by componentry as having a relatively high crystal content (e.g. samples 22, 10; Fig. 4) and those which show a strong trend towards one or more crystals in the chemistry data. Notable in the chemistry data is that sample 20 has a consistent trend towards plagioclase, and sample 8 (the most distal) a trend towards gypsum.

3.2.3. Surface chemistry Results of the surface chemistry analysis are shown as ternary plots of relative mass abundance of FeO–SiO2–Al2O3, FeO–Al2O3– MgO, CaO–Al2O3–MgO, and FeO–SO3–Na2O in Fig. 6a–d. Approxi-

Stromboli volcano (Aeolian Islands, Italy) is well-known for its persistent, mildly explosive ‘Strombolian’ activity, as described by numerous authors from Pliny the Elder in Roman Age to, for example, Chouet et al. (1974) and Patrick et al. (2007). Also well

4. Stromboli case study

119

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

established is the infrequent occurrence of ‘major explosions’ (1/ year), effusive phases (1/decade), and violent ‘paroxysms’ (Barberi et al., 1993), the last which occurred in 1930, 2003, and 2007. Explosions at Stromboli originate from one of three wellestablished sources (Northeast (NE), Central (C), and Southwest (SW)) located 750–800 m a.s.l. Larger tephra (lapilli and bombs) typically either accumulate as small cones along a terrace or roll down a scar known as the Sciara del Fuoco (Fig. 7), while ash is often more widely distributed around the island. Based on observations and thermal imaging at Stromboli, Patrick et al. (2007) classify explosions as Type 1 (ash-poor) and Type 2 (ash-bearing). Why some explosions emit ash and others not remains unclear. Patrick et al. (2007) propose this difference is a function of the presence or absence of brittle material capping the vent, and offer several possible sources of such brittle material (e.g. slumping of crater walls, changing magma rheology). Unlike Etna and despite the abundant literature on Stromboli, no study has focused on the textural properties of its ash, nor is

there an existing componentry scheme for ash from Stromboli. Only Schiavi et al. (2009) analyzed ash emitted during the 2003 eruptive crisis at Stromboli, but from a geochemical perspective. They give a basic description of the ash components as crystals, lithics (blocky, grayish to reddish), and glassy fragments, with a further subdivision of glassy fragments as high porphyritic (HP; e.g. Corsaro et al., 2005; Francalanci et al., 2004) or low porphyritic (LP; e.g. Bertagnini et al., 2003) on the basis of chemistry (with LP matching Stromboli’s less evolved golden pumice; Métrich et al., 2005, 2010) and physical characteristics (HP is amber to dark brown and poorly vesicular; LP is gray–yellow, light to transparent in thin section, and strongly vesicular). Here we present data from seven samples of ash emitted from various sources during, and a few weeks after, the 2007 eruptive crisis at Stromboli (Table 1). We anticipate this will provide a foundation for a more universal textural characterization of Stromboli’s ash and ash from Strombolian activity.

Fig. 7. Digital elevation map of Stromboli showing location each ash sample was collected. The villages of Stromboli and Ginostra are labeled, in addition to the headquarters of the Centro Operativo Avanzato (COA) and the Istituto Nazionale di Geofisica e Vulcanologia (INGV).

Table 1 Stromboli Sample Characteristics. Sample No.

Vitrous clasts

Micro-crystalline clasts

Crystals

Lithics

Aggregates

Adhering particles

Gypsum

Halite

Probable origin

280207 30307 150307 180307a 190307a 100407 290407

XX XX XXX XX X X X

XX XX O X XX XX XX

X X X X XX X X

XX X O XX XX X X

O X X O O XXX XXX

X XX XX X X XX XX

X X XX X X X X

XXX O X XXX O O O

Sea entry Summit craters & ocean entry Paroxsym Sea entry Summit craters? Summit craters Summit craters

O: absent, X: present, XX: abundant, XXX: very abundant.

120

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

4.1. Activity and sampling The volcanic activity during the 2007 effusive eruption and paroxysm is described in detail in Barberi et al. (2009) and Calvari et al. (2010). At 12:00 GMT on 27 Feb 2007 a fissure began propagating within the NE crater. Forty-eight minutes later a lava flow was noted on the northern rim of the Sciara del Fuoco, which reached the sea the same day. Lava effusion continued through 2 April with entry into the sea during most of this time. Occasional small ash explosions at the summit craters occurred before and throughout the period of effusive activity, and an explosive paroxysm occurred on 15 March. The paroxysm ejected meter-sized blocks to a distance of 1.5 km, and was associated with an impulsive increase in the lava effusion rate for 9 min prior and several hours after the event (Barberi et al., 2009). We carried out analysis of seven samples collected at the locations shown on Fig. 8. The sample number refers to the day– month–year it was collected (e.g. sample 280207 was collected on 28 February 2007). All samples were collected in ‘real-time’ (e.g. as ash fallout occurred) with the exception of 280207 and 030307, which collected in a container over a 1–3 day period. Samples ejected after 15 March were collected in Stromboli village due to restricted summit access after the paroxysm. 4.2. Data and observations 4.2.1. General observations A randomly selected subset of grains was examined using a binocular microscope. We found that the componentry scheme used at Etna does not adequately characterize these Stromboli samples, which show a much larger variability and less defined boundaries between particle categories. As a starting point, in this study we considered an end-member of rounded ash particles with a highly altered surface to be recycled, or lithic, while another end-member are those with sharp edges and/or a slightly or not altered surface, as fresh, or juvenile. Juvenile clasts can include the subcategories tachylite, sideromelane, and crystals. In between the two endmembers are grains that appear to be ‘older’ tachylite, sideromelane and crystals (e.g. that are variably altered and/or rounded). However, exceptions, in particular the alteration of some otherwise juvenile particles, were also noted. This was especially evident in the paroxysmal sample 150307 in which some angular but altered particles had the more primitive composition of golden pumice that is associated only with major explosions and paroxysms (Métrich et al., 2005, 2010). The morphoscopic and chemical SEM analyses were performed on a randomly selected subset of 500–1000 grains per sample. Three SEM images of particles in each sample are shown in Fig. 8. Given the high variability among the Stromboli samples, we begin with a brief description of each, which corresponds to characteristics given in Table 1. The composition of surface characteristics was confirmed using the spot function of the EDS. Morphological and whole grain surface chemistry data (Figs. 9 and 10, respectively) will be discussed subsequently. 4.2.1.1. Sample 280207. The 28 February sample, which collected over between 24–48 h, is predominantly composed of variably altered, blocky, non-vesicular clasts that have either a glassy (Fig. 8a) or micro-crystalline (Fig. 8b) groundmass. Crystals of plagioclase, pyroxene, and rarely olivine are present, and the secondary evaporate minerals of halite and carnallite (as inferred from spot SEMEDS analysis) [(KMgCl3
6(H2O)] occur in abundance. Fig. 8c shows a crystal of halite containing a fragment of glass. 4.2.1.2. Sample 030307. The 3 March sample collected over a threeday period. The majority of clasts are either micro-crystalline and

