Correlation of Magma Evolution and Geophysical Monitoring during ...

Correlation of Magma Evolution andGeophysical Monitoring during the 2011^2012 ElHierro (Canary Islands) Submarine Eruption

JOAN MARTI¤ 1,2*, ANTONIO CASTRO2, CARMEN RODRI¤ GUEZ2,FIDEL COSTA3, SANDRA CARRASQUILLA4, ROCI¤ O PEDREIRA5 ANDXAVIER BOLOS1

1GRUPO DE VOLCANOLOGI¤ A, SIMGEO (UB-CSIC), INSTITUTO DE CIENCIAS DE LA TIERRRA JAUME ALMERA, CSIC,

08028 BARCELONA, SPAIN2UNIDAD ASOCIADA DE PETROLOGI¤ A EXPERIMENTAL, CSIC-UNIVERSIDAD DE HUELVA, FACULTAD DE CIENCIAS

EXPERIMENTALES, CAMPUS DEL CARMEN, 21071 HUELVA, SPAIN3EARTH OBSERVATORY OF SINGAPORE, NANYANG TECHNOLOGICAL UNIVERSITY, SINGAPORE 639798, SINGAPORE4DEPARTAMENTO DE MINERALOGI¤ A Y PETROLOGI¤ A, UNIVERSIDAD DE GRANADA, FUENTENUEVA, GRANADA, SPAIN5DEPARTAMENTO DE GEOLOGI¤ A, UNIVERSIDAD DE HUELVA, CAMPUS DEL CARMEN, 21071 HUELVA, SPAIN

RECEIVED JULY 12, 2012; ACCEPTED FEBRUARY 21, 2013ADVANCE ACCESS PUBLICATION APRIL 26, 2013

The application of petrography, mineral chemistry, geochemistry, and

experimental petrology, including mineral^melt thermodynamic and

diffusion modelling, on quenched basanitic magma samples from

the recent (2011^2012) submarine eruption of El Hierro (Canary

Islands) has permitted the identification of major physico-chemical

variations prior to and during magma eruption that correlate in

time with monitored geophysical changes. After nearly 3 months of

seismic unrest the eruption of El Hierro started on October 10, 2011

and ended by late February 2012.We studied 10 lava balloons and

pyroclastic fragments collected floating on the sea surface between

October 15 and lateJanuary. Based on petrological and geophysical

data we distinguish two main eruptive episodes. Magma erupted

from the beginning of the eruption until late November 2011 was an

evolved basanite (�5 wt%MgO), changing to more primitive com-

positions (�8^9 wt %MgO) with time, thus suggesting extraction

from a compositionally zoned magma system. Experimental data

and mineral^melt thermodynamic modelling indicate that the

erupted magma equilibrated at a pressure of about 400MPa, which

corresponds to a depth of 12^15 km. This depth is consistent with

the location of the crust^mantle discontinuity beneath El Hierro

and with the hypocentral location of seismicity during the unrest

episode. Preliminary modelling of the olivine chemical zoning of crys-

tals erupted in this first episode suggests that the time scale for bas-

anite fractionation and magma replenishment in the shallower

reservoir was of the order of a few months.This is within the same

time frame as the duration of the unrest episode preceding the erup-

tion. The first eruption episode coincided with intense seismicity

mostly located north of the island, first at a depth of 20^25 km and

a few days later also at 10^15 km depth, with strong seismic tremor

beneath the vent site. An abrupt change in magma composition and

crystal content was observed at the end of November 2011. After

that, more primitive and less viscous magma erupted contemporan-

eously with a change in the frequency and intensity of seismic

events. During this period, seismicity was mostly north of the island

at depths of 10^15 km. At the same time the tremor intensity at the

eruption site significantly dropped. This marked the onset of the

second eruption episode, which is correlated with an intrusion of

fresh, more primitive magma into the shallow magmatic system that

raised the temperature of the remaining magma. Experiments reveal

that subtle changes in temperature of about 508C (i.e. 1100^

11508C) were enough to produce large changes in the crystal content

(10^60 wt %). This non-linear behaviour between crystal content

*Corresponding author. Telephone: þ34-934095410. Fax:þ34-934110012. E-mail: [email protected]

� The Author 2013. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 54 NUMBER 7 PAGES1349^1373 2013 doi:10.1093/petrology/egt014

Downloaded from https://academic.oup.com/petrology/article-abstract/54/7/1349/1484801by gueston 11 April 2018

and temperature had important effects on magma dynamics during

transport and cooling. Our results suggest the existence of two inter-

connected mafic magma reservoirs during the El Hierro eruption,

which agrees with the pattern shown by the seismicity. Stress re-

adjustments of the plumbing system, caused by decompression

during the eruption, influenced the thermodynamic evolution of the

erupting magma and facilitated the intrusion of the deeper magma

into the shallow reservoir, thus forcing a change in its rheological

characteristics and eruption dynamics.

KEY WORDS: El Hierro eruption; unrest; petrology; seismicity; experi-

mental petrology; diffusion modelling

I NTRODUCTIONMovement and accumulation of magma beneath dormantvolcanoes can be monitored through characteristic seismicsignals and accurate measurements of ground deformation,which are used to forecast eventual volcanic eruptions(Sparks, 2003). The interpretation of these geophysicaland geodetic signals can be significantly improved byknowing the essential physical properties of the magmassuch as viscosity, vesicularity, amount and composition ofdissolved volatiles, crystal content and temperature,among others (Blundy & Cashman, 2008; Lavalle¤ e et al.,2008; Giordano et al., 2010; Longo et al., 2012). These canbe constrained by means of experimental phase equilibriastudies of volcanic products (Martel et al., 1998; Pichavantet al., 2002, 2009). Geophysical monitoring is a fundamentaltool to understand deep-seated processes related tomagma migration and magma chamber recharge that nor-mally precede volcanic eruptions (Sigmundsson et al.,2010). Characterization of magma physics using phaseequilibria and the chemical composition of volcanic prod-ucts offers an excellent opportunity to establish cause^effect relationships between magma dynamics andgeophysical signals (e.g. Kahl et al., 2011; Saunderset al., 2012).Here we present a strong correlation between monitored

geophysical signals, changes in magma physics, and petro-logical properties of products of the 2011^2012 El Hierrosubmarine eruption in the Canary Islands. We have ana-lysed the whole-rock and mineral compositions of magmafragments collected at different times during the eruption.Phase equilibria were determined by using mineral^meltthermometry, thermodynamic modelling and laboratoryexperiments. First-order constraints on the magma resi-dence time were calculated based on olivine diffusion mod-elling. Our results show that a direct (time) correlationbetween magma evolution and geophysical signals can po-tentially be established and used to interpret unrest anderuption dynamics.

GEOLOGICAL SETT INGThe Canary Islands form a chain of seven main volcanicislands located off the northwestern continental shelf ofAfrica, constructed on a Jurassic age oceanic crust(Schmincke & Sumita, 1998; Fig. 1). Several contrastingmodels have been proposed to explain their origin. Theseinclude a hotspot origin (Schmincke, 1982; Hoernle &Schmincke, 1993; Carracedo et al., 1998), a propagatingfracture from the Atlas Mountains (Le Pichon & Fox,1971; Anguita & Herna¤ n, 1975), and mantle decompressionmelting associated with uplift of tectonic blocks (Aran‹ a &Ortiz, 1991). A unifying model has been proposed byAnguita & Herna¤ n (2000), who considered the existenceof a residue of a fossil mantle plume under North Africa,the Canary Islands, and western and central Europedefined through seismic tomography (Hoernle et al., 1995)and its interaction with tectonic processes that affect thelithosphere in this region.El Hierro is the youngest of the Canary Islands, with the

oldest exposed subaerial rocks having an age of 1·12 Ma(Guillou et al., 1996); it is located at the southwestern endof the island chain (Fig. 1). El Hierro lavas range from alka-line basanites and picro-basalt to highly alkaline basaniteand picrite to phonolite (Stroncik et al., 2009; Day et al.,2010). From petrological studies (Neumann et al., 1999;Klu« gel et al., 2005), it is inferred that mafic eruptions inthe Canary Islands are fed by single batches of astheno-sphere-derived magmas. Stroncik et al. (2009) havesuggested that the plumbing system beneath El Hierro ischaracterized by multi-stage ascent of magmas. Geophysi-cal studies (Bosshard & MacFarlane, 1970; Watts, 1994) in-dicate that the base of the crust below El Hierro is at adepth of 13^15 km.

