b.m. kennedy et al- time-and temperature-dependent conduit wall porosity: a key control on degassing...

8/3/2019 B.M. Kennedy et al- Time-and temperature-dependent conduit wall porosity: A key control on degassing and explos

1/12

Time-and temperature-dependent conduit wall porosity: A key control on degassingand explosivity at Tarawera volcano, New Zealand

B.M. Kennedy a,b,, A.M. Jellinek b, J.K. Russell b, A.R.L. Nichols c, N. Vigouroux d

a Geological Sciences, University of Canterbury, Private Bag 4800, Christchurch, 8140, New Zealandb Department of Earth and Ocean Sciences, University of British Columbia, 6339 Stores Road, Vancouver, BC, Canada V6T 1Z4c Institute for Research on Earth Evolution (IFREE), Japan Agency for Marine Earth Science and Technology (JAMSTEC), 2-15 Natsushima-cho, Yokosuka, Kanagawa 237-0061, Japand Department of Earth Science, Simon Fraser University, Burnaby, British Columbia, Canada

a b s t r a c ta r t i c l e i n f o

Article history:

Received 14 February 2010

Received in revised form 18 August 2010

Accepted 23 August 2010

Available online xxxx

Editor: R.W. Carlson

Keywords:

degassing

conduit

magma

volcano

explosive

lava dome

The permeability of volcanic conduit walls and overlying plug can govern the degassing and explosivity of

eruptions. At volcanoes characterized by a protracted history of episodic volcanism, conduit walls are

commonly constructed of quenched magma. During each successive eruptive phase, reheating by ascending

magma can modify the porosity, permeability and H2O content of the conduit wall rocks and overlying plug.

We investigate whether theunusual explosivity of the 1886 basaltic eruption at Tarawera volcanois related to

the heating and degassing of the AD1314 Kaharoa rhyolitic rocks, through which it erupted. We heat cores of

perlitic Tarawera dome rhyolite to 300 C1200 C for 30 min to 3 days at atmospheric pressure. We

characterize time (t)- and temperature (T)-dependent variations in porosity, volatile content and texture

through SEM image analyses. We also directly measure pre- and post-experimental connected and isolated

porosity and water content. We identify four textural/outgassing regimes: Regime 1 ( T800 C, t2 h), with

negligible textural changes and a significant loss of meteoric water (1.40.72 wt.% H2O); Regime 2

(800T1100 C, t6 h), with cracking and vesicle growth and a 510% increase in connected porosity;

Regime 3 (800T1200 C, t30 min), with healed cracks, coalesced and collapsed vesicles, and overall

reduced porosity; and Regime 4 (T1200 C, tN30 min), with a collapse of all connected porosity. These

regimes are governed by the temperature ofthe event (T) relative to the glass transition temperature (Tg) andthe time scale of the event (t) relative to a critical relaxation time for structural failure of the melt (r). We

identify a quantitative transition from predominantly brittle behavior such as cracking, which enhances

connected porosity and permeability, to viscous processes including crack healing and vesicle collapse, which

act to reduce connected porosity. Applied to the 1886 basalt eruption at Tarawera, we show that progressive

heat transfer ultimately reduced the open porosity and permeability of the conduit walls, thereby partially

sealing the conduitand reducing volatile loss. We argue that this mechanismwas an underlyingreasonfor the

exceptional explosivity of the 1886 eruption. We further suggest that textural changes associated with

reheating could explain some of the cyclic deformation and degassing observed at many lava domes

preceding explosive eruptions.

2010 Elsevier B.V. All rights reserved.

1. Introduction

Individual volcanic eruptions can shift rapidly in style from

relatively quiescent lava dome extrusion to explosive eruptions (e.g.

Sparks, 2003; Voight et al., 1999). These shifts are generally attributed

to variations in physical properties such as gas content, viscosity,

vesicularity, wall rock permeability, and crystallinity (e.g. Gonner-

mann and Manga, 2007; Jaupart, 1998; Melnik and Sparks, 2002). To

date experimental and theoretical studies have focussed on variation

of these properties as magma rises and decompresses(e.g. Baker et al.,

2006; Gardner, 2007; Hammer and Rutherford, 2002; Larsen et al.,

2004; Proussevitch et al., 1993; Takeuchi et al., 2009; Yoshimura and

Nakamura, 2008) or is sheared (Gonnerman and Manga, 2003;

Lavallee et al., 2007, 2008; Okamura et al., 2010; Smith et al., 2009;

Tuffen et al., 2003, 2008). Natural pumice, dome rocks and

experimentally decompressed glasses show huge textural variations

in their permeablevesicleand crack networks (Jaupart, 1998; Michaut

and Sparks, 2009; Mueller et al., 2008; Rust and Cashman, 2004; Saar

and Manga, 1999; Takeuchi et al., 2008; Westrich and Eichelberger,

1994; Wright et al., 2009; Yoshimura and Nakamura, in press). Other

experimental studies investigate the effects of temperature on water

speciation, solubility, and magma viscosity (Stolper, 1989; Yamashita,

1999; Zhang et al., 2007). Rocks from conduit walls also exhibit

variation in porosity (Kennedy et al., 2005; Rust et al., 2004; Stasiuk et

Earth and Planetary Science Letters xxx (2010) xxxxxx

Corresponding author. Geological Sciences, University of Canterbury, Private Bag

4800, Christchurch, 8140, New Zealand.

E-mail address: [email protected] (B.M. Kennedy).

EPSL-10550; No of Pages 12

0012-821X/$ see front matter 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.epsl.2010.08.028

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

j o u r n a l h o m e p a g e : w w w . e l s e v i er . c o m / l o c a t e / e p s l

Please cite this article as: Kennedy, B.M., et al., Time-and temperature-dependent conduit wall porosity: A key control on degassing andexplosivity at Tarawera volcano, New Zealand, Earth Planet. Sci. Lett. (2010), doi:10.1016/j.epsl.2010.08.028
http://dx.doi.org/10.1016/j.epsl.2010.08.028http://dx.doi.org/10.1016/j.epsl.2010.08.028http://dx.doi.org/10.1016/j.epsl.2010.08.028mailto:[email protected]://dx.doi.org/10.1016/j.epsl.2010.08.028http://www.sciencedirect.com/science/journal/0012821Xhttp://dx.doi.org/10.1016/j.epsl.2010.08.028http://dx.doi.org/10.1016/j.epsl.2010.08.028http://www.sciencedirect.com/science/journal/0012821Xhttp://dx.doi.org/10.1016/j.epsl.2010.08.028mailto:[email protected]://dx.doi.org/10.1016/j.epsl.2010.08.028


2/12

al., 1996). Yet, no studies have addressed the isobaric time-dependent

textural changes of conduit walls in response to reheating. Here we

address two key questions: 1. How is degassing of the ascending

magma affected by changes in permeability of the conduit walls

during reheating? 2. To what extent can the release of volatiles from

the reheated wall rock contribute towards the eruption?

The vents for many explosive eruptions are often plugged by

fractured, vesicular lava domes or partially filled conduits (e.g.

Johnson and Lees, 2000; Voight et al., 1999). Surprisingly, almost noattention has been given to the effects of the hot rising magma on the

behaviour (i.e. evolution of texture and volatile content) of the lava

that plugs the volcanic conduit. This is despite observations that show

temperature rises in older lava domes prior to eruption (Wooster and

Kaneko, 1997) that correlate with eruption style (Sahetapy-Engel and

Harris, 2009). We propose that reheating can influence the degassing

of both the older plug and the rising magma. Magmatic water and

resorbed meteoricwater dissolved in glass withinthe old plug may be

available for degassing and vesiculation. This vesiculation in turn

affects the porosity and the permeability of the plug and the ability of

the rising hot magma to degas.

The 1886 Tarawera eruption (Cole, 1970; Nairn, 2002), is one of

only a few examples of basaltic plinian eruptions (Houghton et al.,

2004). At Tarawera, basalt erupts through a pre-existing dome

complex and silicic conduit system (Carey et al., 2007). This may be

a common occurrence at bimodal vents, however, descriptions of

bimodal vent exposures are absent in volcanological literature.

Detailed stratigraphic studies at Tarawera have tracked the shifting

eruption centres and fragmentation level and documented interaction

with groundwater and the pre-existing hydrothermal system (Carey

et al., 2007; Houghton et al., 2004; Sable et al., 2006, 2009). The effect

of rhyolitic conduit wall recycling during this eruption has also been

discussed (Rosseel et al., 2006).

Conduit wall permeability is an important variable in explosive

basaltic eruptions(Houghton and Gonnerman, 2008) but has not been

investigated experimentally. We use laboratory experiments on the

Tarawera rhyolitic lava to show that during an eruption the wall rock

permeability is both time- and temperature-dependent, as is the

release of volatiles from the wall rocks into the erupting magma. Weargue that a reduction in the permeability of the conduit walls as a

result of reheating hindered outgassing and increased the explosivity

of the 1886 Tarawera eruption.