blocky, or glassy and elongated/irregularly shaped (Fig. 8d, and e), with the two types present in roughly equal amounts. Crystals of plagioclase and pyroxene are present in small amounts, and gypsum is the most abundant secondary phase. Halite is present but rarely, although an anomalous concentration of chlorine exists in the analyzed glass. Glassy grains > 10 lm composed almost exclusively of silica occur. Micron to sub-micron particles often adhere to larger clasts (e.g. Fig. 8d, and f) and also common are particle aggregates. The grain shown in Fig. 8f, with rounded edges and a high degree of alteration is interpreted to be recycled. 4.2.1.3. Sample 150307. The 15 March sample represents the paroxysm that occurred the same day. It is composed exclusively of glassy shards (Fig. 8g, and h) and crystals (principally plagioclase). Micro-crystalline clasts are absent. Gypsum is ubiquitous as a secondary mineral (Fig. 8i), halite is rare, and glass often has alteration phases dominated by chlorine. Also present are extremely acidic vitrous grains, and grains composed of silica, aluminum, and magnesium in proportions equal to cordierite. Particles of lm to sublm size often adhere to larger particles, and aggregates are common. 4.2.1.4. Sample 180307a. The majority of clasts in the 18 March sample are glassy, moderately vesicular and have sharp edges and slight alteration (Fig. 8j, and k). Crystals of plagioclase and pyroxene are common. Of the secondary minerals, halite is diffuse, and gypsum is present but to a much lesser extent. Micron and sub-micron secondary minerals composed of chlorine and sulfur also exist, as well as smaller particles that adhere to these (e.g. Fig. 8l, elongate crystal containing sulfur and magnesium). 4.2.1.5. Sample 190307a. The 19 March sample consists predominantly of non-vesicular micro-crystalline and glassy clasts with sharp edges and little superficial alteration (e.g. Fig. 8m–o), and single crystals. Aggregates are present, but rare. Recycled clasts are volumetrically subordinate and particularly altered, with a glass composition generally rich in silica. Secondary minerals predominantly of gypsum exist, but rarely; halite is absent. 4.2.1.6. Sample 100407. The sample consists of mostly aggregates altered with gypsum (Fig. 8p, and q) and slightly altered blocky particles (Fig. 8r) that vary from glassy to micro-crystalline. Fluidal and angular particles are rare to absent, such that the sample is inferred to be predominantly lithic. 4.2.1.7. Sample 290407. This sample is composed of nearly all ash aggregates (Fig. 8s) consisting of variably altered blocky to fluidal particles, generally with rounded edges (Fig. 8t, and u). Gypsum is common as an alteration phase. 4.2.2. Grain size and morphoscopy Morphology data on the parameters of Heywood diameter, compactness, and elongation are presented in Fig. 9 – for all samples except 290407. The majority of particles in the 29 April sample are aggregated material (e.g. Fig. 8s), which readily flattened or fell apart when mounting, such that the automated morphoscopic analysis was not performed. The grain-size data show much variation among the samples, while the parameters of compactness and elongation are similar for all samples. Most samples have particles with diameters ranging between a few and 200 microns. Interestingly the 15 March sample, representing the paroxysm, has the smallest mode diameter at 25 lm, and is the best sorted. Sample 180307 has the oddest distribution, with clast diameters ranging between a few and over 350 microns, and no dominant mode. The samples collected over the longer 48–72 h period (280207,

121

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

b

c

d

e

f

g

h

i

j

k

l

m

n

o

p

q

r

s

t

u

290407

100407

190307a

180307a

150307

030307

280207

a

Fig. 8. Select SEM images of each sample from Stromboli 2007 eruption. Scale bar at top left of each image is 10 lm. Ash in images a, d–h, j–k,m–o, t–u were classified as juvenile, b,p,q,s, as lithic, and images c, i, l show alteration phases. A description of each image is given in Section 4.2.1 of the manuscript.