OVERV IEW OF THE 2011 ^ 2 012SUBMAR INE ERUPT IONThe eruption began on October 10, 2011 at 4:30 a.m., at adepth of 900m below sea level (m bsl) on a north^south-trending fissure located along the southern rift zone of theisland (Fig. 1). During the first 4 days the locus of the erup-tion migrated north, establishing a permanent central con-duit and a main vent at 300m depth, around which avolcanic edifice was constructed by accumulation of theerupted material. When the eruption ceased in lateFebruary 2012, the top of this volcanic cone was located at88m bsl (www.ieo.es). During the eruption, highly vesicu-lated lava balloons (see Keuppers et al., 2012) up to 1macross and small pyroclastic fragments sporadically ap-peared at the sea surface, where they remained floatingfor minutes to a few hours before sinking.The eruption was preceded by an episode of unrest ac-

tivity that began on July 17, 2011 and was characterized byheightened seismicity, surface deformation, and gas

JOURNAL OF PETROLOGY VOLUME 54 NUMBER 7 JULY 2013

1350Downloaded from https://academic.oup.com/petrology/article-abstract/54/7/1349/1484801by gueston 11 April 2018

www.ieo.es

emissions (Lo¤ pez et al., 2012). During the first 2 months ofunrest, seismic activity was concentrated at the north ofthe island, with most of the hypocentres located at adepth of 12^15 km (Fig. 2). During the second week ofSeptember the location of seismicity migrated towards thesouth by more than 14 km along a north^south lineartrend, and by the end of September shifted to the east byabout 5 km, always at the same depth (Mart|¤ et al., 2013),until the onset of the eruption, which was preceded by avery shallow seismic swarm (Lo¤ pez et al., 2012). The erup-tion was characterized by continuous tremor of variable in-tensity, originating from the vent site (Fig. 3), and by

tectonic and volcano-tectonic seismicity, although this ac-tivity was confined to the region north of the island (for acomplete catalogue of the seismic activity see www.ign.es).Although there is significant uncertainty in locating the

earthquake sources (see Lo¤ pez et al., 2012; Mart|¤ et al.,2013), the variations in the number and location of seismicevents, as well as in tremor intensity, clearly define two dis-tinct episodes during the eruption. A period of low seismicactivity during the first week of the eruption was followedby intense seismicity, first at 20^25 km depth and then at10^15 km (Fig. 2). During this first episode tremor intensitywas high and oscillating (Fig. 3; Mart|¤ et al., 2013). By

El Golfo

Las Playas

El Julan

0 10km

N

Recent eruptions

Tiñor volcano

El Golfo and Las Playas volcanoes

Rift eruptions

Post-collapse deposits of El Golfo

Upper Pleistocene and Holocene

Lower and middle Pleistocene

Submarine eruption 2011-2012

Rift zones

10/10/2011 Eruption

Fig. 1. Simplified geological map of El Hierro modified from Ancochea et al. (2004).

MARTI¤ et al. EL HIERRO SUBMARINE ERUPTION


www.ign.es

Fig. 2. Epicentral and hypocentral location of seismic events recorded (a) from 17 July to 10 October 2011 (unrest episode), and (b) from 10October 2011 to 5 March 2012 (eruptive episode). Data from IGN Seismic Catalogue (www.ign.es, and Mart|¤ et al., 2013).



www.ign.es

November 22, 2011, a second period started in which seis-mic activity was lower and hypocentres were all locatedat a depth of 10^15 km (Fig. 2). The intensity of the tremorsignal decreased considerably compared with the previousepisode (Fig. 3). These conditions were maintained untilthe end of the eruption in late February 2012. Most of thetectonic and volcano-tectonic seismicity recorded duringthe eruption was located at the north of the island, witha similar epicentral location to that during the first2 months of unrest. The reader is referred to the papers byLo¤ pez et al. (2012) and Mart|¤ et al. (2013) for more informa-tion on the geophysical data recorded during El Hierrounrest and eruption episodes.

METHODSSampling and analytical^experimentaltechniquesSamples of the floating lava balloons and pyroclastic frag-ments were collected for petrological study. The lavaballoons were up to 1m in diameter and consisted of angas-filled cavity surrounded by a thin shell (a few centi-metres thick) of quenched basanite, whereas the pyroclas-tic fragments are highly vesicular scoria of fine lapilli andash size. Floating fragments of quenched basanite magmaappeared on different days during the eruption. Althoughit is not possible to establish an exact correspondence be-tween eruption time and appearance at the sea surface ofthese magma fragments, their origin (see Keuppers et al.,2012) ensures that they represent different times of the erup-tion, so that their study is important to characterize timevariations of the main petrological parameters. The ElHierro eruption produced basanite as well as more evolvedtrachytic and rhyolitic compositions (e.g. Meletlidis et al.,2012; Sigmarsson et al., 2013); here we concentrate only onthe basanites.Approximately 10 g of each sample were ground for

whole-rock geochemistry. Major and trace elements were

determined by X-ray fluorescence (XRF) and inductivelycoupled plasma mass spectrometry (ICP-MS), respectively,at the University of Granada, following the standard pro-cedures described by Baedecker (1987). Precision for majorelements is better than 1%. Analysis of trace elements wasperformed following the method described by Bea (1996);the precision was approximately 2% and 5% error on con-centrations of 50 and 5 ppm, respectively. Chips of basan-ite samples were mounted in epoxy resin, polished, andsubsequently analysed for the major element compositionsof mineral phases and glass using a JEOL JXA-8200Superprobe at the University of Huelva. A combination ofsilicates and oxides were used for calibration. A defocusedbeam of 20 mm diameter was used for glass to minimizeNa migration. The same conditions were applied to theanalysis of the experimental charges. In low melt fractionexperiments, a smaller beam (1 mm) was used to avoid con-tamination with X-rays from surrounding phases. Phasemodes for natural and experimental samples are reportedin wt % and were calculated by least-squares mass balanceusing the method of Hermann & Berry (2002). Viscositycalculations were made from the major element datausing the method of Caricchi et al. (2007).The sample selected as the starting material for the

high-pressure experiments was basanite sample HB-6 (seebelow). Experiments were conducted in a Boyd^Englandtype piston-cylinder apparatus at the University of Huelva(Spain). A 12·5mm (half-inch) diameter talc^Pyrex^peri-clase cell was used in all experiments. Gold^palladium(Au70Pd30) capsules of 3mm diameter and 0·15mm wallthickness were filled, sealed by pressure folding, and intro-duced into MgO pressure containers. Temperatures weremeasured and controlled with Pt100^Pt87Rh13 thermo-couples wired to Eurotherm 808 controllers. Oil pressureswere measured with electronic DRUCK PTX 1400 pres-sure transmitters, connected to OMRON E5CK control-lers. Pressure was corrected manually and maintainedwithin a narrow range of �5 bar oil pressure, equivalent

Fig. 3. Time evolution of the 1h average amplitude module (normalized) of the continuous seismic signal filtered from 1 to 10Hz (from Mart|¤et al., 2013).



to �250 bar on the sample. Heating proceeded at the max-imum rate allowed by the system of 100K min^1. The hot-piston-out technique was used to minimize friction effectsof the talc^Pyrex^periclase cell (McDade et al., 2002).After the desired run time, the experiments were quenchedat a rate of more than 100K s^1 by switching the poweroff. Fast cooling was crucial to avoid the formation of crys-tals during quenching. After quenching, capsules wereexamined under a binocular microscope for tears andchecked for proximity of the thermocouple during the run.Oxygen fugacity was intrinsically buffered by the

Fe3þ/Fe2þ ratio and oxide assemblage of the natural start-ing material, which is close to the Ni^NiO (NNO) bufferaccording to previous determinations for basanite rocksfrom the same eruptive environment (Ablay et al., 1995).This value of oxygen fugacity was tested with the MELTScode (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998) byreproducing mineral assemblages and crystallization se-quences, paying special attention to the Fe-oxides.Furthermore, because metal capsules are permeable toHþ at high temperatures we added a mixture of suitableproportions of Ni and NiO to the MgO pressure mediumsurrounding the capsule with the aim of fixing externallythe fO2 at the NNO buffer. Each experiment was previ-ously equilibrated at 12008C for 1h and then cooled to therun equilibrium temperature. Several sets of runs were per-formed at 0·4GPa and variable temperature from 1050 to12008C. A complete list of run conditions is given inTable 1.

Diffusion modelling techniquesRim to rim electron microprobe traverses were performedacross three olivine crystals in samples HB-2 and HB-3.We measured concentrations of Fe, Mg, Ca, Ni, Mn, Al,and Ti, with a spacing between analyses of 10 mm. Weused Fick’s second law and the concentration-dependentdiffusion coefficients of Fe^Mg and Mn (Dohmen &Chakraborty, 2007), and Ni (Petry et al., 2004) parallel tothe [001] axis to model the data. For Ca we used the dataof Coogan et al. [2005; see review by Chakraborty (2010)for more details on diffusion data in olivine]. The fits weredone visually, using finite differences to solve the equation.Errors in the fit of the data include those related to analyt-ical uncertainty, the initial and boundary conditions onthe pre-eruptive temperature, oxygen fugacity and crystalaxis, and on the diffusion coefficient.Variation of the initialconditions and scatter of the data yield time of diffusionerrors of the order of �50%. Temperature and fO2 uncer-tainties in this study were limited because they were con-strained by experimental data. We used diffusion parallelto the [001] axis which is about six times faster than thatparallel to the [100] or [010] axes; thus, in principle, the dif-fusion times could be up to a factor of six longer. Theerror on the diffusion coefficients is typically the largestone. However, we modelled Fe^Mg, Mn, Ca, and Niusing four independently determined diffusion coefficients

and found times for all elements that were within �50%.This shows that the error on the diffusion coefficient forthese elements and at these conditions should not bemuch larger than the error in the time determinationsthat we obtained (Costa & Dungan, 2005; Costa et al.,2008).