2. Methodolgy

2.1. Sampling

We collected samples that contained rhyolite and basalt from the

proximal deposits of 1886 basalticfissure eruption. The motivationfor

this sampling was to collect samples that show evidence for heat

transfer between basalt and rhyolite. Enclave samples were collected

from along the rim of thefi

ssure on the summit of Tarawera lavadome (Fig. 1). We limited our samples to the crystal rich 1314 AD lava

dome xenoliths/enclaves and excluded the older crystal-poor xeno-

liths (Carey et al., 2007). We chose a single large homogeneous glassy

and perlitic sample of the lava dome to use for our experimental

starting material.From this samplewe drilled cylindrical cores 1 cm in

diameter by 2 cm in length, which were then left in an oven overnight

at 100 C to remove surface water from the samples. The precise

dimensions, density and porosity of samples were measured prior to,

and following, heating experiments. The volume of these porous

cylindrical cores was calculated from averages of replicate (n=3)

measurements of diameter and length. This volume and the

sample mass were used to calculate the bulk density (bulk) of

the core. Skeletal (or framework) density (skeletal) is obtained by

measuring sample volume via helium pycnometry. Connected

porosity (connected) (Table 1) was calculated from skeletal and bulk

density from the relationship: connected= 1(bulk/skeletal). We

obtained values of dense rock equivalent (DRE) density for rock by

crushing three cores and performing pycnometry on the resulting

powders. All experimental cores have the same average DRE density

(2.34 g/cm30.10 g/cm3). Using this average value for powder density

we compute total and isolated porosity (Table 1) as: total=1(bulk/powder), and isolated=(bulk/skeletal)(bulk/powder) (Michol et al.,

2008). Porosity values have an uncertainty of up to 4% associated with

the largest porosities measured by pycnometer (Michol et al., 2008).

2.2. Experiments

We first heated the furnace to 3001200 C (10 C) at

atmospheric pressure. This represents the expected temperaturerange of conduit rocks in contact with the erupting basalt (Rosseel

etal.,2006). Cylindrical samples(described subsequently) in ceramic

crucibles were placed in the centre of the furnace for a specified time

Fig. 1. Tarawera volcanic edifice shown as: a) geological map illustrating sampling locations on the rim of the 1886 eruption fissure (marked by X), and b) aerial photograph of the

fissure and domes looking SW from NE of the Wahanga dome and taken by Lloyd Homer, GNS Science.

2 B.M. Kennedy et al. / Earth and Planetary Science Letters xxx (2010) xxxxxx

http://dx.doi.org/10.1016/j.epsl.2010.08.028http://dx.doi.org/10.1016/j.epsl.2010.08.028


3/12

(30min72 h). This timescale was constrained by (1) the 5 h eruption

duration (Keam, 1988) and (2) a total magma rise time of a ~1 day

(Houghton et al., 2004). We characterized time and temperature

dependent changes in texture and water content between 800 and

1200 C. For temperatures 300700 C, experiments were run for 2 h to

characterize temperature dependent degassing below the calculated

theoretical glass transition (Giordano et al., 2008). Immediately

following each experiment samples were removed, cooled in air,

reweighed and their porosities remeasured with a helium pycnometer.

2.3. S.E.M.

Natural samples and samples from each experiment were coated

in carbon and examined using the Philips XL30 electron microscope

(SEM) (Beam power 15 kV and setting Spot 6) at the Earth and

Ocean Sciences Department of UBC. We compare the pre- and post-

experimental textures of the samples. Whenimaging both natural and

cored samples we avoided surfaces affected by therock saw and corer.

2.4. Water content

We ground and sieved samples, including crystals, to 100 m for

bulk water analysis from experiments lasting 30 min, 2, 4, 17.5 and

36 h. These samples were analyzed at ALS Actlabs using a Leco

induction furnace combined with spectral analysis of the emitted

volatiles. Thesample (~0.3 g) is thermally decomposed in a resistance

furnace (ELTRA CW-800) in a pure nitrogen environment at 1000 C,

causing release of volatiles, including both H2O and H2O+.

Some natural and post-experimental samples were thinned and

polished on both sides to make waferssuitable for analysis of volatiles

with the FTIR. FTIR analysis was undertaken at the University of

Oregon following the technique described in Wright et al. (2007).During analysis, we made every attempt to avoid areas containing

crystals and vesicles. Total H2O contents and the speciation of H2O in

the glass were determined using the absorbances of combination

bands at 5230 cm1 (molecular H2O) and 4520 cm1 (OH). At low

total H2O contents (b0.5 wt.%; Stolper, 1982), molecular H2O is not

detectable, in which case total water content was measured using the

3570 cm1 band, representing the fundamental OH stretching

vibration. Absorbances were converted to concentrations using the

BeerLambert law and absorption coefficients at 5230 cm1 and

4520 cm1 from Zhang et al. (1997) and at 3570 cm1 from Stolper

(1982). A glass density of 2.34 g/cm3 was obtained from pycnometry.

Sample thickness was determined using a micrometer and varied

between 300 and 900 m. Results from the 3570 cm1 band and

totals from the 5230 cm1

and 4520 cm1

bands (Table 1) agree to

Table 1

Results of high temperature experiments performed on natural samples summarized as experimental conditions, and properties of starting materials and run products. The number

in brackets in the FTIR column represents the number of measurements the mean value is based upon.

Sample name Time

(t)

T/Tg Temp.

(T)

Relax.

time

t/r log

visc.

Connected

porosity

Isolated

porosity

Total

porosity

Leco

indution

H2O

FTIR (4520+

5230

peaks)

total H2O

FTIR

(3570

peak)

total H2O

FTIR

(4520

peak)

OH

FTIR

(5230

peak)

mol. H2O

FTIR

image

mol.

H2O

(h) (oC) (h) (Pa s) (frac) (frac) (frac) (wt.%) (wt.%) (wt.%) (wt.%) (wt.%) (wt.%)

Original g 20 0.28 0.00 0.28 1.62 0.81 (4) 0.78 (2) 0.09

(4)

0.73 (4) 0.16

1.30Original h 20 0.27 0.04 0.30 1.72

Original i 20 0.21 0.08 0.29 1.61

Original j 20 0.25 0.04 0.29 1.69

mean original 20 0.25 0.04 0.29 1.660

BMK06T300 2 0.41 300 0.24 0.07 0.31 0.76

BKM06T400 2 2 0.54 400 0.23 0.05 0.28 0.43

BKM06T500 2 2 0.68 500 0.16 0.12 0.28 0.27

BKM06T600 2 2 0.81 600 0.21 0.10 0.31 0.22

BKM06T700 2 2 0.95 700 30 0.5 12.4 0.21 0.08 0.29 0.2

BK06T800 0.5 0.5 1.08 800 30 15 10.3 0.27 0.04 0.31 0.39

BK06T800 2 2 1.08 800 30 60 10.3 0.28 0.03 0.31 0.3

BK06T800 6 6 1.08 800 30 180 10.3 0.39 0.05 0.45 0.058 (1)

BK06T800 17 17 1.08 800 30 510 10.3 0 .29 0.02 0.32 0.32

BK06T800 24 24 1.08 800 30 720 10.3 0 .28 0.04 0.32 0.33

BK06T800 72 72 1.08 800 1 2160 10.3 0.32 0.03 0.34

BK06T900 0.5 0.5 1.22 900 1 476 8.8 0 .32 0.06 0.38 0.41

BK06T900 2 2 1.22 900 1 1905 8.8 0.31 0.03 0.34 0.35BK06T900 4 4 1.22 900 1 3810 8.8 0.36 0.03 0.39 0.29

BK06T900 6 6 1.22 900 1 5714 8.8 0.35 0.06 0.41

BK06T900 17 17 1.22 900 1 1.7 104 8.8 0.37 0.03 0.40 0.26

BK06T900 24 24 1.22 900 1 2.3 104 8.8 0.33 0.06 0.39 0.3

BK06T900 72 72 1.22 900 0.08 6.8 104 8.8 0.33 0.05 0.38

BK06T1000 0.5 0.5 1.35 1000 0.08 1.5 104 7.3 0.34 0.11 0.45 0.37

BK06T1000 2 2 1.35 1000 0.08 6 104 7.3 0.46 0.05 0.51 0.33

BK06T1000 4 4 1.35 1000 0.08 1.2 105 7.3 0.40 0.05 0.45 0.26

BK06T1000 6 6 1.35 1000 0.08 1.8 105 7.3 0.41 0.08 0.50 0.044 (2)