030307) have among the broadest distributions, and the more distal 280207 is better sorted than the more proximal 030307

(as expected). However it is somewhat surprising that 190307, collected the furthest from source is relatively poorly sorted,

122

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

60

% of Clasts

50 40 30 20 10 0 0

50

100

150

200

250

300

350

Stromboli samples. Also plotted is the composition of Stromboli melt and phenocrysts (from Bertagnini et al., 2003; Francalanci et al., 2004). Plots (a and b) show that all samples have values typical for Stromboli melt, with the influence of crystals, especially plagioclase, and that samples 2900407 and to a slight extent 150307 are elevated in MgO. The samples appear quite distinct on plots (c and d), showing that the samples are variably altered in sulfur and sodium. Samples 150307, 290407, and 100407 show high SO3, and samples 280207, 180307, and to some extent 030307 have elevated Na2O. Sample 190307 shows the least alteration of these components. These alteration patterns are consistent with those observed above.

Diameter (µm) 5. Interpretation and discussion

30

% of Clasts

25 20 15

5.1. The Etna case

280207 030307 150307 180307 190307 100407

10 5 0

0 0.2 more irregular

0.4

0.6

0.8

Compactness

1 square

25

% of Clasts

20 15 10 5 0 1 1.5 less elongate

2

2.5

Elongation

3

3.5 4 more elongate

Fig. 9. Results of morphological analysis for Stromboli samples. Legend is the same for all plots.

and that sample 100407, rich in aggregates, has the second tightest distribution. Considering the morphoscopy parameters of compactness and elongation, we note that sample 180307, with the widest grain size distribution, shows slightly anomalous shape distributions, and that sample 190307 shows the least variability in compactness but the most variability in elongation. In general, the distributions in all cases are similar, with samples having a compactness parameter that peaks at 0.6 and an elongation parameter that peaks between 1.7 and 2. 4.2.3. Surface chemistry Results of the surface chemistry analyses are shown as ternary plots of FeO–SiO2–Al2O3, FeO–Al2O3–MgO, Cl–SO3–Na2O, and FeO– SO3–Na2O in Fig. 10(a)–(d). Approximately 50 of 120 randomly selected EDS grains per sample are plotted (to prevent saturation of points on the plots). Only plot (c) is of different composition than shown in Fig. 6 for the Etna study, in order to highlight alteration of Cl (possibly from sea water) as distinct from SO3 among the

The Etna case study shows the results of componentry, morphology, and surface chemistry analyses on eight ash samples from the 24 November 2006 explosive event, collected at variable times and distances from source. Componentry analysis revealed that the majority of ash particles in all samples are lithics, and the surface chemistry data show that grains in all samples have the composition of Etna melt, influenced by crystallization and some sulfur alteration. The componentry data show an expected trend of increasing juvenile content (i.e. inferred decreasing particle density) with distance from source, which is matched by the morphoscopy data that indicates better sample sorting (from grain-size data) with distance. The morphological parameters of compactness and elongation show no clear trends among the samples. The high lithic content is likely related to an unsteady magma flux that allowed erosion or collapse of conduit walls (Andronico et al., 2009a; Keating et al., 2008). Given the model proposed by Taddeucci et al. (2004) wherein tachylite forms at cooler, more stagnant margins of a conduit and sideromelane in a hotter central streamline, the higher sideromelane content in samples 5, 6, and 8 (Fig. 4) may indicate a peak in the ejection of fresh magma soon after the eruption onset (given the sample collection time). It also may be related to the fact that sideromelane is the least dense and most irregularly shaped particle – and therefore can be expected to travel to these more distal locations (due to lower settling velocity). The high content of lithics and crystals in sample 22 probably is due to fractionation of this denser material close to source, with the additional possibility that a higher lithic content was emitted during weaker explosive pulses such that fewer lithics reached distal locations. The abundance of recycled material in sample 10 and not in sample 25 argue against the interpretation that more recycled material was emitted in the final hours of the event (i.e. as winds shifted to the S). In general, we consider that the relative uniformity among samples in the parameters of compactness and elongation (Fig. 5) to be due to the dominance of lithic particles (similar componentry) in all samples. For example, sideromelane is expected to provide the least compact and most elongate particles, however that is not seen here in those samples with more sideromelane (samples 5 and 6) probably because the difference in sideromelane content among all samples is at most 15% (Fig. 4). In the future, the morphoscopic parameters for the different components will be assessed by applying the analysis to separated sets of grains. In this case, we see size sorting and changes in componentry that can be linked with distance from source, but no clear shifts in shape sorting. Notable in the chemistry data is that sample 20 has a consistent trend towards plagioclase, and sample 8 (the most distal) a trend towards gypsum (Fig. 6). The latter point may indicate that alter-

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

(a)

(b)

(c)

(d)

123

Fig. 10. Ternary plots of Stromboli surface chemistry data, with typical values for Stromboli melt (M) and phenocrysts olivine (Ol), clinopyroxene (Cpx), plagioclase (Plag), oxide (Ox) gypsum (Gyp), and halite (Ha); from Bertagnini et al., 2003; Francalanci et al., 2004). Plots (a and b) show in general ash has chemistry of Stromboli glass with some influence of crystals. Plots (c and d) show variable alteration of the samples.

ation increases with distance from source, however given the elevated sulfur content of all samples, including the most proximal sample 22, there seems to be multiple origins for the sulfur alteration. 5.2. The Stromboli case The analyzed 2007 samples are highly variable, particularly in the degree of particle fluidity, groundmass crystallinity, and degree/type of surface alteration. We infer that samples with abundant halite (280207 and 180307 and in part 030307) originate at the sea entry. These three samples also have the most abundant clasts with a micro-crystalline groundmass, with the microcrystallinity likely to form during cooling in a lava flow. Sample 030307 is interpreted to contain a mixture of clasts produced at the summit craters (glassy, fluid, and vesicular) and at the ocean entry (microcrystalline, blocky). Sample 150307 is composed almost solely of glassy clasts interpreted to be sourced from the paroxysm. The absence of halite in sample 190307 likely precludes a ‘sea-entry’

origin, while the sharp edges and lack of surface alteration suggest a recent eruption of relatively fresh magma from the summit craters. Samples 100407 and 290407 were collected after the cessation of the lava flow and therefore must be sourced from summit explosions. The abundance of aggregates and recycled clasts in these samples suggest lack of fresh magma in the upper conduit system by this time. Some observations that warrant further comment include the variability in size and sorting, the presence or absence of aggregates, and the variable amount of sulfur alteration among the samples. The abundance of recycled material erupted with time will also be commented on. The grain-size data show that sample 150307 is by far the best sorted of the samples, and has among the smallest modal diameter at about 25 lm. The small modal particle size and uniform sample size may be an effect of the relatively more distal sampling site, but are also likely related to the higher degree of fragmentation and longer transport times during the paroxysm relative to other activity, in addition to strong winds that occurred