RESULTSWhole-rock compositionsWhole-rock chemical compositions are given inTable 2. Allsamples are basanite in composition, similar to older ElHierro eruptions (Stroncik et al., 2009) and those of theother Canary Islands (Lustrino & Wilson, 2007) (Fig. 4a).They follow a differentiation trend similar to that observedduring previous eruptive events on El Hierro over aperiod of more than 1Myr. The basanites are strongly ves-iculated and contain scarce phenocrysts (42mm diameter)and microphenocrsyts (52mm diameter) of chromite(Chr), ulvo« spinel (Usp), olivine (Ol), clinopyroxene (Cpx)and occasionally plagioclase (Pl). Geochemical variationdiagrams (Fig. 4) show that the erupted magmas werevariably fractionated at shallow depths, within the stabilityfield of olivine. Inter-element variation diagrams showclear evidence of differentiation by crystal fractionation.Refractory-to-incompatible (R/I) element variation dia-grams (Fig. 4e and g) (Maaloe, 1985) indicate a dominantprocess of fractional crystallization rather than one con-trolled by variable degrees of partial melting.The relation-ships between MgO vs Cr, and Ni vs Cr (Fig. 4f and h)show variation trends from more primitive samples(MgO48wt %), in which only chromite is involved inthe fractionation process, to more evolved samples(MgO� 5^7wt %), in which clinopyroxene andolivine are the dominant fractionating minerals. Theerupted products define a continuous differentiationtrend in which olivine and Ti-rich magnetite (ulvo« spinel)played the most important role. In contrast, the nearlyconstant MgO content of the glasses (Fig. 5b) suggeststhat temperature was nearly uniform during the eruptiveprocess

Mineral assemblagesImportant changes in the mineralogy of the erupted prod-ucts with time are observed. A prominent feature is thepresence of chromite as a liquidus phase, predating thecrystallization of olivine and clinopyroxene. The stabilityof chromite prior to olivine implies highly oxidizing condi-tions, close to the NNO buffer during crystallization.Thermodynamic modelling with the MELTS code(Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998) requiresoxidizing conditions at the NNO buffer to reproduce thesame crystallization sequence. A general feature is the dis-equilibrium between olivine and coexisting glass in termsof Fe/Mg ratio (Fig. 6). In all cases, the melt (glass) is



Table1:

Experimentalconditions,modalabundancesandmicroprobeanalysesofexperimentalruns

Runno.:

ACHB-9

ACHB-10C

ACHB-7

ACHB-8

Starting

material:

HB-6þ8%

olivine

HB-6þ8%

olivine

HB-6þ8%

olivine

HB-6þ8%

olivine

P(G

Pa):

0·4

0·4

0·4

0·4

T(8C):

1200

1150

1100

1050

fO2buffer:

NNO

NNO

NNO

NNO

Duration(h):

34

44

Modal

(%):

Glass

(100)

SD

Glass

(92)

SD

Ol(7)

Chr-TiM

t(1)

Glass

(47)

SD

Cpx(32)

Ol(15)

Pl(55)

Chr-TiM

t(3·5)

Glass

(40)

SD

Cpx(42·7)

Chr-TiM

t(11)

Pl(55)

wt%

SiO

244·28

0·23

43·00

0·14

39·54

0·00

49·54

0·97

41·21

38·83

56·74

0·00

55·48

0·08

43·50

0·00

56·92

TiO

24·78

0·26

4·90

0·12

0·07

9·94

3·18

0·31

5·85

0·13

0·00

16·26

1·70

0·2

4·52

14·64

0·00

Al 2O3

14·46

0·20

14·72

0·08

0·00

13·21

18·57

0·26

10·46

0·00

26·39

7·61

21·67

0·17

8·67

4·72

26·92

FeO

6·77

0·32

10·70

0·10

16·03

49·40

6·82

0·46

8·83

18·46

0·00

62·02

3·47

0·12

9·06

67·31

0·00

MgO

9·00

0·33

6·39

0·20

43·51

10·71

3·20

0·28

10·00

41·16

0·00

7·41

1·08

0·06

11·10

6·62

0·00

MnO

0·17

0·03

0·17

0·02

0·23

0·22

0·13

0·05

0·17

0·32

0·00

0·58

0·08

0·02

0·13

0·65

0·00

CaO

10·96

0·12

11·26

0·07

0·44

0·36

5·85

0·61

21·55

0·40

8·07

0·49

1·74

0·04

21·37

0·21

7·87

Na 2O

3·53

0·11

3·69

0·07

0·00

0·00

4·51

0·10

0·79

0·00

6·07

0·00

4·52

0·92

0·64

0·00

6·08

K2O

1·52

0·06

1·58

0·04

0·00

0·00

3·23

0·28

0·00

0·00

0·83

0·00

4·71

0·66

0·00

0·00

0·80

P2O5

0·89

0·07

0·90

0·04

0·16

0·00

1·55

0·09

0·18

0·00

0·09

0·00

0·73

0·11

0·24

0·00

0·00

Cr 2O3

0·11

0·04

0·06

0·08

0·17

12·35

0·03

0·03

0·05

0·04

0·00

1·33

0·03

0·03

0·07

1·19

0·00

NiO

0·04

0·03

0·04

0·05

0·07

0·19

0·04

0·05

0·03

0·12

0·00

0·15

0·00

0·00

0·03

0·12

0·00

Total

16·51

97·4

100·20

96·39

100·00

99·11

99·47

98·18

95·85

95·21

99·34

95·45

98·58

Mg#/mol%

anorthite

0·70

0·52

0·83

0·46

0·67

0·80

40·26

0·36

0·69

39·70

NNO,Ni–NiO

buffer,fixedbyintrinsicbufferingan

dexternal

buffer

withad

ditionofNi–NiO

totheMgO

container.Modal

(%)was

calculatedbymassbalan

ce;

residual

totalerrorisfourforACHB-10C

,threeforACHB-7an

dzero

forACHB-8.Fivean

alyses

werecarriedoutper

sample

foreach

phase.

Totalironis

given

asFeO

.Formass-balan

cecalculationFe3þin

Splwas

determinated

bystoichiometry

andFe3þin

theglass

was

assumed

tobe25%

oftheFeO

totalforNNO

buffer.

Mg#¼molarMgO/(MgOþFeO

).Clinopyroxene–liquid

temperature

and

KFe=Mg

DCpx�

liqwere

calculated

afterPutirka

(2008).Possible

errors

ofmeasuremen

tare

reported

asonestan

darddeviation(SD)forglass.



Table 2: Whole-rock compositions of the erupted lava fragments and bombs

Sample: HB-1 HB-2 HB-3 HB-4 HB-5 HB-6 (Ash) HB-6 HB-8 HB-9 HB-10 HB-11

Date of

emission:

15/10/2011 31/10/2011 31/10/2011 31/10/2011 27/11/2011 5/12/2011 6/12/2011 5/1/2012 18/1/2012 21/1/2012 28/1/2012

Major elements (wt %)

SiO2 44·65 43·05 41·88 43·36 40·13 43·76 42·84 42·47 42·87 42·86 43·02

TiO2 4·64 4·83 4·73 4·59 4·87 4·68 4·74 4·7 4·77 4·78 4·78

Al2O3 13·51 14·09 14·17 13·88 13·60 14·36 13·43 13·56 12·98 13·05 13·03

FeOtot* 12·60 12·70 13·82 13·45 17·07 12·85 13·34 12·17 13·61 13·65 13·62

MgO 5·56 6·91 7·17 7·31 7·60 6·25 7·83 8·66 8·67 8·6 8·6

MnO 0·21 0·37 0·25 0·20 0·19 0·20 0·19 0·42 0·188 0·189 0·189

CaO 10·34 11·06 11·16 10·50 10·53 10·97 10·99 10·49 11·18 11·16 11·29

Na2O 4·42 5·49 4·54 3·97 3·28 4·10 3·89 6·02 3·51 3·55 3·5

K2O 1·14 1·71 1·57 1·55 1·49 1·59 1·46 1·56 1·39 1·4 1·4

P2O5 0·88 n.d. n.d. 0·92 0·78 0·95 0·84 n.d. 0·786 0·79 0·78

LOI n.d. n.d. n.d. 0·25 0·50 0·60 0·43 n.d. 0·74 0·61 0·65

Total 97·95 100·22 99·29 99·98 100·05 100·31 99·98 100·05 100·69 100·64 100·86

Mg# 0·44 0·49 0·48 0·49 0·44 0·46 0·51 0·56 0·53 0·53 0·53

Trace elements (ppm)

