0% 20% 40% 60% 80% 100%

ES T4

ES T3A

ES T1A

Glassy Microlitic Crystalline Lithics Crystal fragments

Exploring the mechanisms of �ne ash generation:methods of ash characterisation

Emma J. Liu, Katharine V. Cashman, & Alison C. RustDepartment of Earth Sciences, University of Bristol, Wills Memorial Building, Bristol, BS8 1RJ

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

Axi

al ra

tio

500µm

Axi

al ra

tio

Concavity index, CI

ElongateEquant

Blocky Bubblydcb

a

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

e

50 µm

Brittle: angular fractured interior

Ductile: fluid outer surface

Introduction

The disruption during the recent Icelandic eruptions of Eyjafjallajokull (2010) and Grimsvotn (2011) was caused by the large volume of fine ash produced during the early phreatomagmatic phases of both eruptions. This highlighted gaps in our knowledge regarding the fragmentation mechanisms that generate volcanic ash in basaltic systems, particularly the role of magma-water interactions in controlling the size and shape of the resulting pyroclasts.

By quantifying the variation in morphology and componentry with grain size, and comparing this relationship to the corresponding vesicle textures, we explore the potential role of bubbles in phreatomagmatic fragmentation.

Making Measurements

2011 Grimsvotn eruption, Iceland

�ermal Stresses in Phreatomagmatic Fragmentation Tephra Correlations - Deception Island, Antarctica

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100

40

30

Bubble diameter/Grain size, L (µm)

N (%

) > L

Proportion dense fragments (%

)

70

80

90

50

60

G3G1

G8G7G6G4

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.2 0.4 0.6 0.8 1

Conv

exity

Solidity

Eyja_dense crystalline

Eyja_ dense glassy

G_dense glassy

Eyja_shard glassy

G_shard glassy

Eyja_vesicular glassy

G_vesicular glassy

Eyja_vesicular microcrystalline

Eyja_vesicular crystalline

glassy

crystalline

densebubbly

crys

tallin

ity

Myrdalssandur

Kirkjubaejarklaustur

Skaftafell

Grimsvotneruption

64’N

63’30’N

19’W 17’W

G1

G7G8

G3

G6

G4

N

0

2

4

6

8

10

12

14

16

18

0.05 0.

10.

15 0.2

0.25 0.

30.

35 0.4

0.45 0.

50.

55 0.6

0.65 0.

70.

75 0.8

0.85 0.

90.

95 1

Num

ber %

Fig. 2: Comparison of results for ‘Form Factor’ obtained by backscattered electron BSE-SEM (cross-sectional), secondary electron SE-SEM (projected), and Morphologi optical particle analyser (projected).

In collaboration with the Holoantar paleoclimate group - Oliva, M., Giralt, S., Antoniades, D., & Geyer, A.

T1

T2

T3

Grimsvotn G1 (115 km)

Grimsvotn G6 (71 km)

Grimsvotn G8 (50 km)

Eyja�allajokull (21 km)

Eyja�allajokull (60 km)

0

10

20

30

40

50

60

70

80

90

100

-10-50510

Cum

ulati

ve m

ass %

Grain size (φ)

Grimsvotn (1) - 115 km

Grimsvotn (6) - 71 km

Grimsvotn (8) - 50 km

Eyja 21km

Eyja 60km

Typical Strombolian (proximal scoria cone depostis)

Phreatomagmatic(proximal fall deposits)

Kiluaea Iki fountain (TGSD)

Grainsize Distributions for Ma�c EruptionsMafic eruptions involving magma-water interactions typically produce finer grained deposits than equivalent dry eruptions.

The 2011 Grimsvotn eruption generated a large proportion of fine ash.

Fig. 1: Compilation of grainsize distributions for mafic eruptions (data from Houghton and Hackett, 1983; Houghton and Schminke, 1989; Houghton and Nairn, 1991; Nemeth and Cronin, 2011; Gudmundsson et al., 2012; Parfitt, Olsson et al. 2013)

We analysed ash samples collected during the first 24 hours of the 2011 eruption, from sites between 30 and 115 km from the vent. Samples were sieved and mounted for SEM analysis as individual size fractions.

Grimsvotn ash is morphologically heterogeneous. Particles can be classified into two main components, which can be effectively discriminated using shape parameters sensitive to the ‘convex hull’

(1) dense fragments (blocky, fracture-bounded) (2) bubbly shards and vesicular grains (irregular, vesicle-bounded)

SEM Analysis

Automated Componentry

Linking external morphology to internal texture

We measured the distribution of 2D vesicle diameters from those grains where >50% of the bubble wall is preserved (solid black line).

We observe a systematic change in ash morphology (symbols) as the particle size approaches the modal bubble size (Fig. 9). Ash particles smaller than the mode of the bubble size distribution are dense, and represent the melt (glass) interstices between bubbles.

modal bubble size

b We have been exploring the role of residual stresses in phreatomagmatic fragmentation through the analogue of Prince Ruperts Drops (PRDs) - these quenched glass drops undergo explosive self-sustained fragmentation when broken and residual stresses are released.