124

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

during the paroxysm. Sample 180307 has by far the most unusual distribution, being essentially unsorted. Primarily given the abundance of halite in this sample we infer the ash source to be the sea entry, where fragmentation dynamics can fluctuate on a seconds to minute basis as largely dependent on wave action. Also note that the sea entry activity generated abundant steam but no significant explosions, implying that the correspondent ash was mainly formed by thermal granulation of the lava or was simply resuspended material. Gypsum occurs in all the samples, however it is most abundant in samples 150307, 100407, and 290407. First considering sample 150307 from the paroxysm, gypsum is ubiquitous as an alteration phase also on grains that have the composition of Stromboli’s golden pumice and therefore must be newly erupted. This implies that such secondary crystallization can occur in matter of seconds, in the plume, between eruption and deposition. In the case of sample 150307, sulfate alteration may be prevalent due to increased sulfur content associated with the paroxysmal events at Stromboli (Aiuppa et al., 2009; Burton et al., 2009). The abundance of gypsum in samples 100407 and 290407 may occur for different reasons. Given that the majority of clasts in both of these samples are recycled, it is possible that the alteration occurred over time, with gypsum precipitated during passive degassing on the ash particles that were subsequently erupted. However, the abundance of aggregates that also occurs in these samples may suggest the presence of a high amount of gas/vapor during eruption (since humidity would tend to bind particles together), which could also contribute to the sulfur alteration. Along the same lines, it is somewhat surprising that aggregates are absent from samples 280207 and 180307a, both with an origin at the ocean entry, given the abundance of vapor expected to be present in this environment. In this case, the large size of particles might have been unfavorable for the formation of aggregates. Of interest is the amount of recycled material with time. Note that recycled ash is present but not common in the samples collected before (280207, 030307) and just after (180307a, 190307a) the paroxysm, while it is ubiquitous in the samples collected weeks later (100407, 290407; Table 1). Also consider that effusion of lava was more or less continuous between 27 February and 2 April, but had ceased by the time the latter samples were collected. Combined, these observations suggest that the influx of fresh magma was high and Stromboli’s shallow conduit system relatively open in the weeks before and days after the paroxysm and in concomitance with the effusive activity, but had become relatively closed and plugged with recycled material within about a month (i.e. by 10 April). 5.3. Comparison between the Etna and Stromboli ash particles Here we compare the automated morphoscopic and surface chemistry data presented for the two volcanoes, considering that all studied samples originated from relatively weak basaltic explosive activity but acknowledging that the Etna samples are from a well-constrained event and Stromboli samples from a variety of complex activity. In general, we note that particles originated by the different eruptive processes show similar morphological features but distinct surface chemistry differences. Figs. 5 and 9 show the morphoscopic parameters for Etna and Stromboli, respectively. Modes in the SEM-derived particle size distributions mirror well differences in the eruption intensity, with values between 25–75 microns 1–3 km from the vent (excluding sample 180307) for Stromboli and between 60 and 200 microns 6–24 km away from the vent for the much stronger Etna eruption. Also the range of the distributions (i.e. particle sorting) gives results that are consistent with their emplacement dynamics. Compactness factor is similar in the two cases (modes 0.6 and

between 0.6 and 0.7 for Stromboli and Etna, respectively), as well as the elongation parameter, with all samples showing relatively uniform distributions that peak between 1.7 and 2. This similarity is in agreement with visual observations that most of the investigated particles are rather blocky in shape (Figs. 3 and 8). In our case sample variability is larger than analytical error (Fig. 5), but by using a higher magnification and/or a better resolution for image acquisition even slighter differences could be highlighted. In general, and despite the abundance of lithic particles in the Etna case and secondary alteration in the Stromboli case (both which are expected to be associated with a blocky particle morphology), we propose that the dominant morphological control is a similar, probably brittle, fragmentation process. The surface chemistry data (Figs. 6 and 10) indicate clearly the basaltic melt composition of both magmas, also suggesting some differences in the abundance of crystals between samples (e.g. a lower content of plagioclase and a higher abundance of oxide crystals in the Etna samples as compared to Stromboli). Even more important is the effectiveness of surface chemistry in highlighting the differences in particle alteration. In our study cases, surface chemistry effectively distinguished slight differences in sulfur alteration among the Etna samples as well as the dominant phases of alteration (gypsum vs. halite) and their abundances among the Stromboli samples.