Li 9·22 8·07 8·92 8·20 6·04 9·18 8·56 7·27 7·78 8·34 7·30

Be 2·84 2·45 2·67 2·48 2·03 2·81 2·59 2·19 2·31 2·52 2·19

Sc 16·4 20·1 22·33 20·2 19·4 24·4 26·5 24·8 26·1 29·0 24·9

V 261 277 338 302 298 321 342 337 361 399 345

Cr 23·3 123 297 179 154 110 217 354 374 340 329

Co 37·1 82 45·3 40·2 61·9 80·3 81·7 46 49·3 53·6 46·5

Ni 41·2 74·9 96·9 80·2 89·9 59·6 114 141 154 166 142

Cu 71·2 68·3 77·2 69·8 72·9 80·0 87·6 86·9 93·8 103 103

Zn 151 140 146 134 130 154 154 130 136 151 129

Ga 25·9 23·5 25·7 23·5 20·5 26·2 25·7 22·5 23·6 25·8 22·2

Rb 37·6 34·3 35·5 30·7 26·9 35·6 34·5 27·9 28·6 31·4 26·8

Sr 1079 982 993 901 707 1117 1060 831 865 951 823

Y 36·7 33·2 37·1 33·1 24·6 37·9 35·9 30·8 31·9 34·8 30·3

Zr 291 286 402 367 261 410 381 379 371 371 371

Nb 78·1 72·1 72·7 66·7 50·4 80·3 73·9 61·3 63·8 68·4 59·9

Cs 0·64 0·61 0·48 0·41 0·47 0·50 0·49 0·37 0·40 0·44 0·38

Ba 414 376 426 391 331 426 396 359 379 403 355

La 66·9 59·7 62·6 57·3 50·2 65·3 59·8 49·3 52·1 55·2 48·4

Ce 135 121 132 120 104 138 126 104 110 118 103

Pr 16·3 14·7 16·6 15·1 12·9 17·4 15·9 13·2 14 14·9 13·03

Nd 64·9 59·2 69·4 62·6 52·8 71·3 66·1 56·1 59·4 63·7 55·05

Sm 12·5 11·6 13·9 12·4 11·0 14·3 13·5 11·5 12·0 13·1 11·2

Eu 3·83 3·57 4·32 3·99 3·37 4·40 4·13 3·68 3·87 4·16 3·62

Gd 10·2 9·49 11·6 10·6 9·09 11·9 11·2 9·74 10·3 11·1 9·62

Tb 1·37 1·28 1·47 1·36 1·21 1·51 1·44 1·28 1·33 1·42 1·26

Dy 6·99 6·57 8·39 7·58 6·36 8·39 7·78 6·98 7·32 8·01 6·88

Ho 1·25 1·18 1·41 1·27 1·11 1·38 1·31 1·18 1·24 1·33 1·16

Er 3·00 2·73 3·42 2·99 2·57 3·28 3·04 2·78 2·92 3·14 2·70

Tm 0·40 0·37 0·40 0·37 0·34 0·40 0·38 0·35 0·36 0·38 0·34

(continued)



more evolved than predicted by the olivine composition,which implies that some fraction of olivine (5^8wt %, de-pending on the sample) was removed.Crystal habits, modal abundances and microprobe

analyses for the ten analysed samples are given in Figs 7and 8 and Table 3. Ulvo« spinel usually displays skeletalhabit and appears as inclusions in olivine (Fig. 7a, e, hand l). Olivine shows a large variety of shapes and zoningpatterns including skeletal, euhedral normally zoned(decreasing Fo content), virtually unzoned, reverselyzoned, and partially dissolved crystals (Fig. 7a, b, d, h, j, oand p). Euhedral and skeletal olivine is present in mostsamples. In contrast, clinopyroxene is euhedral, withfrequent concentric and patchy zoning and quenching tex-tures (Fig. 7c, f, g, i, k and n). Quenched pyroxenes arefound in several samples, typical of ultra-rapid cooling insubmarine environments (Allen et al., 2008).These quench-ing textures are represented by star-like (Fig. 7g) and her-ringbone crystals (Fig. 7h). Plagioclase is present only asmicrolites (Fig. 7h). Textural analysis and density measure-ments of samples indicate that vesicularity is always pre-sent but in different proportions, nearly double in themore evolved (poorer in MgO) samples (HB-1, HB-2,HB-3 and HB-7) compared with the more primitive(richer in MgO) samples (HB-8 to HB-11). Flow structuresare highlighted by the orientation of plagioclase microlitesaround olivine phenocrysts (Fig.7o) and by the orientationof vesicles. Clinopyroxene phenocrysts may form synneusistextures and appear zoned with concentric and patchingzoning (Fig. 7).

Olivine chemical zoning and time scalesof processesBackscattered electron (BSE) images show that olivine dis-plays a variety of zoning patterns and shapes. Rim com-positions become progressively more Fo-rich with time in

the eruption sequence, consistent with the increase inMg# of the erupted rocks (Fig. 4). Given the variety ofzoning patterns and shapes, a detailed study of olivine tex-tures and compositions is beyond the scope of this paper,and will be the subject of future work. Here we report thefirst results on samples from the early eruption episode toaddress whether crystal zoning patterns can be integratedwith the rest of the seismic and experimental dataset wereport.BSE images of olivine crystals from the early episode

(samples HB-1 to HB-5) appear to have a normal zoningpattern that shows up as dark cores surrounded by brightrims, implying that Fe increases towards the rim.However, three quantitative electron microprobe profilesshow that between the cores and the rims there is a re-versely zoned part of higher Fo and Ni content (Fig. 9).Such reverse zoning patterns can be interpreted as theresult of the arrival of more primitive magma into anevolved reservoir followed by crystal fractionation (e.g.Kahl et al., 2011). Moreover, the traverses also show a var-iety of core plateau compositions from Fo83 to Fo77, whichsuggest that the mafic magmas below El Hierro were vari-ably differentiated before a new mafic magma intrusionarrived from depth.Given enough time and high magmatic temperatures,

the different parts of the olivine crystals will tend to equili-brate owing to volume diffusion. Here we use the approachdescribed by Costa et al. (2008) and Kahl et al. (2011) to es-timate magma residence times from the compositionalzoning of these crystals. The initial conditions were deter-mined by the shape of the profiles and the observationthat the olivine phenocrysts have skeletal rims, indicatinga final fast growth event with constant composition andthus justifying the use of the plateaux for the last step.The maximum concentrations for the initial profile of thereversely zoned area were taken to be the same as or

Table 2: Continued

Sample: HB-1 HB-2 HB-3 HB-4 HB-5 HB-6 (Ash) HB-6 HB-8 HB-9 HB-10 HB-11

Date of

emission:

15/10/2011 31/10/2011 31/10/2011 31/10/2011 27/11/2011 5/12/2011 6/12/2011 5/1/2012 18/1/2012 21/1/2012 28/1/2012

Yb 2·37 2·18 2·31 2·19 1·90 2·42 2·24 1·92 2·11 2·31 1·90

Lu 0·35 0·31 0·32 0·29 0·26 0·32 0·31 0·26 0·28 0·29 0·26

Hf 4·94 5·93 10·1 9·41 7·02 10·6 9·50 8·40 9·00 9·59 8·35

Ta 5·29 5·16 5·59 5·02 4·19 5·91 5·3 4·91 5·12 5·27 4·69

Pb 4·25 3·67 4·28 3·69 3·13 4·79 3·89 3·12 3·38 3·64 3·11

Th 6·68 5·81 5·69 5·14 5·07 5·97 5·60 4·50 4·76 5·06 4·43

U 1·91 1·57 1·60 1·41 1·35 1·70 1·56 1·25 1·35 1·44 1·24

*Total Fe given as FeO.n.d., not determined. Mg#¼molar MgO/(MgOþ FeO).



0.0

0.2

0.4

0.6

0.8

1.0

0 50 100 150 200

Mol

ar M

gO/(M

gO+F

eO)

Ni (ppm)

(c)

0

4

8

12

16

40 45 50 55 60 65 70 75 80

Na 2

O+K

2O

SiO2 (wt%)

(a)(a)

MgO (wt%)

Mol

ar M

gO/(M

gO+F

eO)

HB1

HB2

HB3

HB4

HB5 HB6 ash

HB6 HB8

HB9

HB10

HB11

Legend

Volcanic rocks of the Canary Islands

1

10

100

1000

La Ce

Pr Nd

Sm Eu

Gd Tb

Dy Ho

Er Tm

Yb Lu

(d)

Roc

k/ch

ondr

ite

0

20

40

60

80

100

120

140

160

180

0 1 2 3 4 5 6 7

Ni (

ppm

)

Ta (ppm)

0

20

40

60

80

100

120

140

160

180

0 50 100 150 200 250 300 350 400 450

Ni (

ppm

)

Cr (ppm)

0

1

2

3

4

5

6

7

8

9

10

0 50 100 150 200 250 300 350 400 450

MgO

(wt%

)

Cr (ppm)

Ni (

ppm

)

Nd (ppm)

(e) (f)

(h)(g)

p.m.

p.m.

f.c.

Chromite and Clinopyroxene

Chromite

Oliv

ine

Clinopyroxenef.c.