BK06T1000 17 17 1.35 1000 0.08 5.1 105 7.3 0.43 0.04 0.47 0.32

BK06T1000 24 24 1.35 1000 0 7.2 105 7.3 0.37 0.08 0.45 0.31

BK06T1100 0.5 0.5 1.49 1100 0 1.9 105 6.2 0.43 0.05 0.49 0.29

BK06T1100 2 2 1.49 1100 0 7.5 105 6.2 0.38 0.04 0.42 0.28

BK06T1100 4 4 1.49 1100 0 1.5 106 6.2 0.32 0.04 0.36 0.29

BK06T1100 6 6 1.49 1100 0 2.3 106 6.2 0.28 0.10 0.37 0.012 (4)

BK06T1100 17 17 1.49 1100 0 6.3 106 6.2 0.28 0.04 0.32 0.26

BK06T1100 24 24 1.49 1100 0 9.0 106 6.2 0.23 0.09 0.32 0.25

BK06T1200 0.5 0.5 1.62 1200 0 1.5 106 5.3 0.34 0.06 0.40 0.29BK 06T1200 24 24 1.62 1 200 1.4 106 7 .2 107 5.3

Minimum 0.5 300 0.16 0.00 0.28 0.20

Maximum 72 1200 0.46 0.12 0.51 1.72

3B.M. Kennedy et al. / Earth and Planetary Science Letters xxx (2010) xxxxxx



4/12

within 0.004 wt.%. Multiple spots were analyzed where possible and

the number of spots per measurement is shown in Table 1 using

number in brackets after the mean measurement.

Additionally, we conducted FTIR spectroscopic imaging of a

fragment of the original dome material to investigate molecular H2O

distribution. This fragment broke naturally along perlitic cracks

during further thinning of the samples, which was carried out to

reducethe numberof crystals andvesiclesin the wafers. Images(each

350350 m) of the fragment were collected using a Varian Inc.Lancer Focal Plane Array (FPA) camera attached to a Varian FTS

Stingray 700 Micro Image Analyser spectrometer and UMA 600

microscope at the Institute for Research on Earth Evolution (IFREE),

Japan Agency for Marine Earth Science and Technology (JAMSTEC).

For more detailed discussion of FTIR spectroscopic imaging see

Wysoczanski and Tani (2006). The thickness of the fragment was

determined using reflective light spectra and the wavelength of the

resulting interference fringes following the method of Wysoczanski

and Tani (2006) and Nichols and Wysoczanski (2007). A refractive

index of 1.50 was used for rhyolite (Liu et al., 2005; Long and

Friedman, 1968). From 89 reflection spectra across the image, the

average thicknesswas 62 m (1= 1 m), and this value was used for

all calculations. The glass density used was the same as for the spot

analyses. Variations in glass density as a result of variations in H2O

concentration result in a maximum error of 0.01 wt.% on the

molecular H2O contents. Owing to the thinness of the sample, the

combination bands at 5230 and 4520 cm1 used to measure

molecular H2O and OH were below detection (see Supplementary

Figure A1). As a result, the spectroscopic image for molecular H2O

represents the absorbance of the fundamental bending of molecular

H2O at ~1630 cm1. See Supplementary Figure A1 for additional

discussion of the use of this band. Total H2O contents were obtained

from the peak at 3570 cm1 as described earlier.

3. Description of natural samples

Our 30 natural samples vary in relative proportions of rhyolite and

basalt (Fig. 2). In contrast to Rosseel et al., 2006, we have chosen to

classify samples on the basis of vesicularity rather than bomb type.The samples described subsequently are all from the AD1314 lavas or

dykes; wall rocks from older Tarawera lavas (Carey et al., 2007) were

not sampled. These young lithics can be split into four vesicularity

types. Type 1 are glassy samples of the Tarawera lava dome that we

use in the experiments (Fig. 2a). Types 24 are samples from the 1886

plinian basaltic fall deposit and contain both rhyolite from the AD

1314 Tarawera lava dome and basalt from the 1886 eruption. The

samples vary from rhyolite lithics coated in basaltic scoria to

completely remelted rhyolite enclaves entirely contained within

basaltic spatter (Rosseel et al., 2006) (Fig. 2bd).

Type 1 samples are representative of the Tarawera lava dome; these

samples were collected from outcrops withinthe 1886 craterwithin the

Tarawera lava dome (Fig. 1). Samples are variably devitrified and

texturally perlitic and show a range of vesicularities. Vesicles show arange of vesicle sizes andshapes (Fig.2a).We chosea typicallargeglassy

block without obvious devitrification but with perlitic cracks (Fig. 2a)

for our experiments. All the drilled cores are from the one sample and

show a total porosity variation of 2830 vol.%. The dome is 1535%

crystalline(Cole, 1970), and contains phenocrysts of plagioclase,quartz,

biotite and minor amphibole, orthopyroxene and FeTi oxides. SEM

images show curvilinear perlitic microcracks at 50 m intervals in the

glass (Fig. 2a(ii)) and unknown secondary minerals on glass surfaces.

Water content of the starting material measured by FTIR is 0.8 wt.%and

by Leco induction is 1.6 wt.% (see Fig. 5b).

Type 2 samples were collected from the basaltic fall deposits

draping the Tarawera lava dome, exposed on the margin of the crater.

Samples are 110 cm chunks of vesicular rhyolite (Fig. 2c), partially

coated in (b

10% by volume) basaltic scoria (b

1 cm clast diameter).

The basaltic scoria is weakly attached to the surface of the rhyolite.

Wherebasaltic scoria is absent, cracks arevisible on thesurfacesof the

rhyolite. The rhyolite has total porosity estimated to range between

40 and 60% (using proportion comparison charts), only one sample

was measured at 53% (Table 1). SEM images show almost no

microcracks and complex vesicle shapes with evidence of vesicle

wall retraction and vesicle collapse (Fig. 2c (ii)).

Type 3 samples were collected from the same basaltic fall deposits

as Type 2 and the talus slopes below this fall deposit. Pyroclastscontain 1090 vol.% basalt. Samples are up to 30 cm in diameter, and

visibly very vesicular N60%, one sample was measured at 73% total

porosity with individual vesicles up to 5 mm. The surfaces of these

samples are covered in basaltic scoria and spatter which is strongly

attached to thesurface of therhyolite(Fig. 2c).Adjacent to thebasaltic

spatter, the outer few mm of the rhyolite is vesicle poor and appears

locally as black obsidian (Fig. 2c). SEM images show no microcracks

and in contrast to the Type 1 dome rock, the vesicles are spherical

(Fig. 2c (ii)).

Type 4 samples were collected from spatter-rich areas in the

basaltic fall. These pyroclasts are N90% basalt (Fig. 2d). These samples

contain small (b5 cm) dense enclaves of rhyolite within basaltic

spatter. The rhyolite blebs are generally too small to measure

porosities, however, we estimate the total porosity is consistently

b25%. SEM images show a range of textures, with prominent small

spherical and irregular vesicles (Fig. 2d (ii)).

4. Experimental results

4.1. Qualitative results: SEM

We present the results of the experimental heating of 32 Type 1

lava dome cores with the aim of understanding the potential

consequences of reheating conduit filling material. SEM images

showing time- and temperature-dependent changes in the micro-

structure of the samples are shown in Figure3. Generally, the textures

progress from cracked vesicular glassy original (Type 1) textures, to

an increase in vesicles with cracked walls, to a connected network of

un-cracked vesicles, and finally to less vesicular isolated vesicles.Experimentslasting 2 h at 300700 C showno noticeabledifference

in texture from the original samples. However, the time series from

experiments at 8001200 C show remarkable differences in the

geometry and size of cracks and vesicles relative to the initial state.

At 800 C, samples heated for ~6 h are characterized by open cracks.

In addition, the widest cracks are spatially associated with dome

structures (related to vesicle growth) (Mungall et al., 1996) on the

surface of the glass which are 100 m in diameter (Fig. 3a). In samples

heated for 24 h cracks are still apparent and some cracks appear more

curved and crack edges appear slightly rounded (Fig. 3b).

At 900 C, samples heated for ~6 h (Fig. 3c) aretexturally similar to

the 800 C runs (Fig. 3a). However, at 24 h, cracks are commonly

curved. The dome structures appear to have surfaces made of a thin

film of glass and relict cracks and rafts of original surfaces exist onsome domal structures (vesicles) (Fig. 3d), suggesting that films of

melt have allowed vesicles to grow.

In experiments conducted at 1000 C, the edges of cracks are

distinctly rounded indicating (re-)melting (Fig. 3e). In the sample

heated for 24 h at 1000 C most of the glass is smooth with angular

cracks only occurring in crystals. In addition, completely smooth

domes occur with only a vague hint of relict cracks (Fig. 3f). The

domes appear as connected vesicles, however these vesicles are not

bulbous, and some appear deflated (Fig. 3f).