6. Conclusions and future work Overall, the semi-automated methods used in this study provide relatively fast and useful quantitative data on a large number of ash particles, which, especially when used in tandem with other types of observations, enables great insight into the volcanic process of ash production. Of the morphological parameters, Heywood diameter proved effective in delivering robust estimates of the grain size distribution of ash samples. The shape parameters (compactness and elongation) point toward a general uniformity in the morphology of the investigated ash samples, even if subtle differences are present, perhaps suggesting a similar fragmentation process for most of our samples. The surface chemistry data for both case studies were useful in highlighting bulk sample composition plus the degree of alteration and presence/influence of magmatic crystals. In particular, particle alteration, as evidenced by surface chemistry, is a good indicator of the source of ash. Even if surface chemistry may be the main factor determining the color of ash grains, such differences often go undetected using binocular microscope analyses, while more traditional techniques for the study of ash alteration (e.g. leachate X-ray diffractometric analyses) will not provide information on glass composition or crystal abundance. Simultaneous information on magma chemistry, crystal abundance, and surface alteration suggest that the proposed technique for the characterization of the surface chemistry of ash particles may be an efficient way to identify unknown ash sources and to perform long-range tephrostratigraphic correlations. Surface chemistry analysis also may be relevant for environmental hazard mitigation by detecting potentially dangerous pollutants on ash particles. The two case studies shown above illustrate well the diversity of information stored in volcanic ash:  subtle differences in ash deposited at various locations were revealed from even a small event at Etna, thus outlining relatively small-scale plume zonations;  different sources of ash during the 2007 crisis at Stromboli have distinctive textures and alteration signatures, and can be distinguished from ash erupted during typical Strombolian activity;

125

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

 FE-SEM makes obtaining quantitative morphological and chemical data for ash more relatively straightforward. The morphological and visual similarity between most of the analyzed ash particles points out a more general problem, which is the applicability of a clear-cut distinction in component classes of ash particles from weak volcanic activity. In particular, when activity is weak, frequent, and long-lasting, the basic distinction between juvenile and lithic components tends to fade, as true wall-rock lithic clasts, linked to much earlier eruptions, may be minor in abundance or absent, and the ejecta can range from truly juvenile clasts, quenched at fragmentation, to material that was quenched minutes to weeks prior to its explosive ejection. This picture gets even more complicated by particle recycling through multiple explosions (Houghton and Smith, 1993), in-conduit crystallization, and the possibility that ash formed at a sea entry. The three processes have been reported at basaltic volcanoes (Houghton and Gonnermann, 2008), almost ubiquitously accompanying more conventional Strombolian or Hawaiian eruptive activity,

and often marking important changes in the eruptions and heralding new waning or waxing phases (Taddeucci et al., 2002). Even if they do not represent a major volumetric source of ash, it is thus important, from a volcano monitoring perspective, to be able to track and interpret correctly the products of such ‘‘ancillary’’ activities. Componentry data for the Etna case study proved useful and enabled more confident volcanological interpretations to be made, but the Stromboli case clearly points out the need to establish a new componentry scheme for such activity. In the near future we aim to obtain and analyze more samples from these volcanoes, analyze samples of individual components, develop a comprehensive classification scheme for products at Stromboli, and identify whether differences in ash are discernable during more subtle variations in typical activity at Stromboli. In the longer term, we will extend this type of textural/chemical analysis of ash to other case studies, including more silicic volcanoes and experimental ash samples formed under controlled fragmentation and surface alteration conditions.

Table A1 Chemical analysis of basaltic glass standard.

a b c d

Oxide

Jarosewich et al., 1977bb

EPMAc

EDS 15 kV

1 rd

EDS 10 kV

1rd

SiO2 TiO2 Al2O3 FeOa MnO MgO CaO Na2O K2O P2O5 Total

50.81 1.85 14.06 12.06 0.22 6.71 11.12 2.62 0.19 0.20 99.84

51.00 1.39 13.31 12.06 0.22 6.80 10.70 2.61 0.21 0.23 98.53

51.59 2.04 13.09 11.75 0.61 7.02 10.70 2.86 0.16 0.18 100.00

0.76 0.22 0.27 0.32 0.18 0.19 0.24 0.16 0.04 0.24

51.60 1.89 13.40 11.80 0.25 6.64 11.40 2.48 0.22 0.25 100.00

0.34 0.30 0.17 0.33 0.59 0.26 0.69 0.09 0.16 0.20

Total iron. See referenced paper for error description. Single analysis. Standard deviation over 5 analyses.

Fig. A1. FE-SEM images of the same particle (from sample Etna 22) after four 90° rotations. The purple rectangle marks the scanning area used to acquire the X-ray spectra, which was converted in the analyses shown in Table A2 (particle 4). (For color interpretation in this figure legend the reader is referred to see the web version of this article.)

126

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127

Table A2 Repeated surface analyses after particle rotation. Oxide SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O P2O5 SO3 Cl Total