0,0

0,2

0,4

0,6

0,8

0 2 4 6 8 10 12

Glasses

0

20

40

60

80

100

120

140

160

180

0 50 100 150 200

(b)

Fig. 4. Geochemical variation diagrams showing the characteristic features of the basanites. These are compared with typical volcanic rockcompositions from the Canary Islands. (a) TAS diagram. (b) Molar Mg/(MgþFeO) vs MgO wt %). (c) Molar Mg ratio vs Ni (ppm).(d) Chondrite-normalized REE patterns. Grey dots in (a)^(c) and the shaded area in (d) represent data for Canary Islands basanites compiledfrom the GEOROC database (http://georoc.mpch-mainz.gwdg.de/georoc/). (e, g) R/I geochemical diagrams showing large variations in the re-fractory (R) element Ni, compared with small variations in the incompatible (I) elementsTa and Nd; vector labeled f.c. represents a fractionalcrystallization trend and vector labeled p.m. would be the effect of a partial melting process. Clinopyroxene vector in (f) and (h) represents50% equilibrium crystallization of a parental liquid with the initial composition of Cr¼ 300 ppm using a crystal^melt partition coefficient of17, the mean value between the maximum and minimum experimentally determined values for basanite systems (Adam & Green, 2006).(f) The coupled variation of Ni and Cr is the result of joint fractionation of olivine (Ol) plus chromite (Chr), compatible with mineralogical ob-servations and compositional observations. (h) MgO^Cr diagram indicating the effect of chromite and clinopyroxene crystallization.



http://georoc.mpch-mainz.gwdg.de/georoc/

slightly higher than the measured values. Two out of thethree crystals were amenable for modelling, given the spa-tial resolution and analytical precision of our analyses.The results show that calculated times overlap with 50%error for all elements (Fe^Mg, Ca, Ni, Mn); both crystalsgive values of between 1·5 and 3 months. Moreover, thebest-fit zoning is obtained in two steps: (1) core to reversezone (Fig. 9) gives about 1^2 months; (2) reversed tonormal skeletal rims takes about 1 month. Other initialconditions were tried, but the total diffusion times werealways of about a few months.The first step could be related to intrusion of deep primi-

tive magma into a shallower reservoir that contained vari-ably evolved and differentiating mafic magma. Thesecond step represents olivine skeletal growth and partlyre-equilibration with an evolved liquid that occurredabout a month before eruption (Fig. 9g). The skeletalnature of the crystals and the short times could be asso-ciated with magma movement from the shallow reservoirtowards the surface, and perhaps crystallization and re-equilibration within a dike system. These processes are areflection of the early eruptive episode and not of thesecond replenishment event. Future work on olivine crys-tals of the overall eruption sequence will allow us tobetter track the different processes and confirm, or not,those that we propose here.

Experimental resultsThe intensive variables of the magma prior to eruptionwere determined using different approaches. Pressure wasfixed at 400MPa according to the depth of the earth-quakes associated with the unrest episode that precededmagma eruption (Lo¤ pez et al., 2012). Consequently, weassume that all petrological processes in the first stageor step occurred within a magma system at a depth of12^15 km, the average depth of pre-eruptive seismicity.Because olivine^liquid thermometry cannot be appliedbecause of the general olivine^liquid disequilibrium men-tioned above, we use the clinopyroxene^liquid thermom-eter of Putirka (2008) (Table 3). The presence ofclinopyroxene crystallized in the latest stage duringquenching implies advanced fractionation of the liquid.Consequently, we decided to perform an experimentalstudy to more precisely determine the pre-eruptivemagma temperature, phenocryst assemblage and modalproportions.Several attempts were made to fit the observed pheno-

cryst assemblage of the erupted basanite with the experi-mental mineralogy. Initially, we used the basanite lavaballoon sample HB-6 as the starting material. This wasthe most primitive material available at the time of thestudy, which was made during the eruption and before theeruption of more primitive magma in its last days. Results

0

1

2

3

4

5

6

0 20 40 60 80 100 120

Time (days)

MgO

(wt %

)

0.4

0.5

0.6

0 20 40 60 80 100 120

Time (days)

Mol

ar M

gO/(M

gO+F

eO)

0

0.2

0.4

0.6

0.8

0 20 40 60 80 100 120

Cry

stal

linity

(φ =

mas

s cr

ysta

l fra

ctio

n)

HB1HB2HB3HB4HB5

HB6 ashHB-8HB-9HB-10HB-11

HB6

Legend

Whole rocks

Glasses

Rec

harg

e

Whole rocks

Glasses

(a)

(b)

(c)

(d)

14 Oct. 2011

Rec

harg

e

4

5

6

7

8

9

0 20 40 60 80 100 120

MgO

(wt%

)

1

56

8 910 11

Fig. 5. Time variations in whole-rock compositions of lava fragments. (a, b) Variation of the MgO contents of whole-rocks (a) and glasses (b).(c) Variation of crystallinity with time marking the discontinuity interpreted as due to recharge of the plumbing system. (d) Variation of theMg# [molar MgO/(MgOþFeO)] showing a sharp increase that marks the transition from the first eruptive episode to the second, accompa-nied by magma recharge.



at different pressures (not shown) demonstrated that thissystem was unable to reproduce the characteristic crystal-lization sequence of the studied basanites: chromite !ulvo« spinel ! olivine ! clinopyroxene ! plagioclase.The reason is that sample HB-6, like the rest of the ElHierro basanites, has experienced olivine fractionation(Fig. 6). Experiments with this composition systematicallyproduced the sequence spinel! clinopyroxene! olivine.We reconstructed the original composition by addition ofthe fraction of forsteritic olivine required to fit the oliv-ine^liquid equilibrium at KD(Fe^Mg)ol^liq¼ 0·3�0·03(Roeder & Emslie, 1970; Putirka, 2008). Accordingly, weadded 8wt % of forsterite to HB-6 (Fig. 6). The results wereport here (Table 1; Fig. 10) correspond to experimentswith this doped starting material. These reproduce thecrystallization sequence and mineral compositionsobserved in the erupted basanites fairly well.Representative images of textures and mineral assem-

blages in experimental runs are compared with naturallava fragments (Fig. 10) to constrain the pre-eruptive andsyn-eruptive intensive variables. Experimental results

show a sudden increase in mass crystal fraction (�) (asdetermined from mass balance), from very low valuesof �¼ 0·08 at 11508C (Fig. 10a) to �¼ 0·52 at 11008C(Fig. 10b). At 10508C, the crystal content is too high(�40·6), overpassing the rheological threshold at whichcrystals begin to interact mechanically with each other,and the system starts to behave like a solid framework,thus reducing the ability of the magma to flow (Arzi, 1978;Vigneresse et al., 1996; Caricchi et al., 2007).

DISCUSS IONPetrology and magmatic evolutionOlivine is the most important early phase crystallizing inthe basanite system, and thus we use MgO as a geochem-ical parameter to evaluate magma evolution, and episodesof magma recharge and differentiation. The MgO contentchanges from 5wt % in the first samples collected inOctober 2011 to about 8wt % in the latest ones collectedin January 2012 (Fig. 5). This change was not continuousand two main episodes in magma evolution can be recog-nized. The first episode, extending from the initiation ofthe eruption to late November 2011, was characterized bya continuous increase in MgO and crystal content, whichsuggests the evacuation of a zoned, differentiating magmasystem. The second episode started in early December2011 and was characterized by an increase in crystallinityand MgO with time, but also by the absence of crystalfractionation. The basanites from the second eruptive epi-sode are the most primitive of the whole eruption and donot appear to have evolved in a shallow magma system ashave the previous magmas. The change between the firstand second eruptive episodes is compatible with a rechargeof the shallow magma system with fresh magma comingfrom a deeper reservoir.Of the two possible magma reservoirs detected by

the two maxima in the depth of seismic foci (Mart|¤ et al.,2013) (Fig. 2), we propose that the shallower one(12^15 km), formed during the pre-eruptive stage and wasthe place of magma differentiation. The available min-eral^melt equilibrium relations for the basanite rocks (e.g.the pyroxene^liquid thermobarometer) cannot resolvepressure differences of about 2 kbar, so the presence of adeeper reservoir located at a depth of 20^25 km, whichhas been identified from the location of seismicity duringthe eruption, cannot be confirmed with petrological data.However, it is most likely that deeper magma was injectedseveral times into the shallow reservoir during unrest anderuption as recorded by the reverse zoning patterns in oliv-ine erupted in the first episode.The modeling of olivine compositional profiles shows

that olivine crystallized and was incompletely re-equili-brated in a total time of a few months, during progressiveaccumulation of fresh magma in a shallow magma reser-voir, from the initiation of unrest to its final eruption

70

75

80

85

90

30 40 50 60 70

100xM

g#

Oliv

ine

100xMg# Liquid

Olivine removal

HB1HB2HB3HB4HB5

HB-8HB-9HB-10HB-11

HB6

Legend

Addi

tion

of 5

-8%

oliv

ine

Fig. 6. Mg# of olivine^whole rock (liquid) pairs (the Rhodes dia-gram). Olivine plots to the left of the equilibrium curve (continuousline), which is based on values of KD(Fe^Mg)ol^liq of 0·30� 0·03(Roeder & Emslie, 1970; Putirka, 2008). Open star marks the positionof HB-6, which is the sample used as the starting material for the ex-periments after the addition of 8 wt % forsterite.



Ol

Spl

Glass

100 µm

HB-1

200 µm

Ol

Spl

Cpx

Cpx

Pl

HB-5

HB-6

HB-7

500 µm500 µmCpx

Pl

Spl

Ol

Ol

Cpx

(a)

100 µm

HB-2

Spl

Ol

Glass

Cpx (e)1 mm

HB-2

Ol

(d)HB-3 HB-3

500 µm 200 µm

OlCpx

Spl (b) (c)

100 µm

HB-4

Cpx

Glass

Ol

Spl

(g)

200 µm

Cpx

Pl

Spl

HB-5

(i)

(h)

(f )

(j)

200 µm

Ol

Fig. 7. Back-scattered electron (BSE) images showing representative textural relationships in each sample. Mineral abbreviations after Kretz(1983). (See text for further details.)