At 1100 C, a sample heated for 40 min shows similar textures to

the sampleheatedfor 24 h at 1000 C (Fig. 3f and g). The surface of the

sample is smooth with bulbous vesicle-like domes, and no relict

cracks (Fig. 3g). The surface of the sample heated for 6 h at 1100 C is

completely smooth and cracks are no longer visible in the glass. Most




5/12

of the glass is bulbous and appears to be connected vesicles ( Fig. 3g).

The glass in the sample heated to 1100 C for 24 h also is smooth and

has no cracks. Vesicles are not as apparent on the surface of the

sample (Fig. 3h).

Finally, the sample heated for 30 min for at 1200 C (Fig. 3i) shows

similar textures to the sampleheated for 6 h at 1100 C (Fig. 3g). Glass

is smooth and unfractured and contains connected vesicles (Fig. 3i).

The sample heated at 1200 C for 24 h has smooth surfaces similar to

that heated to 1100 C (Fig. 3h), however fresh surfaces show some

isolated vesiclesexist beneath thesmooth surface (Fig. 3j). All original

textures were obliterated.

In summary, experiments at lower temperatures and/or for short

times generally produced samples with open cracks. Experiments at

higher temperatures and/or for longer times show evidence of

melting, crack annealing, the growth offilms ofmelt, and the inflation

and deflation of vesicles.

Fig. 2. (i) Photographs and (ii) corresponding scanning electron micrographs of rhyolitic samples from Tarawera illustrating 4 main types of vesicularity. (a) Type 1: (i) vesicularity in

vesicular lavadomeused as thestarting material forour experiments; (ii)S.E.M.imageshowing curvilinearmicrocracksand varietyof vesiclesizes andshapes.(b) Type2: (i)vesicularity

withina rhyolitic pyroclast partially coated in basalticscoria;(ii) S.E.M. image showingsmoothsurfaces,retracting vesiclewalls,largeconnectedvesicles andno cracks.(c) Type3: (i)high

vesicularity in a rhyolitic pyroclast coated in basaltic spatter, in some areas 13 mm thick obsidian occurs at the boundary between rhyolite and basalt; (ii) S.E.M. image showing large

connected spherical vesicles and no cracks. (d) Type 4: (i) vesicularity in basaltic spatter containing dense enclaves of rhyolite; (ii) S.E.M. image showing a dense rhyolitic enclave with

some isolated irregular vesicles towards its margin.




6/12

Fig. 3. SEM images of experimental run products produced by heating of natural samples over fixed periods of time. a) Experimental run product for 800 C over 6 h is largely

texturally unchanged from the original. However, cracks have begun to open especially in areas around domed glass associated with subsurface vesicles. b) Run product for 800 C

over 24 h shows similar cracking associated with sub-surface vesicles;also shows convexareasand some cracksappear to be rounded. c) Runproductfor 900 C over 6 h shows open

cracks associated with sub-surface vesicles. d) Heating at 900 C for 24 h causes formation of new vesicles that feature films of melt that heal old fractures and create rafts of the old

vesicle wall on the surface of the new vesicle. e) Sample heated at 1000 C for 6 h shows angular fracture edges rounding and annealing. g) Sample heated at 1000 C for 24 h shows

almost no vestiges of the earlier cracked surfaces, however, existing dome-like structures appear deflated. g) Samples heated at 1100 C for 6 h feature connected inflated vesicles

within the original vesicle wall septa, and no evidence of the original cracked glass. h) Heating at 1100 C for 24 h produces a sample on which the surface shows no evidence of the

original bubbly and cracked surface. i)Run productfor 1200 C over30 min is similar to g), vesicle wallscontain inflated connected vesicles. j)Heating at 1200 Cfor 24 h causes the

sample to collapse and flow; material had to be chipped out of the crucible and internally it shows spherical isolated vesicles.




7/12

4.2. Porosity

Time series of fractional porosity evolution during the experiments

are given in Table 1 and illustrated in Figure 4. Both total and

connected porosity initially increase with time at all experimental

temperatures. The time taken to reach a maximum in total porosity

decreases as the experimental temperature increases. This increase in

porosity is then followed by a reductionin porosity,wherethe specific

total vs. connected porosity path is dependent upon temperature.Isolated porosity is illustrated in Figure 4c by the vertical distance

between any point and the dashed, fully connected porosity line. At

b700 C, overall porosity and volume were indistinguishable within

error from the unheated samples (Tables 1 and Supplementary Table

A2).

At 800 C theconnected porosity reaches a maximum of 0.39 in the

experiment conducted for 6 h (Fig. 4a). Experimental times greater

than 6 h correspond with lower connected and total porosities.

Isolated porosity remains less than 0.05 in all experiments at 800 C.

The time series at 900 C also shows an initial increase in

connected and overall porosity, connected porosity then fluctuates

and reaches a maximum at 17.5 h of 0.37. Thereafter the connected

porosity drops monotonically to 0.33 at 36 h (Fig. 4b). Again, the

isolated porosity remains similar during this time series (Table 1).

The experiments at 1000 C, have a maximum total and connected

porosity of 0.46 at 2 h. Longer experiments correspond with smaller

porosities (Fig. 4c). This time series shows higher isolated porosities

(up to 0.11) compared to 800and 900 C,particularly theexperiments

for 2, 6 and 24 h.

At 1100 C a maximum porosity of 0.43 occurs at 30 min.

Thereafter, the connected porosity drops sharply to 0.23 by 24 h.

Despite the consistent decrease in connected and overall porosity,

isolated porosity varies considerably and shows a substantial peak

(~0.1) at 6 h (Fig. 4d).

Samples heated at 1200 C collapsed and lost their cylindrical form

(textural collapse), the sample heated for 30 min remained generally

cylindrical whilst the sample heated for 24 h collapsed completely.

In summary, at all experimental temperaturesan initial increase in

total porosity is followed by a decrease.

4.3. Water content

The water content of the glass in the original sample (starting

material) varies between 1.6 and 0.8 wt.% as measured by Leco

induction furnace and by FTIR spot analyses (N=4), respectively

(Fig. 5b, Table 1). These differences in measured water contents

between Leco Induction furnace and FTIR are absent at higher

temperatures and lower water contents. The FTIR spectroscopic

image of the absorbance for the molecular H2O band (Fig. 5a)

supports this, with absorbance increasing, by up to a factor of 7

compared to the interior of the fragment, at the perlitic margins.

Notwithstanding the qualifying statements of Zhang et al. (1997)(described in the figure caption of the Supplementary Fig. A1) this

represents an increase in molecular H2O content from approximately

0.16 to 1.30 wt.%. All time series show that N75% of bulk water

measured by Leco induction was lost within 0.5 h. Over 2 h at 500 C

the bulk water content shows a systematic decrease as temperature

increases from 1.6 wt.% to around 0.22 wt.%, which is close to the

detection limit of this instrument (Fig. 5b). Within the uncertainty of

the measurements, repeated experiments at temperatures above

500 C showed no change in bulk water content. FTIR spot measure-

ments using the 3550 cm1 band support the same general pattern of

rapid water loss and show a mean initial water content of 0.78 wt.%

was reduced to 0.06% after 6 h at 800 C (Fig. 5b). Additionally, these

FTIR spot analysis support continued degassing to 0.01 wt.% after 6 h

at 1100 C.

4.4. Results summary

Our observations and measurements of the rhyolite lava prior to

experimentation illustrate that the lava has an intricate network of

perlitic cracks (Fig. 2a) and connected vesicles (connected porosity

0.26 and isolated porosity 0.04). In addition, a significant portion of

the water contents of the original dome glass are not accounted for by

the water species measured by either spot or FTIR mapping. Water

contents that are effectively mapped by FTIR are shown to be stronglyenriched along the margins of the perlitic cracks (Fig. 5a). Our

experimental results show that our samples, where heated above

500 C lose most of their water within 2 h. However, changes in total

porosity do not occur until 800 C. All time series above 800 C

showed a small but continued degassing and an initial increase in

connected porosity associated with cracking and vesicle growth,

followed by a decrease in porosity associated with vesicle collapse.

Generally, the hotter the experiment the earlier in the time series the

porosity decrease occurred. Textural changes are summarized in

Fig. 6, experiments are subdivided into samples with (1) no textural

changes, (2) cracking/ inflation,(3) vesicle collapse/deflation,and (4)

textural obliteration. This subdivision is used to develop degassing

and deformation regimes which are discussed below.