1 44.79 1.66 15.83 14.42 0.57 2.55 8.59 5.99 2.62 1.26 1.60 0.12 100

42.51 1.95 14.66 16.38 0.68 3.63 10.62 4.96 2.25 0.95 1.09 0.32 100

2 39.01 2.02 13.00 20.46 1.26 4.44 10.96 3.59 2.66 0.81 1.79 0.00 100

44.07 2.47 14.73 14.68 0.47 3.72 10.30 5.06 2.56 0.29 1.58 0.08 100

44.19 1.89 14.45 14.43 0.55 3.80 10.63 5.12 3.02 0.94 0.93 0.05 100

44.84 2.00 14.58 13.94 0.06 4.01 10.87 4.79 3.09 0.48 1.28 0.05 100

3 43.75 1.92 15.68 15.49 0.17 3.24 10.21 4.55 3.08 1.19 0.73 0.00 100

43.04 1.89 14.94 15.23 0.32 3.72 10.31 5.10 3.23 1.21 0.78 0.22 100

36.81 3.36 12.07 24.32 0.98 3.23 9.59 4.32 2.87 0.71 1.74 0.00 100

Acknowledgements NL thanks generous support from the National Science Foundation International Postdoctoral Research Program, grant OISE0754423. Funding was partially provided by FIRB-MIUR Project ‘‘Research and Development of New Technologies for Protection and Defense of Territory from Natural Risks’’. Authors are grateful to Antonio Cristaldi for having shared preliminary studies of the 24 November 2006 event. We also thank multiple colleagues of the Italy’s National Civil Defense that some way facilitated collection of ash samples during Stromboli’s 2007 crisis. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. Appendix A. Analytical error in the surface chemistry data To quantify the analytical error for the surface chemistry data we first must determine the EDS system error. Table A1 reports a comparison between (a) published chemical analyses of a thin-sectioned basaltic glass standard (vg2 glass of Jarosewich et al., 1977a, b), and (b) the analyses performed in our laboratory with the FESEM EDS standardless system at the following conditions: 10 and 15 kV acceleration voltages, beam current of 0.85 and 55 nA, live time 30 s, and 10 mm working distance. For comparison, the table also shows an analysis of the same standard performed with a JeolJXA8200 EDS-WDS Electron Microprobe at 15 kV accelerating voltage and 10 nA beam current (see Del Gaudio et al., 2010, for further analytical details). EDS data are the average of 5 analyses. Particle surface roughness is another source of error in the surface chemistry data. This roughness causes the X-rays generated by the beam-particle interaction to reach the detector under variable geometries, mostly different from the ideal case. To quantify this effect we performed specific tests by measuring the same particle four times, each time rotating the particle by 90° and thus changing the particle surface-detector geometry (Fig. A1). Table A2 illustrates the results for five different particles. These data are used to calculate the error bars in Figs. 6 and 10 for each ternary plot. Each error bar represents the largest 2r standard deviation over the five test particles investigated, for the relative abundance of the three elements in each plot. References Aiuppa, A., Federico, C., Giudice, G., Giuffrida, G., Guida, R., Gurrieri, S., Liuzzo, M., Moretti, R., Papale, P., 2009. The 2007 eruption of Stromboli volcano; insights from real-time measurement of the volcanic gas plume CO2/SO2 ratio. Journal of Volcanology and Geothermal Research 182 (3–4), 221–230. Alparone, S., Andronico, D., Lodato, L., Sgroi, T., 2003. Relationship between tremor and volcanic activity during the southeast crater eruption on Mount Etna in early 2000J. Geophysical Research 108 (B5), 13.