1361

(continued)

Downloaded from https://academic.oup.com/petrology/article-abstract/54/7/1349/1484801by gueston 11 April 2018

(Fig. 9). This is in agreement with inferences about the for-mation of the shallow magma reservoir based on hypo-centre locations of seismicity during the unrest episode(Lo¤ pez et al., 2012). Additional evidence in favour of rapiddifferentiation, zonation, and new magma intrusion from

depth in the erupted basanite during the first episode isthe overall compositional disequilibrium between glassand olivine in all samples. Even the skeletal olivine rimshave Mg# higher than expected for equilibrium (Fig. 6),although such disequilibrium could be due to the fast

HB-8

HB-9

100 µm

HB-9

100 µm

100 µm

100 µm

HB-10

HB-11

100 µm

HB-11

100 µm

Ol

Ol

Ol

OlOl

Ol

Ol

Cpx

Cpx

Cpx

Cpx

Cpx

Spl

Spl

Spl

Spl

Spl

Spl

Spl

Pl

Pl Pl

Pl

Glass

Glass

GlassGlass

(k) (l)

(m) (n)

(o) (p)

Fig. 7. Continued.



Glass Ol Chr Usp Cpx Pl

Olivine unzoned

Ol zoned

Ol resorption

Ol skeletal 1

Ol skeletal 2

Ol synneusis

Clinopyroxene unzoned

Cpx zoned

Cpx synneusis

Plagioclase

Sample Modal abundances (wt%)Ol Cpx Pl

HB1

HB2

HB3

HB4

HB5

HB6

Crystalline habits

HB8

HB9

HB10

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

HB11

0 20 40 60 80 100

Legend

Fig. 8. Schematic diagram showing the habits of phenocrysts (Ol, Cpx) and microlites (Pl). Modal abundances of crystals and glass are de-picted in the bars to the right. Modes for natural and experimental samples are wt % of phases calculated by least-squares mass balance usingthe method of Hermann & Berry (2002).



Table3:

Modalabundancesandmicroprobeanalysesfrom

glassandmineralsoferuptedlava

fragments

Sam

ple:

HB-1

HB-2

HB-3

HB-4

Modal

(%):

Glass

(93)

Ol(4)

Usp

(2)

Cpx(1)

Glass

(86)

Ol(6)

Cpx(6)

Usp

(1)

Chr-TiM

t(1)

Glass

(87)

Cpx(5)

Ol(4·5)

Usp

(3)

Chr-TiM

t(0·5)

Glass

(83)

Cpx(11)

Ol(5)

Usp(1)

wt%

SiO

245·07

39·53

0·19

43·31

44·95

38·70

41·16

0·10

0·13

45·30

41·38

37·88

0·06

0·10

46·10

41·57

36·56

0·07

TiO

24·44

0·83

17·19

4·72

4·68

0·08

5·63

17·50

13·37

4·70

5·68

0·09

17·08

14·20

4·69

4·45

0·02

18·41

Al 2O3

14·53

1·46

5·87

9·66

16·31

0·04

9·82

5·43

8·08

14·88

8·81

0·03

5·58

7·21

15·61

8·04

0·03

5·28

FeO

tot*

13·02

18·31

68·70

9·68

11·24

18·61

8·57

64·71

59·83

11·75

9·18

19·89

65·98

61·61

11·57

13· 79

19·11

61·97

MgO

4·99

38·43

7·54

13·04

4·43

41·50

10·82

7·18

7·92

5·06

11·30

42·48

7·75

7·85

4·73

11·78

41·53

7·18

MnO

0·50

0·41

0·30

0·18

0·18

0·29

0·10

0·32

0·32

0·18

0·10

0·26

0·35

0 ·34

0·16

0·10

0·26

0·36

CaO

10·43

1·00

0·20

17·51

11·22

0·33

22·02

0·16

0·03

11·35

22·38

0·27

0·00

0·00

10·01

21·32

0·19

0·00

Na 2O

4·24

0·57

0·00

1·06

4·06

0·01

0·58

0·01

0·01

4·72

0·59

0·00

0·00

0·00

5·11

0·56

0·00

0·00

K2O

1·64

0·20

0·00

0·34

1·78

0·01

0·05

0·03

0·01

1·69

0·02

0·00

0·00

0·00

1·78

0·02

0·00

0·00

P2O5

1·14

0·43

0·00

0·50

1·11

0·06

0·12

0·00

0·02

0·34

0·00

0·00

0·00

0·00

0·25

0·01

0·00

0·00

Cr 2O3

n.d.

n.d.

n.d.

n.d.

0·01

0·04

0·04

0·15

6·87

0·01

0·04

0·00

0·08

5·07

0·01

0·03

0·02

0·20

NiO

n.d.

n.d.

n.d.

n.d.

0·03

0·14

0·01

0·07

0·00

0·02

0·00

0·14

0·15

0·17

0·00

0·00

0·18

0·14

Total

100 ·00

101·17

99·98

99·98

100·00

99·82

98·92

95·67

96·58

100·00

99·46

101·03

97·03

96·54

100·00

101·65

97·90

93·61

Mg#/mol%

anorthite

0·41

0·79

0·71

0·41

0·80

0·69

0·43

0·69

0·79

0·42

0·60

0·79

T(8C)(Cpx–liq)/

KDFe/MgCpx–liq

1206

0·67

1135

0·48

1030

0·22

1080

0·28

Sam

ple:

HB-5

HB-6

HB-8

Modal

%:

Glass

(52)

Cpx(29)

Ol(14)

Pl(2)

Usp(2)

Chr-TiM

t(1)

Glass

(82)

Cpx(7)

Ol(6)

Usp

(2)

Pl(1·5)

Chr-TiM

t(0·5)

Glass

(84)

Ol(8)

Cpx(4)

Pl(2)

Usp

(2)

wt%

SiO

245·37

44·90

39·81

51·55

0·11

0·11

45·23

46·90

39·11

0·07

51·23

0·07

44·50

39·00

44·30

50·88

0·28

TiO

24·75

4·23

0·03

0·23

17·58

14·54

4·93

2·47

0·04

19·04

0·29

17·88

4·81

0·05

3·83

0·23

19·50

Al 2O3

15·78

6·87

0·04

30·01

5·86

7·12

15·39

5·74

0·03

5·76

29·51

6·04

15·90

0·06

7·32

30·03

5·96

FeO

tot*

11·14

8·20

18·74

0·77

63·99

61·63

12·68

8·02

17·03

64·83

0·87

63·10

11·76

18·08

7·91

0·84

61·59

MgO

4·50

12·23

41·37

0·11

7·18

7·52

4·37

13·04

44·00

7·32

0·12

7·07

4·88

42·31

12·56

0·15

8·07

MnO

0·23

0·12

0·30

0·00

0·39

0·33

0·24

0·17

0·24

0·38

0·00

0·37

0·21

0·25

0·09

0·02

0 ·39

CaO

11·00

22·54

0·28

13·36

0·00

0·00

9·43

22·37

0·21

0·00

13·18

0·00

10·40

0·30

22·14

12·71

0·04

Na 2O

5·21

0·48

0·17

3·98

0·00

0·00

5·60

0·53

0·01

0·00

4·18

0·00

4·45

0·02

0·47

3·90

0·06

(continued

)



Table3:

Continued

Sam

ple:

HB-5

HB-6

HB-8

Modal

%:

Glass

(52)

Cpx(29)

Ol(14)

Pl(2)

Usp(2)

Chr-TiM

t(1)

Glass

(82)

Cpx(7)

Ol(6)

Usp

(2)

Pl(1·5)

Chr-TiM

t(0·5)

Glass

(84)

Ol(8)

Cpx(4)

Pl(2)

Usp

(2)

K2O

1·70

0·00

0·00

0·28

0·00

0·00

2·08

0·00

0·00

0·00

0·32

0·00

1·92

0·01

0·04

0·33

0·01

P2O5

0·33

0·00

0·00

0·00

0·00

0·00

0·01

0·00

0·00

0·00

0·00

0·01

1·12

0·03

0·08

0·03

0·00

Cr 2O3

0·00

0·02

0·00

0·00

1·45

5·22

0·02

0·04

0·04

0·55

0·01

2·79

0·02

0·09

0·03

0·03

0·30

NiO

0·00

0·00

0·00

0·00

0·16

0·19

0·01

0·01

0·19

0·11

0·00

0·09

0·03

0·17

0·02

0·01

0·02

Total

100·00

99·56

100·72

100·28

96·71

96·66

100·00

99·29

100·91

98·06

99·70

97·42

100·00

100·35

98·78

99·15

96·21

Mg#

0·42

0·73

0·80

65·00

0·38

0·74

0·82

63·50

0·43

0·81

0·74

64·30

T(8C)(Cpx–liq)/

KdFe/MgCpx–liq

1050

0·41

1142

0·23

1125

0·26

Sam

ple:

HB-9

HB-10

HB-11

Modal

%:

Glass

(75)

Cpx(15)

Ol(7)

Usp

(2)

Chr-TiM

t(1)

Pl(51)

Glass

(76)

Cpx(14)

Ol(7)

Chr-TiM

t(4)

Pl(51)

Glass

(74)

Cpx(15)

Ol(6)

Chr-TiM

t(3)