5. Discussion

5.1. Comparison between natural and experimental samples

By comparing the textures of the naturally heated samples and the

experimentally heated samples we constrain the porosity and, by

inference, thepermeability history of thedome and conduit wall rocks

in response to reheating. In general, there is a striking similarity

between the textures of type 24 samples and the highertemperature

and long timescale experiments we performed (Figs. 2 and 3). The

textural similarities support our estimates of the timescales and

temperatures of reheating that were derived from the eruption

timescale (Keam, 1988). However, we do not imply that the thermally

driven textural changes in the natural samples occurred in-situ at theconduit margin; most of these erupted as bombs and probably

followed the thermal history described in Rosseel et al. (2006). The

open cracks and cracked vesicle walls of lower temperature and

shorter timescale experiments were not seen in our suite of naturally

heated samples, although this may be due to sampling bias.

The type 2 samples (Fig. 2b) have textures and porosity similar to

experiments for 24 h at 1000 C (Fig. 3f). In both these rocks, glass

surfaces are smooth, crackshave healed but vesicles remain small and

only partially inflated.

The type 3 samples (Fig. 2c), appear similar to experiments above

1100 C (Fig. 3g and h) and show no evidence of the original textures

of the type 1 rock (Fig. 2a). However, the experiments do not exhibit

the large vesicle sizes and high porosities of the natural type 3

samples (Fig. 2d). In our experiments above 1100 C, at timescaleslonger than 40 min, vesicles coalesced, connected with the ambient

pressure and collapsed. The large size of the natural samples allowed

more vesicle growth and coalescence, and the developments of a

larger isolated porosity before depressurizing to ambient pressure.

Such an internal pressure could prevent collapse of the sample and

loss of porosity (Yoshimura and Nakamura, 2008). Alternatively,

vesiculation of these samples could have been aided by decompres-

sion as they erupted.

The type 4 samples are similar to experiments at 1200 C for

timescales greater than 30 min, no original vesicle textures resem-

bling the type 1 sampleremain. Theamoeboid outer shapes and dense

textures of the type 4 samples imply that they fully melted. Vesicles

are generally small, implying that larger vesicles coalesced, collapsed

or escaped by migration through the melt.




8/12

The similarity of textures shared by the experimental and natural

rocks implies that reheating drove vesiculation and vesicle collapse in

the natural samples. Natural type 2 samples found in the basaltic

scoria reached temperatures of 1000 C for at least 24 h, whereas type

3 and 4 samples found in spatter imply higher temperatures and/or

longer hot residence times (Carey et al., 2008; Rosseel et al., 2006;

Sable et al., 2009). In summary, the comparison of Figures 2 and 3

show that the temperatures and timescales of reheating during our

experiments were appropriate for the eruption, and that largevariations in conduit wall porosity occurred during the eruption. An

ongoing study of the scoria and spatter filled basaltic dyke margins

exposed in the base of the fissure support this porosity variation.

Additionally, this fieldwork constrains the thermal impact of the

basalt to be between a few millimetres to tens of centimetres, in some

areas we identified a distinct 5 cm thick welded portion; and these

detailed field descriptions will be the subject of a future publication.

5.2. Degassing, and deformation regimes

Degassing and deformation proceeded differently in our experi-

ments relative to other experimental studies involving hydration of

melts (Baker et al., 2006; Gardner, 2007; Larsen et al., 2004; Takeuchi

et al., 2009; Yoshimura and Nakamura, 2008). Our samples were

composed predominantly of glass naturally hydrated by water along

perlitic cracks at temperatures well below Tg (Denton et al., 2009). As

a result this water can be outgassed at temperatures well below Tg(Tuffen et al., 2010). Isotopic studies indicate that this water is likely

to be meteoric in origin (DeGroat-Nelson et al., 2001; Friedman and

Smith, 1958; Friedman et al., 1966; Shane and Ingram, 2002). An

additional contrast to previous experiments is that our experimental

sample already had a high proportion of connected porosity through

cracks and vesicles. This distinction is important because therelationship between permeability and porosity is different during

vesicle growth and vesicle collapse (Michaut and Sparks, 2009; Rust

and Cashman, 2004). A sample that has undergone vesicle collapse

will have a higher ratio of permeability to porosity than a sample of

similar porosity that has not undergone vesicle collapse (Michaut and

Sparks, 2009; Mueller et al., 2008). The complexities of this

relationship are most apparent when illustrated by up to 6-fold

increase in permeability over the porosity range of 3040% (Wright

et al., 2009). The permeability/porosity relationship is further

complicated by the presence of cracks with the ability to open and

to heal (Yoshimura and Nakamura, in press). However, all our

experiments had the same starting texture and we are confident that

incremental changes in open porosity correlate with changes in

permeability.

Fig. 4. Experimental results summarized as total porosity vs. connected porosity and labelled by experimental times. Open headed arrows follow a temporal evolution of increasing

experimental time. A dotted 1:1 line indicating equality between connected and total porosity is shown on all plots; the vertical distance between each data point and this line

measures theisolated porosity withineach sample. a) Time seriesat 800 C;insetof solid headedarrows shows howporosityis affected by various relevant processes.b) Time series

at 900 C. c) Time series at 1000 C. d) Time series at 1100 C.




9/12

On reheating, our experiments imply that the porosity and

permeability structure of lava domes or plugs will pass through a

series of regimes, the nature of which depend on 1) the temperature

of the event; 2) the time spent at that temperature (Fig. 6a); and 3)

the volatile content and original vesicularity of the rock being heated.

Physically, the character of the change in porosity and, by inference

the permeability, depends on whether dome rocks respond in a

viscous or brittle way to heat transfer from the newly erupting

magma. Thus, the key parameter is the glass transition, which

depends on the temperature of the event relative to the glass

transition temperature Tg (Knoche et al., 1994) and the time scale ofthe event relative to a critical relaxation time for structural failure of

the melt. Consistent with experimental results, we take this critical

time to be proportional to the viscous relaxation time r=/G (Webb

and Dingwell, 1990) (Table 1). Here, G =1010 Pa is the elastic

modulus, is the melt viscosity at a given temperature and water

content and C is a constant to be determined from our experiments.

We take Tg to be the temperature corresponding to a viscosity of

1011.4 Pa s which correlates well with the calorimetric Tg (Giordano

et al., 2008). The value ofTg is calculated to be 740 C for this rhyolite

glass composition (Nairn et al., 2004) at 0.1 wt.% H2O using the

viscosity model of Giordano et al. (2008) (see Supplementary Table

A2). Thetransition from brittle to viscous behaviour(Regimes1 to 3 in

Fig. 6) depends on the temperature (T) relative to Tg and the time

scale for the experiment (t) (or eruptive event).

To gain additional insight we replot the data in Figure 6a in terms

ofTnormalized to Tgand tnormalized to r(Fig. 6b). Thedata collapse

to a power law defining the transition from brittle to viscous

behaviour of the form T/ Tg=C(t/r)0.046, where C ~ 0.86. This result

shows that whether or not brittle processes govern the final porosity

of the sample depends on the response time of the melt, which is

governed by its strongly temperature-dependent viscosity, relative to

the time scale of the thermal forcing applied in the experiment.

We use our experimental degassing and deformation data to

propose four degassing and deformation regimes (Fig. 6). In Regime 1

(T800, t2 h; T/TgNCt/r0.046) degassing occurs without significantdeformation (Fig. 6). The rock maintains its vesicle structure

connected by microcracks and ruptured vesicle walls (Fig. 2a, b)

and consequently maintains a high proportion of connected to total

porosity (Fig. 4). No visible changes to this structure could be seen

with the SEM. Between 0 and 700 C we attribute this initial H2O

degassing to (1) release of water captured in micropores and/or low

temperature alteration hydrous minerals (Denton et al., 2009), and

(2) diffusion of resorbed meteoric water out of the perlitic margins

(Fig. 5a) of the glass (Tuffen et al., 2010). 0.200.05 wt.% magmatic

water remains dissolved in the glass interior (Fig. 5). This initial

period of degassing had no impact on the textures of the rock as the

rock was not sufficiently above its calculated glass transition

temperature (Giordano et al., 2008) and ductile deformation did not

occur.

Fig. 5. a) FTIR spectroscopic images of a fragment of unheated original sample that broke along perlitic cracks: (i) photomicrograph of fragment; (ii) planar spectroscopic map of

absorbance of the molecular H2O band at 1630 cm1 (see Supplementary Fig. A1) across the fragment; the fragment is resting on a H2O-free IR-invisible KBr disk (absorbance=0);

(iii) three-dimensional view of same image indicating that there is some absorbance in the interior of the fragment. Note that one of the edges of the fragment is not enriched in

molecular H2O, suggesting that this edge did not form along a perlitic crack. Colour scale applies to both images. Molecular H2O contents are estimated using a constant molar

absorption coefficient and an average thickness and density for the whole fragment (also refer to Supplementary Fig. A1). b) Weight percent water plotted against temperature for

experiments lasting 2 h. Black squares show water contents of glass measured with FTIR spectroscopy and black circles show bulk water content measured by Leco induction.

Vertical bars show the range in water content from multiple analyses and are due to variation within samples.