39.64 2.89 13.12 21.54 0.93 3.90 9.09 4.03 2.68 0.66 1.48 0.04 100

4 41.53 2.89 14.05 18.17 0.74 3.95 9.47 4.7 2.84 0.95 0.63 0.10 100

39.62 2.31 14.46 20.91 1.06 3.45 9.36 4.66 2.56 0.67 0.91 0.03 100

45.14 1.58 18.67 12.46 0.65 2.50 9.15 4.88 2.67 0.86 1.26 0.17 100

43.85 1.99 16.41 13.62 0.69 2.91 10.45 5.06 2.64 1.15 1.00 0.24 100

5 43.59 1.70 15.75 14.05 0.40 3.01 10.89 4.99 2.98 1.15 1.38 0.12 100

43.43 1.93 16.80 13.77 0.64 2.62 10.46 5.44 2.67 1.13 1.04 0.08 100

47.85 1.79 15.93 11.45 0.31 3.31 9.25 5.12 3.49 1.16 0.35 0.00 100

48.13 1.53 16.24 10.46 0.50 3.56 9.00 5.69 2.80 1.25 0.60 0.24 100

45.64 2.39 16.37 12.15 0.64 3.57 9.42 5.13 2.66 1.05 0.84 0.14 100

45.53 2.03 15.75 13.26 0.00 2.79 9.45 4.92 3.50 1.47 1.14 0.17 100

Alparone, S., Andronico, D., Sgroi, T., Ferrari, F., Lodato, L., Reitano, D., 2007. Alert system to mitigate tephra fallout hazards at Mt. Etna Volcano, Italy. Natural Hazards 43 (3), 333–350. Andronico, D., Caruso, S., Corsaro, R., Cristaldi, A., Del Carlo, P., Miraglia, L., 2005. Caratterizzazione dei prodotti esplosivi emessi dallo Stromboli nel periodo luglio-dicembre 2004. INGV Report No. UFVG2005/52. Andronico, D., Cristaldi, A., Lo Castro, D., Scollo, S., Taddeucci, J., 2007. The 24 November 2006 paroxysm at south-east crater, Mt. Etna. In: Proceedings of the XXIV General Assembly of the IUGG. Andronico, D., Scollo, S., Cristaldi, A., Ferrari, F., 2009a. Monitoring ash emission episodes at Mt. Etna: the 16 November 2006 case study. Journal of Volcanology and Geothermal Research 180 (2–4), 123–134. doi:10.1016/ j.jvolgeores.2008.10.019. Andronico, D., Spinetti, C., Cristaldi, A., Buongiorno, M.F., 2009b. Observations of Mt. Etna volcanic ash plumes in 2006: an integrated approach from ground-based and polar satellite NOAA-AVHRR monitoring system. Journal of Volcanology and Geothermal Research 180 (2–4), 135–147. doi:10.1016/ j.jvolgeores.2008.11.013. Andronico, D., Cristaldi, A., Del Carlo, P., Taddeucci, J., 2009c. Shifting styles of basaltic explosive activity during the 2002–03 eruption of Mt. Etna, Italy. Journal of Volcanology and Geothermal Research 180 (2-4), 110–122. Barberi, F., Rosi, M., Sodi, A., 1993. Volcanic hazard assessment at Stromboli based on review of historical data. Acta Vulcanologica 3, 173–187. Barberi, F., Civetta, L., Rosi, M., Scandone, R., 2009. Chronology of the 2007 eruption of Stromboli and the activity of the Scientific Synthesis Group. Journal of Volcanology and Geothermal Research 182 (3–4), 123–130. Bertagnini, A., Métrich, N., Landi, P., Rosi, M., 2003. Stromboli volcano (Aeolian Archipelago, Italy); an open window on the deep-feeding system of a steady state basaltic volcano. Journal of Geophysical Research 108 (B7), 15. Calvari, S., Lodato, L., Steffke, A., Cristaldi, A., Harris, A.J.L., Spampinato, L., Boschi, E., 2010. The 2007 Stromboli eruption: event chronology and effusion rates using thermal infrared data. Journal Geophysical Research. Solid Earth 115 (B4, B04201). doi:10.1029/2009JB006478. Cas, R.A.F., Wright, J.V., 1987. Volcanic Successions Modern and Ancient. Allen and Unwin, London. 528 pp. Cashman, K.V., Hoblitt, R.P., 2004. Magmatic precursors to the 18 May 1980 eruption of Mount St. Helens, USA. Geology 32 (2), 141–144. Chouet, B., Hamisevicz, N., McGetchin, T., 1974. Photoballistics of volcanic jet activity at Stromboli, Italy. Journal of Geophysical Research 79 (32), 4961–4976. Cioni, R., D’Oriano, C., Bertagnini, A., 2008. Fingerprinting ash deposits of small scale eruptions by their physical and textural features. Journal of Volcanology and Geothermal Research 177, 277–287. Coltelli, M., Pompilio, M., Del Carlo, P., Calvari, S., Pannucci, S., Scribano, V., 1998. Mt. Etna – 1993–95 Eruptive activity. Acta Vulcanologica 1. Coltelli, M., Del Carlo, P., Pompilio, M., 2000. Etna: Eruptive activity in 1996. Acta Vulcanologica 12 (1), 63–67. Corsaro, R.A., Pompilio, M., 2004. Magma dynamics in the shallow plumbing system of Mt. Etna as recorded by compositional variations in volcanic of recent summit activity (1995–1999). Journal of Volcanology and Geothermal Research 137, 55–71. Corsaro, R.A., Miraglia, L., Zanon, V., 2005. Petrologic monitoring of glasses in the pyroclastites erupted in February 2004 by the Stromboli Volcano, Aeolian Islands, Southern Italy. Journal of Volcanology and Geothermal Research 139, 339–443. De Rosa, R., 1999. Compositional modes in the ash fraction of some modern pyroclastic deposits: their determination and significance. Bulletin of Volcanology 61, 162–173. doi:10.1007/s004450050269. Del Gaudio, P., Mollo, A., Ventura, G., Iezzi, G., Taddeucci, J., Cavallo, A., 2010. Cooling rate-induced differentiation in anhydrous and hydrous basalts at 500 MPa: Implications for the storage and transport of magmas in dikes. Chemical Geology 270, 164–178. Dellino, P., La Volpe, L., 1996. Cluster analysis on ash particles morphology features to discriminate fragmentation dynamics in explosive eruptions. Acta Vulcanologica 8 (1), 31–39.

N.C. Lautze et al. / Physics and Chemistry of the Earth 45–46 (2012) 113–127 Dellino, P., Liotino, G., 2002. The fractal and multifractal dimension of volcanic ash particles contour: a test study on the utility and volcanological relevance. Journal of Volcanology and Geothermal Research 113 (2), 1–18. Delmelle, P., Lambert, M., Dufrene, Y., Gerin, P., Oskarsson, N., 2007. Gas/aerosol-ash interaction in volcanic plumes; new insights from surface analyses of fine ash particles. Earth and Planetary Science Letters 259 (1–2), 159–170. Ersoy, O., 2010. Surface area and volume measurements of volcanic ash particles by SEM stereoscopic imaging. Journal of Volcanology and Geothermal Research 190, 290–296. Ersoy, O., Gourgaud, A., Aydar, E., Chinga, G., Thouret, J-C., 2007. Quantitative scanning-electron microscope analysis of volcanic ash surfaces: application to the 1982–1983 Galunggung eruption (Indonesia). GSA Bulletin 119 (5/6), 743– 752. doi:10.1130/B26048.1. Fisher, R.V., Schmincke, H.U., 1984. Pyroclastic Rocks. Springer-Verlag, Berlin, Heidelberg, New York, Tokyo. 472 pp. Francalanci, L., Tommasini, S., Conticelli, S., 2004. The volcanic activity of Stromboli in the 1906–1998 AD period: mineralogical, geochemical and isotope data relevant to the understanding of the plumbing system. Journal of Volcanology and Geothermal Research 131, 179–211. Heiken, G., Wohletz, K., 1985. Volcanic Ash. University of California Press, Berkeley. 246 pp. Horwell, C.J., Fenoglio, I., Ragnarsdottir, K.V., Sparks, R.S.J., Fubini, B., 2003. Surface reactivity of volcanic ash from the eruption of Soufrière Hills volcano, Montserrat, with implications for health hazards. Environmental Research 93, 202–215. Houghton, B.F., Gonnermann, H.M., 2008. Basaltic explosive volcanism: constraints from deposits and models. Chemie der Erde 68 (118), 117–140. Houghton, B.F., Smith, R.T., 1993. Recycling of magmatic clasts during explosive eruptions: estimating the true juvenile content of phreatomagmatic volcanic deposits. Bulletin of Volcanology 55, 414–420. Jarosewich, E., Nelen, J.A., Norberg, J.A., 1977a. Electron microprobe reference samples for mineral analyses. In: Fudali, R.F. (Ed.), Mineral Sciences Investigations 1976–1977, Smithsonian Contributions to the Earth Sciences, vol. 22, pp. 68–72. Jarosewich, E., Parkes, A.S., Wiggins, L.B., 1977b. Microprobe analyses of four natural glasses and one mineral: an interlaboratory study of precision and accuracy. In: Fudali, R.F. (Ed.), Mineral Sciences Investigations 1976–1977, Smithsonian Contributions to the Earth Sciences, vol. 22, pp. 53–67. Keating, G., Pelletier, J., Valentine, G., Statham, W., 2008. Evaluating suitability of a tephra dispersal model as part of a risk assessment framework. Journal of Volcanology and Geothermal Research 177 (2), 397–404. Kueppers, U., Perugini, D., Dingwell, D.B., 2006. Explosive energy during volcanic eruptions from fractal analysis of pyroclasts. Earth and Planetary Science Letters 248, 800–807. Maria, A., Carey, S., 2002. Using fractal analysis to quantitatively characterize the shape of volcanic particles. Journal of Geophysical Research 107 (B11), 2283. doi:10.1029/2001JB000822. Martin, R.S., Mather, T.A., Pyle, D.M., Power, M., Allen, A.G., Aiuppa, A., Horwell, C.J., Ward, E.P.W., 2008. Composition-resolved size distributions of volcanic aerosols in the Mt. Etna plumes. Journal of Geophysical Research 113, D17211. doi:10.1029/2007JD009648. Métrich, N., Clocchiatti, R., Mosbah, M., Chaussidon, M., 1993. The 1989–1990 activity of Etna magma mingling and ascent of H (sub 2) O–Cl–S-rich basaltic magma; evidence from melt inclusions. Journal of Volcanology and Geothermal Research 59 (1–2), 131–144.