Pl(1)

wt%

SiO

245·06

45·78

39·31

0·07

0·08

51·26

44·60

45·02

39·28

0·11

51·13

45·12

45·60

39·46

0·09

50·81

TiO

25·15

3·35

0·06

20·58

5·94

0·29

4·86

4·07

0·03

7·52

0·31

4·80

3·63

0·05

14·94

0·24

Al 2O3

15·89

6·75

0·04

6·21

15·02

29·65

16·17

6·74

0·03

14·04

30·17

15·75

6·74

0·05

8·15

30·29

FeO

tot*

11·88

7·26

16·61

62·77

40·11

0·89

11·60

7·91

16·97

45·03

0·85

11·45

7·21

16·56

56·49

0·82

MgO

4·78

12· 94

43·31

7·59

10·70

0·13

4·76

12·49

43·02

9·74

0·12

4·78

12·86

43·81

7·65

0·10

MnO

0·20

0·10

0·24

0·34

0·22

0·00

0·19

0·12

0·21

0·22

0·00

0·18

0·10

0·24

0·35

0·01

CaO

10·31

22·47

0·30

0 ·20

0·02

12·82

10·54

22·28

0·30

0·05

13·34

10·38

22·63

0·28

0·05

13·28

Na 2O

3·67

0·45

0·02

0·00

0·01

3·87

4·17

0·43

0·00

0·01

3·58

4·30

0·44

0·01

0·04

3·61

K2O

1·95

0·01

0·00

0·01

0·00

0·35

2·00

0·01

0·00

0·00

0·32

2·02

0·02

0·00

0·01

0·30

P2O5

1·07

0·03

0·03

0·00

0·00

0·04

1·06

0·06

0·01

0·00

0·02

1·18

0·02

0·02

0·00

0·03

Cr 2O3

0·03

0·12

0·04

0·05

25·82

0·01

0·04

0·01

0·05

19·61

0·02

0·06

0·04

0·00

9·39

0·03

NiO

0·03

0·03

0·20

0·09

0·29

0·01

0·02

0·01

0·21

0·23

0·00

0·00

0·02

0·17

0·11

0·02

Total

100·00

99· 31

100·16

97·90

98·19

99·33

100·00

99·15

100·12

96·56

99·86

100·00

99·31

100·65

97·26

99·51

Mg#/mol%

anorthite

0·42

0·76

0·82

64·60

0·42

0·74

0·82

67·30

0·43

0·76

0·83

67·00

T(8C)(Cpx–liq)/

KDFe/MgCpx–liq

1153

0·26

1119

0·24

1143

0·28

Modal

abundan

ceswerecalculated

bymassbalan

ceusing

minim

um

squares.Fivean

alyses

werecarried

outper

sample

foreach

phase.

Mg#¼molarMgO/

(MgOþFeO

).Clinopyroxene–liquid

temperature

andK

Fe=Mg

DCpx�

liqwerecalculatedafterPutirka(2008).

*Totalirongiven

asFeO

.Formass-balan

cecalculationFe3þin

Splwas

determined

bystoichiometry

andFe3þin

theglass

was

assumed

tobe25%

oftheFeO

total

forNNO

buffer.



80

82

84

86 Mol% Forsterite

0 10 20 30 40 50

0.05

0.10

0.15

0.20

0.25

Wt% NiO

0 10 20 30 40 50

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0 10 20 30 40 50

Wt% MnO

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Wt% CaO

0 10 20 30 40 50

(c) (d)

(e) (f)

Distance (x10 µm)Distance (x10 µm)

C

R Z

S R

3 petS2 petS1 petS

500 µm

500 µm

Mg

(a)(b)

Liquid L0

Equilibrium crystallizationLiquid L

1

Magma replenishmentLiquid L

2

Fractionation

Olivine grew in equilibrium with a liquid (L

0) with

Mg#=0.61

Olivine was entrapped as a xenocryst into a more primitive liquid (L

1) with

Mg#=0.63evolved liquid (L

2) with Mg#=0.54,

formed by in situ fractionation.

(g)Skeletal rim

Eruption

About 30-40 days About 20-60 days Total timebefore eruption:About 1.5 to 3 months

Fig. 9. Compositional profiles and diffusion model fits for olivine phenocrysts in basanite HB-2. (a) Back-scattered electron (BSE) image ofzoned olivine phenocrysts. Red line indicates the location of the composition profiles. (b) X-ray map of Mg distribution. (c^f) Profiles showingvariations in olivine composition vs distance across the crystal. Dots represent point analyses. Black fine line is the initial profile. Pink line isthe best fit (estimated by eye). Three zones are distinguished in the model fit: C, core; R Z, reverse zone; S R, skeletal rim. All diffusionmodels were performed atT¼11508C (temperature according to experiments with 90wt % liquid and only chromite and olivine), fO2 at theNNO buffer, and parallel to [001] axis. (g) Schematic representation of magmatic events revealed by zoning in olivine. The best fit for all elem-ents is obtained in two steps as explained in the main text. Times for Fe^Mg, Ni, and Mn diffusion overlap with each other, and are about 1month for the first step and about 3 weeks for the second step. Times for Ca zoning are also about 1 month for the first step, but almost 2months for the second. This time is longer than for the other elements because the dependence on Fo content of Ca diffusivity is not accountedfor. Another crystal from sample HB-3 is not shown here but shows similar zoning patterns (i.e. reverse followed by normal zoning) but lowerFo content. The diffusion times for Fe^Mg for this other crystal are also about 2 months in total. (See Methods section for more details aboutthe modelling parameters.)



ACHB-8, 0.4 GPa, 1050 °C

200 µm

ACHB-7, 0.4 GPa, 1100 °C

100 µm

ACHB-10c, 0.4 GPa, 1150 °C

(a)

(b)

(c)

Ol (new)

Glass

Ol (old)

Cpx

Glass

Pl

Spl

Spl

Ol

Ol

Ol

Cpx

100 µm

200 µm

Ol

Spl

Cpx

Cpx

Pl

HB-5

(e)

Ol

Ol

Spl

Glass

100 µm

HB-1

(d)

Experiments Rocks

Temperature (°C)

0

0.2

0.4

0.6

0.8

1.0

700 800 900 1000 1100 1200 1300 1400

MELTS model crystallinity curveP=0.4 GPa, NNO, 0.5 wt% water

Experimental curveP=0.4 GPa, NNO, 0.5 wt% water

Critical T range

(f )

Cry

stal

linity

(φ =

mas

s cr

ysta

l fra

ctio

n)

Cpx-liquid equilibria

Fig. 10. Back-scattered electron (BSE) images of experimental charges (a^c) and lava fragments (d, e). (a^c) Representative images ofexperimental runs at 0·4GPa and the NNO buffer at temperatures of 1150, 1100 and 10508C, respectively (Table 1). (d) Typical textural rela-tions of skeletal olivine and ulvo« spinel in vesiculated basanite HB-1, the first erupted lava in October 2011. (e) Textural relations of HB-5 basan-ite marking the end of the first eruptive episode, showing olivine phenocrysts, with abundant spinel (chromite) inclusions, in a microlite-richmatrix with clinopyroxene and plagioclase. (f) Crystallinity vs temperature diagram based on the experimental data for El Hierrobasanites. The interpolated experimental curve is very close to the results from thermodynamic modeling (MELTS code; Ghiorso & Sack,1995; Asimow & Ghiorso, 1998) using the same intensive parameters as the experimental runs. The clinopyroxene^liquid equilibria are notwell constrained within the critical temperature range (critical T range), which is between 1100 and 11508C, and crystallinity (�) between0·08 and 0·53.



growth rates implied by the skeletal shape. Variabledepletion by 5^8wt % forsterite is required to bringolivine^liquid pairs to the equilibrium ratioKD(Fe^Mg)ol^liq¼ 0·30� 0·03 (Putirka, 2008). Olivinefractionation was an important process in the earlyerupted magma (MgO� 5wt %), compared with the lessfractionated samples of the late magma emissions (MgO�8wt %). The observed geochemical variation trends(Fig. 3) can be accounted for by crystal fractionation invol-ving the most important phases present in the system (i.e.olivine, chromite and clinopyroxene). Other processessuch as mixing of magmas and crystal capture (e.g. olivinexenocrysts) may have contributed to a minor extent to theobserved variations.The shift of seismic foci to shallower depths at the end of

the first eruptive episode is coincident with a shift to moreprimitive compositions of the erupted magmas. Crystalfractionation was not operative during the second episode.We suggest that the shallow magma system was exhaustedat this time and that magma was transported directlyfrom the deeper one to the surface.Moreover, it can be inferred from the experimental re-

sults that for temperatures of around 10508C, the rheo-logical threshold that determines the ability of a magmato flow is surpassed, limiting the probability of eruption(Marsh, 1981) because of the large viscosity increase pro-duced by friction between crystals. The implication is thatthe critical transition from fluid magma to crystal-richmush behaviour can occur by decreasing the temperatureby only 508C, as our experimental results show. A directcomparison between the modal mineralogy of the experi-ments and the natural rocks (Fig. 11) indicates that theerupted magma in the 2011^2012 El Hierro eruption wasconstrained to a narrow temperature interval of about708C (from 11808C of HB-1 to 11108C of HB-5), half of thetemperature range predicted by clinopyroxene^liquid ex-change thermometry (Fig. 11, right panel). The suddenchange in viscosity owing to the critical rheological transi-tion over a narrow temperature interval observed at theend of the first eruptive episode may have reduced the mo-bility of magma from the shallow system. This may havealso contributed to the stress readjustment of the plumbingsystem, increasing the internal pressure in the deeper partand providing the excess pressure necessary to cause anew intrusion event.