Fragment morphology is spatially related to pre-existing bubbles in the glass, and share many of the shape and surface characteristics seen in Grimsvotn volcanic ash.

Grimsvotn ash shows evidence of both brittle and ductile fragmentation, often in the same clast (Fig. 10).

Differential cooling of magma interacting with water will develop residual thermal stresses in quenched pyroclasts, which may contribute to explosive thermal granulation of vesicular pyroclasts and amplify fine ash production.

The stress field around vesicles would provide a local control on brittle fracture propagation. The shape and size of the resulting ash particles would therefore be controlled by the size and spatial distribution of bubbles in the quenched primary pyroclasts.

Fig.11: (a) Cross-section through fractured PRD, showing the influence of bubbles on fracture propagation; (b) residual stress field revealed by polarised filters.

Fig.6: Sampling locations for the Grimsvotn ash

Fig. 8: Componentry as a function of grainsize. Automated componentry shown as shaded bars. Manual results illustrated by dotted lines.

Shape parameter measurements provide a rapid, reproducible and unbiased approach to estimate the proportions of different components within a sample. We observe good agreement between manual and automated results.

Our results demonstrate that the relative proportion of dense fragments to bubbly grains increases with decreasing size fraction.

Componentry measurements based on a single size fraction are not sufficient to characterise a deposit.

a

Shape parameters provide numerical measures of particle shape. They are typically non-dimensional, formed from ratios of different measures of particle size (e.g. perimeter, feret diameter, convex hull).

Understanding how shape parameters translate to physical properties is key to their use as a tool in ash characterisation.

Resolution: critical pixel density >750 pixels per particleSample size: >1000-2000 particles statistically significant

Instrument: we find fundamental differences in the results obtained between cross-section and projected view techniques

Image aquisition

BSE

Morphologi

SE

Form Factor

Parameters sensitive to the convex hull (perimeter of an elastic band stretched around a particle) are very effective at discriminating based on vesicularity:

(a) Shards: particle-scale (morphological) concavities. Low solidity. (b) Vesicular grains: many small (textural) concavities. Low convexity. (c) Dense fragments: few concavities High convexity and solidity.

Parameters that compare to an equivalent-area reference shape (e.g. circle) and axial ratio are poor discriminators of vesicularity.

What about crystals?

A microlitic groundmass changes the bubble texture from smooth isolated spheres to irregular and convoluted. Interconnectivity also increases.

The effect of this is to significantly increase convexity(perimeter-based roughness), with relatively little increase in solidity (excess area). Scattered crystals in a glassy matrix are indistinguishable from crystal-free glass, however, as the crystals do not influence the bubble texture.

Shape Parameter Selection

Shape parameters are sensitive to different aspects of particle morphology.

We used cluster analysis to explore the relationships between commonly-used shape parameters (Fig. 3). The clustergram structure is independent of the morphology of the particles analysed.

0

50

100

150

Eucl

idea

n di

stan

ce

ELRD

Axial Ratio

DelA

AxlR ConvexityForm Factor

C

EXT

RTSolidity CP

PFCVX_F

DelA

Form (elongation)

Area-based roughness

Perimeter-based roughness

Fig. 4: Variation in (a) convex hull and (b) best-fit ellipse for shards, vesicular grains and dense fragments; example distributions of (c) solidity and (d) axial ratio, for subsample of Grimsvotn ash (4 φ size)

a b

c d

Fig. 9

Fig. 3

Fig. 7 Fig. 9

Fig. 10

Solidity Axial Ratio

Num

ber %

Grain size (φ)

Cum

ulati

ve m

ass %

T1

T3

T3

T2

T2

T2

T1

T1

T1

T3T1

T4T5

T3b

CH120301 CH120501 LIM08 ES120302 ES120301G CN120301G

Glassy Microlitic Crystalline Lithics Crystal fragments

FINE COARSE

Fig. 12 (left): Relative proportions of components of different crystallinity within tephra layers of ES120302. Fig. 13 (above): Tephra chronological framework for Byers Peninsula, Antarctica,

Red bars show syn-collapse sediment deformation.

Componentry and textural measurements can aid correlations when chemical compositions are similar (e.g., single source volcano)

Combined with paleoenvironmental evidence, tephra stratigraphy in lake cores from Byers Peninsula, Antarctica, has constrained the timing of the Deception Island caldera collapse to ~4000 ka.

We measure an abrupt change in crystallinity, morphology and composition between T3 and T1/T2 layers. T3 is chemically similar to proximal deposits of the first ignimbrite phase of the Deception Island caldera collapse unit.

0

10

20

30

40

50

60

70

80

90

100

16 32 63 91

Md Dst

Freq

uenc

y %

Grainsize (µm)

G8 G7 G6 G4 G3 G1