10/12

Regime 2 (800T1100 C, t6 h; T/TgC t/r0.046) is a transi-

tional regime showing both brittle and viscous deformation (Fig. 6).

Degassing of magmatic water (b0.1 wt.%) continues (Fig. 5) and

vesicle and crack expansion cause 520%increasesin overall porosity(Fig. 4). Observations show the porosity changes are due to local

viscous vesicle growth and coalescence with concurrent brittle

cracking in areasof high strain (Fig. 3ac). Our experimental samples

have initial porosities of 0.290.31, and connected porosity of 0.27;

the growth of isolated vesicles frequently impinge on other vesicles

(Fig. 3ac), resulting in coalescence and increasing connected

porosity with relatively small increases in isolated porosity

(Fig. 4a). In addition, the opening of cracks (Fig. 3b, c) connects

isolated vesicles, increasing the connected porosity and reducing

isolated porosity (Fig. 4a).

In Regime 3 (800T1200 C, t30 min; T/TgbCt/r0.046) porosity

loss occurs due to vesicle deflation (Fig. 6). This regime is

characterized by crack sealing and vesicle collapse leading to a

transient but generally decreasing overall porosity (Fig. 4). Both

processes involve viscous relaxation and appear to occur at similar

temperatures and timescales. Vesicle collapse is driven as the internal

pressure of the vesicle can no longer balance the surface tension and

gas leaks out of the vesicles through the connected porosity. Crack

healing occurs as melt connects opposite sides of a crack (Fig. 3e,f).

Healing of cracks reduces connected porosity and increases isolated

porosity (Fig. 4). However, generally, in our experiments observations

of crack healing correlate with decreases in overall porosity which can

only be explained by vesicle collapse (Figs. 3 and 4). Occasional smallincreases in porosity are seen in this regime and we interpret these to

be due to continued degassing and vesicle growth, as larger coalesced

vesicles collapse.

In Regime 4 (T1200 C, tN30 min; T/Tg NNCt/r0.046) textural

collapse occurs due to viscous flow of the melt in response to gravity

(Fig. 6). Although we could not measure the residual porosity of these

run-products, SEM image analysis shows a significantly reduced

porosity comprising isolated pores (Fig. 3k) implying a much reduced

permeabilty.

The temperature/time space of each of these regimes is ultimately

controlled by the viscosity of the melt/glass framework in the dome

lava which is itself strongly influenced by the dissolved water content

(inset Fig. 6a). The calculated dry value for Tg of this melt (see

Supplementary Table A2) is 770 C. The equilibrium volatile content

of the glass will be controlled by pressure and therefore the

openness and porosity of the volcanic system. An open, degassed

system such as the system reproduced by our experiments has a high

Tg of 740 C (~0.1 wt.% H2O) and is relatively difficult to remelt and

initiate Regimes 2, 3 and 4. However, a closed system with a higher

ambient pressure and equilibrium volatile content will have a

depressed Tg. For example, this melt with 1 wt.% H2O has a calculated

Tg of 606 C (see Supplementary Table A2, Giordano et al., 2008).

Therefore, if such a system is reheated it may achieve Regimes 2, 3,

and 4 at lower temperatures.

Thestyle of hydration, the original texture, andthe amount of time

a reheated rock/magma spendsin Regime 1 will strongly influence the

relationship between water content and its calorimetric Tg, and ability

to continue to other regimes. Similarly the amount of time a rock

spends in permeable Regimes 1 and 2 will influence the ability of anymagma beneath it to degas. For these reasons heating rates and

partially closed systems become important considerations and

avenues for future research.

6. Conclusions and implications for Tarawera

The frozen magma forming the conduit walls will be reheated

during the ascent and eruption of fresh magma. The time and

temperature dependent textural changes in the plug or walls have

implications for the monitoring of degassing and deformation of

active volcanoes, and on the resultant style and magnitude of an

eruption. In particular, a key issue is whether reheating leads to the

production or destruction of permeability in the conduit walls. For

example, enhanced permeability facilitates outgassing, which canreduce the overpressure in the intruding magma, and favour effusive

volcanism (e.g., Quane et al., 2009). In contrast, reduced permeability

can inhibit outgassing, leading to greater overpressure in the new

magma and an increased likelihood for explosive volcanism.

From our experiments, the evolution of wall rock texture during an

eruption depends on the temperature and duration of the event (Fig. 6).

In Regime 1 (T/TgNCt/r0.046), there is degassing of rehydrated meteoric

water from the rhyolite. For Tarawera dome rock, up to 1.4 wt.% water

could be released by conduit wall and old dome rocks into the erupting

magma during reheating. In Regime2 (T/TgCt/r0.046) both ductile and

brittle deformation cause increases in connected porosity and inflation.

This increase in connected porosity should correlate with an increase in

permeability and hence aid the degassing of the fresh magma below.

Regime 3 (T/Tgb

Ct/r0.046

) is a result of vesicle collapse and porosity

Fig. 6. a) Sample deformation style separated into Regime 1 no deformation, Regime 2

cracking/ inflation, Regime 3 vesicle collapse/ deflation, and Regime 4 textural

obliteration plotted in time and temperature space.The dashedlines markapproximate

boundaries between regimes. The arrow suggests a heating trajectory based on the

thickness of the remelted dyke margin measured in the field. The curved solid line

represents the calorimetric Tg which increases as water is lost; below this line

conditions are equivalent to Regime 1. The line is calculated approximating time to a

decreasing H2O content (measured by FTIR spectroscopy, Fig 5b) and from the

calorimetric Tg calculated using Giordano et al. (2008; see inset and Supplementary

Table A2). b) Tnormalized to Tg and tnormalized to a critical strain rate for structural

failure of the melt (r). This plot shows that transitional Regime 2 plots at T/Tg~t/r0.046,

below this critical power law degassing occurs through pre-existing structures and

above power law viscous behaviour allows vesicle collapse.




11/12

reduction. Although a connected pore structure still remains, perme-

ability will be reduced progressively closing the degassing system

(Westrich and Eichelberger, 1994; Yoshimura and Nakamura, 2008;

Yoshimura and Nakamura, in press). Regime 4 produces a generally

isolated pore structure and permeability would be significantly

decreased causing pressure build up, that could give rise to explosive

eruptions.

The magnitude of thermally-induced permeability change is

expected to depend on the length scale L = 2

ffiffiffiffiffi

t

p

, where the thermaldiffusivity k of potentially foamed rhyolitic dome rocks is order

107 m2 s1 (Bagdassarov and Dingwell, 1994). For Tarawera, over the

proposed 524 hour magma rise and eruption time (Houghton et al.,

2004; Keam, 1988) L is likely to be approximately 825 cm (Fig. 6a).

Indeed, on-going field studies of the margins of basalt dykes in the

lava domes reveala 5 cm thick remelted layer of rhyolite with reduced

porosity. This 5 cm thick remelted layer is remarkably consistent with

the eruption timescale and thermal diffusion length scale (Fig. 6a).

However, detailed descriptions of these dyke margins are not yet

complete and beyond the scope of this paper.

For the 1886 eruption, we propose a reheating trajectory of 5 h

(Fig. 6a) which is consistent with (1) the eruption timescale

(Houghton et al., 2004; Keam, 1988), (2) the presence of all three

naturally heated sample types, and (3) the thicknesses of remelted

dyke margins.

The volume of AD 1314 Tarawera rhyolite affected is difficult to

estimate dueto uncertainties in thesubsurfacegeometryof these and

older rhyolite lavas and feeder dykes, and their relationship to the

basalt dyke (e.g. Fig. 4 in Carey et al., 2007). We take the basalticfissure to be 8 km long (Fig. 1) in map view and 0.28 km deep (the

lower bound assumes that the basalt interacts only with the shallow

dome; the upper bound implies interaction with a conduit over the

full depth to the 8 km rhyolitic magma source (Shane et al., 2008)).

Over the 5 houreruption duration the intruding magma will heatand

degas the surrounding rhyolitic wall rocks to a distance of 8 cm

(Fig. 6a). From these dimensions and assuming that the wall rocks

contain 1.4 wt.% H2O, reheating would release around 6.4106

2.6108 kg of H2O. From our experiments, the majority of this

volume would be released within the first 2 h of reheating and maysignificantly contribute to precursory signs of eruption, although it

may not be significant in terms of the total volatile budget of the

eruption.

The similarity between the textures in the rhyolite erupted from

Tarawera volcano and the experiments defining Regime 4 imply that

conduit-wall permeability may have been catastrophically reduced

during the early stages of the Tarawera eruption. We propose that

such a sealing of the system may have produced a closed system and

aided the unusually explosive basaltic eruption of Tarawera in 1886.