127

Métrich, N., Bertagnini, A., Landi, P., Rosi, M., Belhadj, O., 2005. Triggering mechanism at the origin of paroxysms at Stromboli (Aeolian Archipelago, Italy); the 5 April 2003 eruption. Geophysical Research Letters 32 (10), 4. Métrich, N., Bertagnini, A., Di Muro, A., 2010. Conditions of magma storage, degassing and ascent at Stromboli: new insights into the volcano plumbing system with inferences on the eruptive dynamics. Journal of Petrology 51, 603– 626. doi:10.1093/petrology/egp083. Mori, T., Burton, M., 2009. Quantification of the gas mass emitted during single explosions on Stromboli with the SO2 imaging camera. Journal of Volcanology and Geothermal Research 188 (4), 395–400. Patrick, M., Harris, A.J.L., Ripepe, M., Dehn, J., Rothery, D.A., Calvari, S., 2007. Strombolian explosive styles and source conditions: Insights from thermal (FLIR) video. Bulletin of Volcanology, 69. doi:10.1007/s00445-006-0107-0. Riley, C.M., Rose, W.I., Bluth, G.J.S., 2003. Quantitative shape measurements of distal volcanic ash. Journal of Geophysical Research 108 (B10), 2504. doi:10.1029/ 2001JB000818. Sagnotti, L., Taddeucci, J., Winkler, A., Cavallo, A., 2009. Compositional, morphological, and hysteresis characterization of magnetic airborne particulate matter in Rome, Italy. Geochemistry Geophysics Geosystems 10, Q08Z06. doi:10.1029/2009GC002563. Schiavi, F., Kobayashi, K., Moriguti, T., Nakamura, E., Pompilio, M., Tiepolo, M., Vannucci, R., 2009. Degassing, crystallization and eruption dynamics at Stromboli: trace element and lithium isotopic evidence from 2003 ashes. Contributions to Mineralogy and Petrology. doi:10.1007/s00410-009-0441-2. Sheridan, M.F., Marshall, J.R., 1983. Interpretation of pyroclast surface features using sem images. Journal of Volcanology and Geothermal Research 16, 153– 159. Simkin, T., Siebert, L., 2000. Earth’s volcanoes and eruptions: an overview. In: Sigurdsson, H. (Ed.), Encyclopedia of Volcanoes. Elsevier, New York. Sulpizio, R., Bonasia, R., Dellino, P., Di Vito, M., La Volpe, L., Mele, D., Zanchetta, G., Sadori, L., 2008. Discriminating the long distance dispersal of fine ash from sustained columns or near ground ash clouds: The example of the Pomici di Avellino eruption (Somma-Vesuvius, Italy). Journal of Volcanology and Geothermal Research 177, 263–276. Taddeucci, J., Pompilio, M., Scarlato, P., 2002. Monitoring the explosive activity of the July–August 2001 eruption of Mt. Etna (Italy) by ash characterization. Geophysical Research Letters 29 (8), 1230. doi:10.1029/2001GL014372. Taddeucci, J., Pompilio, M., Scarlato, P., 2004. Conduit processes during the July– August 2001 explosive activity of Mt. Etna (Italy); inferences from glass chemistry and crystal size distribution of ash particles. Journal of Volcanology and Geothermal Research 137 (1–3), 33–54. Taddeucci, J., Scarlato, P., Andronico, D., Cristaldi, A., Buettner, R., Zimanowski, B., Kueppers, U., 2007. Eos, Transactions, American Geophysical Union 88 (24), 253–256. Watanabe, K., Danhara, T., Terai, K., Yamashita, T., 1999. Juvenile volcanic glass erupted before the appearance of the 1991 lava dome, Unzen volcano, Kyushu, Japan. Journal of Volcanology and Geothermal Research 89, 113–121. Wohletz, K.H., 1987. Chemical and textural surface features of pyroclasts from hydrovolcanic eruption sequences. In: Marshall, J.R. (Ed.), Clastic particles: scanning electron microscopy and shape analysis of sedimentary and volcanic particles. Van Nostrand Reinhold Company, New York, pp. 79–97.