This offers an additional and complimentary explanation to the

driving mechanisms suggested by previous authors e.g., phreato-

magmatic interactions and bubble/microlite coupling (Carey et al.,

2007; Sable et al., 2009).

Finally, our work also has implications for eruptions related to re-intrusion of similar temperature magma. A closed pressurized lava

dome and conduit with 13 wt.% H2O (e.g. Burgisser et al., 2010)

could potentially depress Tg (Fig. 6a inset) and hence allow melting

associated with the re-intrusion of magmas of similar composition

(Giordano et al., 2008), i.e. a 700 C rhyolite could re-intrude and melt

a water rich rhyolite with a Tg depressed to 600 C. Our work predicts

that as an old lava dome or plug heats up it will initially inflate,

partially degas, then deflate, and seal up leading to continued

pressurization and possibly eruption (Fig. 6). This sequence may be

followed by another reheating event and lead to the cyclic

deformation observed at many lava domes ( Johnson et al., 2008;

Matthews et al., 1997; Voight et al., 1999).

Supplementary materialsrelated to this article canbe found online

at doi:10.1016/j.epsl.2010.08.028.

Acknowledgements

Funding for MJ was provided by NSERC, Canadian Institute for

advanced Research, and Marsden (UOC0508). Funding for BK was

provided by Marsden Fast start (09-UO-017C). Help with measuring

water contents was provided by Paul Wallace, and John Stix.

Additional help with field access was provided by Ken Ruaeti and

the Ruawahia 2b Trust and Judy Collins of Mt Tarawera tours, Paul

Ashwell, Felix VonAulock, and FabianWadsworth. We would also liketo thank reviewers Dr. Heather Wright and Dr. Hugh Tuffen for their

detailed and insightful reviews that significantly improved the

manuscript.

References

Bagdassarov, N., Dingwell, D., 1994. Thermal properties of vesicular rhyolite. J. Volcanol.Geoth. Res. 60, 179191.

Baker, D.R., Lang, P.G., Robert, G., Bergevin, J.F., Allard, E., 2006. Bubble growth inslightly supersaturatedalbite meltat constant pressure. Geochim. Cosmochim.Acta70, 18211838.

Brooker, R.A., Kohn, S.C., Holloway, J.R., McMillan, P.F., 2001. Structural controls on the

solubility of CO2 in silicate melts: Part II: IR characteristics of carbonate groups insilicate melts. Chem. Geol. 174, 241254.Burgisser, A., Poussineau, S., Arbaret, L., Druitt, T.H., Giachetti, T., Bourdier, J.-L., 2010.

Pre-explosive conduit conditions of the 1997 Vulcanian explosions at SoufrireHills Volcano, Montserrat: I. Pressure and vesicularity distributions. J. Volcanol.Geoth. Res. 194, 2741.

Carey, R.J., Houghton, B.F., Sable, J.E., Wilson, C.J.N., 2007. Contrasting grain size andcomponentry in complex proximal deposits of the 1886 Tarawera basaltic Plinianeruption. Bull. Volc. 69, 903926.

Carey, R.J., Houghton, B.F., Thordarson, T., 2008. Contrasting styles of welding observedin the proximal Askja 1875 eruption deposits II; local welding. J. Volcanol. Geoth.Res. 171, 2044.

Cole, J.W., 1970. Structure and eruptive history of the Tarawera complex. NZ J. Geol.Geophys. 13, 879902.

DeGroat-Nelson, P.J., Cameron, B.I., Fink, J.H., Holloway, J.R., 2001. Hydrogen isotopeanalysis of rehydrated silicic lavas: implications for eruption mechanisms. EarthPlanet. Sci. Lett. 185, 331341.

Denton, J.S., Tuffen, H., Gilbert, J.S., Odling, N., 2009. The hydration and alteration ofperlite and rhyolite. J. Geol. Soc. 166, 895904.

Friedman, I., Smith, R., 1958. The deuterium content of water in some volcanic glasses.Geochem. Cosmochem. Acta 15, 218228.

Friedman, I., Smith, R.L., Long, W.D., 1966. Hydration of natural glass and formation ofperlite. Bull. Geol. Soc. Am. 77, 323328.

Gardner, J.R., 2007. Bubble coalescence in rhyolitic melts during decompression fromhigh pressure. J. Volcanol. Geoth. Res. 166, 161176.

Giordano, D., Russell, J.K., Dingwell, D.B., 2008. Viscosity of magmatic liquids. EarthPlanet. Sci. Lett. 271, 123134.

Gonnerman, H.M., Manga, M., 2003. Explosive volcanism may not be an inevitableconsequence of magma fragmentation. Nature 426, 432435.

Gonnermann, H.M., Manga, M., 2007. The fluid mechanics inside a volcano. Ann. Rev.Fluid Mechs. 39, 321356.

Hammer, J.E., Rutherford, M.J., 2002. An experimental study of the kinetics ofdecompression-induced crystallisation in silicic melt. J. Geophys. Res. 107, 124.

Houghton, B.F., Gonnerman, H.M., 2008. Basaltic explosive volcanism: constraints fromdeposits and models. Chem. Erde 68, 117140.

Houghton, B.F., Wilson, C.J.N., Del Carlo, P., Coltelli, M., Sable, J.E., Carey, R., 2004. Theinfluence of conduit processes on changes in style of basaltic Plinian eruptions:Tarawera 1886 and Etna 122 BC. J. Volcanol. Geoth. Res. 137, 1 14.

Jaupart, C., 1998. Gas loss from magmas through conduit walls during eruption. In:Gilbert, J.S., Sparks, R.S.J. (Eds.), The Physics of Explosive Volcanic Eruptions: Geol.Soc. Sp. Pubs, 145, pp. 7390.

Johnson, J.B., Lees, J.M., 2000. Plugs and chugs seismic and acoustic observations ofdegassing explosions at Karymsky, Russia and Sangay, Ecuador. J. Volcanol. Geoth.Res. 101, 6782.

Johnson, J.B.,Lees, J.M.,Gerst, A., Sahagian, D., Varley, N., 2008.Long-period earthquakesand co-eruptive dome inflation seen with particle image velocimetry. Nature 456,377381.

Keam, R.F., 1988. Tarawera: the volcanic eruption of 10 June 1886, p. 472. Published bythe author, Auckland, New Zealand.

Kennedy, B., Spieler, O., Scheu, B., Kueppers, U., Taddeucci, J., Dingwell, D.B., 2005.Conduit implosion during Vulcanian eruptions. Geology 33, 581584.

Knoche, R., Dingwell, D.B., Seifert, F.A., Webb, S.L., 1994. Non-linear properties ofsupercooled liquids in the system Na2OSiO2. Chem. Geol. 116, 116.

Larsen, J.F., Denis,M.H.,Gardner,J.E.,2004.Experimental study ofbubble coalescence inrhyolitic and phonolitic melts. Geochim. Cosmochim. Acta 68, 333344.

Lavalle, Y., Hess, K.-U., Codonnier, B., Dingwell, D.B., 2007. Non-Newtonian rheological

law for highly crystalline dome lavas. Geology 35, 843

846.




12/12

Lavallee, Y., Meredith, P.G., Dingwell, B.D., Hess, K.-U., Wassermann, J., Cordonnier, B.,Gerik, A., Kruhl, J.H., 2008. Seismogenic lavas and explosive eruption forecasting.Nature 453, 507510.

Liu, Y., Zhang, Y., Behrens, H., 2005. Solubility of H2O in rhyolitic melts at low pressuresand a new empirical model for mixed H2OCO2 solubility in rhyolitic melts. J.Volcanol. Geoth. Res. 143, 219235.

Long, W., Friedman, I., 1968. The refractive index of experimentally hydrated rhyoliteglass. Am. Mineral. 53, 17541756.

Matthews, S.J., Gardeweg, M.C., Sparks, R.S.J., 1997. The 1984 to 1996 cyclic activity ofLascar Volcano, northern Chile; cycles of dome growth, dome subsidence,degassing and explosive eruptions. Bull. Volcanol. 59, 7282.

Melnik, O., Sparks, R.S.J., 2002. Modeling of conduit flow dynamics during explosiveactivity at Soufrire Hills Volcano, Montserrat, West Indies. In: Druitt, T.H.,Kokelaar, B.P. (Eds.), The Eruption of the Soufrire Hills Volcano, Montserrat, from1995 to 1999: Geological Society of London Memoir, 21, pp. 307 317.

Michaut, B.D., Sparks, R.S.J., 2009. Ascent and compaction of gas rich magma and theeffects of hysteretic permeability. Earth Planet. Sci. Lett. 282, 258267.

Michol, K.A., Russell, J.K., Andrews, G.D.M., 2008. Welded block and ash flow depositsfrom Mount Meager, British Columbia, Canada. J. Volcanol. Geoth. Res. 169,121144.

Mueller, S., Scheu, B., Spieler, O., 2008. Permeability control on magma fragmentation.Geology 36, 399402.

Mungall, J.E., Bagdassarov, N.S., Romano, C., Dingwell, D.B., 1996. Numerical modellingof stress generation and microfracturing of vesicle walls in glassy rocks. J. Volcanol.Geoth. Res. 73, 3346.

Nairn, I.A., 2002. Geology of the Okataina Volcanic Complex. Inst. Geol, Nuc. Sci,geological map 25, 1 sheet +150p. Lower Hutt, NZ.

Nairn, I.A., Shane, P.R., Cole, J.W., Leonard, G.J., Self, S., Pearson, N., 2004. Rhyolitemagma processes of the VAD 1315 Kaharoa eruption episode, Tarawera volcano,New Zealand. J. Geotherm. Volcanol. Res. 131, 265294.

Newman, S., Stolper, E.M., Epstein, S., 1986. Measurement of water in rhyolitic glasses:calibration of an infrared spectroscopic technique. Am. Mineral. 71, 15271541.

Nichols, A.R.L., Wysoczanski, R., 2007. Using micro-FTIR spectroscopy to measurevolatile contents in small and unexposed inclusions hosted in olivine crystals.Chem. Geol. 242, 371384.

Okamura, S., Nakamura, M., Nakano, T., Uesugi, K., Tsuchiyama, A., 2010. Sheardeformation experiments on vesicular rhyolite: implications for brittle fracturing,degassing, and compaction of magmas in volcanic conduits. J. Geophys. Res. 115.

Proussevitch, A.A., Sahagain,D.L.,Kutolin, V.,1993.The stabilityof foamsin silicic melts.J. Volcanol. Geoth. Res. 59, 161178.

Quane, S., Russell, J.K., Freidlander, B., 2009. Timescales of compaction in volcanicsystems. Geology 37, 471474.

Rosseel, J.-P., White, J.D.L., Houghton, B.F., 2006. Complex bombs of phreatomagmaticeruptions: role of agglomeration and welding in vents of the 1886 Rotomahanaeruption, Tarawera, New Zealand. J. Geophys. Res. 111, 124.

Rust, A.C., Cashman, K.V., 2004. Permeability of vesicular silicic magma: inertial andhysteresis effects. Earth Planet. Sci. Lett. 228, 93107.

Rust, A.C., Cashman, K.V., Wallace, P., 2004. Magma degassing buffered by vapor flowthrough brecciated conduit margins. Geology 32, 349352.

Saar, M.O., Manga, M., 1999. Permeabilityporosity relationship in vesicular basalts.Geophys. Res. Lett. 26, 111114.

Sable, J.E., Houghton, B.F., Wilson, C.J.N., Carey, R.J., 2006. Complex proximalsedimentation from Plinian plumes: the example of Tarawera 1886. Bull. Volcanol.69, 89103.

Sable, J.E., Houghton, B.F., Wilson, C.J.N., Carey, R., 2009. Eruption mechanisms duringthe climax of the Tarawera 1886 basaltic Plinian inferred from microtexturalcharacteristics of deposits. In: Thordarson, T., Larsen, G., Rowland, S.K., Self, S.,Hoskuldsson, A. (Eds.), Studies in Volcanology: The Legacy of George Walker, pp.129154.

Sahetapy-Engel, S., Harris, A., 2009. Thermal structure and heat loss at the summitcrater of an active lava dome. Bull. Volc. 71, 1528.

Shane, P., Ingram, N., 2002. D values of hydrated volcanic glass: a potential record ofancient meteoric water and climate in New Zealand. NZ J. Geol. Geophys. 45,453459.

Shane, P., Smith, V.C., Nairn, I., 2008. Millennial timescale resolution of rhyolite magmarecharge at Tarawera volcano: insights from quartz chemistry and melt inclusions.Contrib. Mineral. Petrol. 156, 397411.

Smith, R., Sammonds, P.R., Kilburn, C.R.J., 2009. Fracturing of volcanic systems.Experimental insights into pre-eruptive conditions. Earth Planet. Sci. Lett. 280,211219.

Sparks, R.S.J., 2003. Dynamics of degassing. Geol. Soc. Sp. Pub. 213, 522.Stasiuk, M.V., Barclay, J., Carroll, M.R., Jaupart, C., Sparks, R.S.J., 1996. Degassing during

magma ascent in the Mule Creek vent (USA). Bull. Volcanol. 58, 117

130.Stolper, E., 1982. Water in silicate glasses: an infrared spectroscopic study. Contrib.Mineral. Petrol. 81, 117.

Stolper, E., 1989. Temperature dependence of the speciation of water in rhyolitic meltsand glasses. Am. Min. 74, 12471257.

Takeuchi, S., Nakashima, S., Tomiya, A., 2008. Permeability measurements of natural andexperimental volcanic materialswith a simple permeameter:toward an understandingof magmatic degassing processes. J. Volcanol. Geoth. Res. 177, 329339.

Takeuchi, S., Tomiya, A., Shinohara, H., 2009. Degassing conditions for permeable silicicmagmas; implications from decompression experiments with constant rates. EarthPlanet. Sci. Lett. 283, 101110.

Tuffen, H., Dingwell, D.B., Pinkerton, H., 2003. Repeated fracture and healing of silicicmagma generate flow banding and earthquakes? Geology 31, 10891092.

Tuffen, H., Smith, R., Sammonds, P.R., 2008. Evidence for seismogenic fracture of silicicmagma. Nature 453, 511514.

Tuffen, H., Owen, J., Denton, J., 2010. Magma degassing during subglacial eruptions andits use to reconstruct palaeo-ice thicknesses. Earth Sci. Revs. 99, 118.

Voight, B., Sparks, R.S.J., Miller, A.D., Stewart, R.C., Hoblitt, R.P., Clarke, A., Ewart, J.,Aspinall, W.P., Baptie, B., Calder, E.S., Cole, P., Druitt, T.H., Hartford, C., Herd, R.A.,

Jackson,P., Lejeune, A.M., Lockhart,A.B.,Loughlin, S.C., Luckett, L.,Lynch, R.,Norton,G.E., Robertson, R., Watson, I.M., Watts, R., Young, S.R., 1999. Instability and cyclicactivity at Soufriere Hills Volcano, Montserrat, British West Indies. Science 283,11381142.

Webb, S.L., Dingwell, D.B., 1990. The onset of non-Newtonian rheology of silicate melts.A fiber elongation study. Phys. Chem. Minerals 17, 125132.

Westrich, H.R., Eichelberger, J.C., 1994. Gas transport and bubble collapse in rhyoliticmagma: an experimental approach. Bull. Volcanol. 56, 447458.

Wooster, M.J., Kaneko, T., 1997. Satellite thermal analyses oflava dome effusion rates atUnzen Volcano, Japan. J. Geophys. Res. 103 (B9), 20,93520,947.

Wright, H.M.N., Cashman, K.V., Rosi, M., Cioni, R., 2007. Breadcrust bombs as indicatorsof Vulcanian eruption dynamics at Guagua Puchincha volcano, Ecuador. Bull.Volcanol. 69, 281300.

Wright, H.M.N., Cashman, K.V., Gottesfeld, E.H., Roberts, J.J., 2009. Pore structure ofvolcanic clasts: measurements of permeability and electrical conductivity. EarthPlanet. Sci. Lett. 280, 93104.

Wysoczanski, R., Tani, K., 2006. Spectroscopic FTIR imaging of water species in silicicvolcanic glasses and melt inclusions: an example from the Izu-Bonin arc. J.Volcanol. Geoth. Res. 156, 302314.

Yamashita,S., 1999.Experimental study of the effect of temperature on watersolubilityin natural rhyolite melt to 100 MPa. J. Pet. 40, 14971507.

Yoshimura,S., Nakamura,M., 2008. Diffusive dehydrationand bubble resorption duringopen-system degassing of rhyolitic melts. J. Volcanol. Geoth. Res. 178, 7280.

Yoshimura S., Nakamura M., in press, Fracture healing in a magma: an experimentalapproach and implications for volcanic seismicity and degassing. J. Geophys. Res.doi:10.1029/2009JB000834.

Zhang, Y., Belcher, R., Ihinger, P.D., Wang, L., Xu, Z., Newman, S., 1997. New calibrationof infrared measurement of dissolved water in rhyolitic glasses. Geochim.Cosmochim. Acta 61, 30893100.

Zhang, Y., Xu, Z., Zhu, M., Wang, H., 2007. Silicate melt properties and volcaniceruptions. Revs. Geophys. 45, 127.


Please cite this article as: Kennedy, B.M., et al., Time-and temperature-dependent conduit wall porosity: A key control on degassing and

b.m. kennedy et al- time-and temperature-dependent conduit wall porosity: a key control on degassing...

Documents