Relations with seismic signalsThe change in the bulk MgO content of the erupted prod-ucts is a record of magma fractionation and, hence, ofmagma accumulation in a reservoir. Such compositionalchanges may be also related to seismic signals indicatingevacuation and recharge of the magma system. In a simi-lar way, pre- and syn-eruptive variables such as tempera-ture, crystal content and viscosity, determined by thepetrological study, can be compared with geophysical

signals. According to these specific features, it is possibleto identify two main eruptive episodes that correspond tothe changes in monitored seismic signals explained above(Fig. 12).Seismicity during the unrest stage (pre-eruptive stage;

Fig. 12, upper panel) started on July 17, 2011, and reflectsthe emplacement of deeper basanitic magma into a shal-lower reservoir, located at the crust^mantle boundary be-neath the northern side of El Hierro island (Lo¤ pez et al.,2012). The maximum frequency of earthquakes (Fig. 12;grey histogram in lower panel) occurs in the early, pre-eruptive stage during magma accumulation within theoceanic crust, with most hypocenters (Fig. 12; shaded areain the background) located at 10^15 km depth duringmost of the pre-eruptive stage (July and August 2011).Accumulation of a minimum of 0·2 km3 of magma in thisshallow reservoir took about 2 months and preceded thelateral migration of magma towards the SE for nearly20 km until eruption started on October 10, 2011 (Mart|¤et al., 2013). As noted above, the first eruptive episodefrom October 10 to the end of November 2011 is character-ized by strong seismicity located first at a depth of 20^25 km and later also at a depth of 10^15 km. This periodcorresponds to eruption of magma with increasing crystal-linity from less than 5wt % (sample HB-1) to around50wt % crystals (sample HB-5). This is interpreted as theevacuation of the shallow magma reservoir that was al-ready zoned after experiencing some crystal fractionation(Fig. 12; upper panel). The first magma emissions fromthis episode are the most fractionated, as they correspondto the magma accumulated in the upper part of the shal-lower reservoir for almost 3 months, from the beginningof unrest to the onset of the eruption.As eruption progressed, the plumbing system started to

decompress and become subjected to stress readjustments,first in the deeper parts and later in the shallower reser-voir, as indicated by the pattern of tectonic and volcano-tectonic seismicity during this first eruptive episode(Mart|¤ et al., 2013). This episode ended with the emissionof the most crystalline and viscous magma (HB-5) of thewhole eruption. In late November 2011, seismicity changeddrastically in both location and frequency (see Fig. 12).This change in seismicity marks the initiation of thesecond eruptive episode in which less fractionatedmagmas were emitted until the end of the eruption, show-ing a nearly constant degree of crystallinity (Fig. 12). Weconsider that the transition from the first to the seconderuptive episode corresponds to an event of magma re-charge of the shallower reservoir with fresh magmas prob-ably coinciding with the partial collapse of the deeperparts of the plumbing system. As a result of this new inputof fresh magma the remaining magma was reheated, chan-ging its rheology and crystal^melt equilibrium, as indi-cated by the changes in composition and degree of



1200

1175

1150

1125

1100

1075

1050

ACHB-9

ACHB-10C

ACHB-7

ACHB-8

HB-9

HB-1

HB-10

HB-5

Experiments Rocks

9.3

6.5

3.3

1.1

0.67

0.69

4.8

4.6

5.0

0.76

0.74

0.71

1225

1025

Tem

per

atu

re (º

C)

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0.733.7

Chr Usp Ol Cpx Pl

Crystallization sequence

HB-1

HB-5

9.3

Liquid (wt% MgO)

Ol

Chr

5 Usp

Cpx (Mg#)

Pl

0.6Cpx-liquid equilibriatemperature uncertainties

Legend

Fig. 11. Comparison of modal abundances of minerals and liquids between experiments and representative rocks of the El Hierro eruption.Thewhole temperature range recorded by the clinopyroxene^liquid equilibrium at 0·4 GPa is shown. The lowest temperature in the natural rocks,10508C, is underestimated and incompatible with the high crystal fraction (�¼0·6) of the experiment at the same temperature (ACHB-8).According to the experimental phase relations, sample HB-5 with �¼ 0·5 was at equilibrium at a temperature of 11108C. Similarly, the highesttemperature according to the clinopyroxene^liquid equilibria, which is represented by sample HB-1, is overestimated. According to experi-ments, this sample was at equilibrium at 11858C, with �40·9. Consequently, the temperature range recorded by the products of the El Hierroeruption is from 1185 to 11108C, only 758C, half the range predicted by clinopyroxene^liquid equilibrium.These discrepancies between tempera-tures based on clinopyroxene^liquid thermometry (Putirka, 2008) and those based on experimental determinations of phase equilibria are dueto two factors. One is the intrinsic error of Px^liquid determinations based on empirical regression of experimental data. The other, and themost important, is that most samples contain abundant quench crystals of Px that were generated by fast cooling as typically happens in sub-marine eruption. Quench crystals are not part of the mineral^melt syn-eruptive assemblage but they strongly fractionate the liquid, givingrise to disequilibrium partitioning between phenocrysts and liquid (glass).



crystallinity of the samples collected from early December2011. During this second eruptive episode seismicity wasmostly located at a depth of 10^15 km, suggesting the pro-gressive collapse of the shallow part of the plumbingsystem, which would not become completely blocked untilthe end of the eruption in late February 2012. It seemslikely that during most of this eruptive episode magmawas transferred directly to the surface from the deep-seated magma reservoir, with minimum residence or nonein the shallow reservoir, as it is indicated by the

petrological homogeneity of the samples from this episodeand the absence of fractionation.These relationships between major changes in seismicity

and petrology can also be observed by comparison withchanges in the intensity of the tremor signal, which showsa drastic decrease coinciding with the transition betweenthe two eruptive episodes (Fig. 3). The amplitude of vol-canic tremor may be related to the size of the eruption con-duit and vent (Chouet, 1996; Jellinek & Bercovici, 2011)and/or variations in the source (Lahr et al., 1994; McNutt,

Erup

tion

begi

ns 1

0/10

/201

1

July August September October November December January February

400

500

300

100

200

0

Dep

th o

f hyp

ocen

ters

( )

(km

)

0

5

10

15

20

25

HB1

HB2

HB3

HB4

HB5

HB6

HB8

HB9

HB10

HB11

Num

ber of earthquakes ( )

36

18

1

Acc

umul

ated

ene

rgy

(x1

011 Jo

ules

)

Seismic AseismicMagma chamber growing

14 km

Upper mantle

Oceanic crust

2011 2012

Samples

Pre-eruptive stage Eruptive stage

Log

visc

osity

(

Pa

s)

3

0

1

2

Magma fractionationin pre-eruptive

reservoir

Deep magmareservoir ? ?

Magmareservoirdeflation

Exhaustedmagmareservoir

Magmapassesover

0

0.2

0.4

Rech

arge

Cry

stal

frac

tion

(f)

Not at scale

N S

EL HIERRO EL HIERRO EL HIERRO

N S N S

Fig. 12. Diagram showing the time evolution of seismic and petrological features (lower panel) and the magma chamber model (upper panel).Frequency of earthquakes is represented by the grey histogram, the hypocentres by the yellow field, and the accumulated energy by the reddots and line. The earthquake depth component (yellow area) represents the total depth range. Viscosity calculations were made from majorelement data using the method of Caricchi et al. (2007). Geophysical data were taken from the Spanish Geographic Institute (www.ign.es) andMart|¤ et al. (2013).



www.ign.es

2005). Changes in the amplitude of the tremor signalduring the El Hierro eruption (Fig. 3) may be interpretedin terms of changes in the structure of the plumbingsystem, as seemed to occur in late November, but theymay also reflect variations in magma rheology coincidingwith these structural changes. In fact, variations inmagma crystallinity, composition, temperature, volatilecontent and vesicularity were sufficiently important tocause significant changes in magma viscosity (up to fourorders of magnitude) (Figs 5 and 12), which should havecaused significant changes in magma rheology anddynamics.

CONCLUDING REMARKSThe major objective of this study was the reconstruction ofthe magma plumbing system for the El Hierro eruptionbased on temporal variations in petrological features cor-related with seismic signals. From the data collected it isstill difficult to confirm whether the magma reactedmainly to changes in the local stress field or to changes inmagma rheology, or both. On the one hand, significantrheological changes (increase in viscosity owing to the in-crease in crystallinity) in the magma could be caused bysmall variations in temperature (5508C) and in turn havean impact on magma dynamics. This could have induceda pulsating movement of magma from the reservoir to thevent, thus explaining the oscillation observed in thetremor amplitude during the whole eruptive process. Onthe other hand, there is little doubt that changes in thelocal stress field, promoted by the variations in the internalpressure of the magma, controlled its movement in theplumbing system.To answer this question better, the petro-logical study should be extended to the full succession oferupted products, which are by now inaccessible. The re-sults we report are, however, still relevant to show how atime comparison between petrological and geophysical in-dicators during monogenetic eruptions can provide im-portant clues about eruption dynamics and contribute to abetter understanding of the eruption precursors in termsof the geological processes that accompany magma move-ment. This is, in turn, one of the main goals to improveeruption forecasting and, consequently, to reduce volcanicrisk.

ACKNOWLEDGEMENTSThis is the first contribution of the Unidad Asociada dePetrolog|¤a Experimental (CSIC-UHU). Mar|¤a Jose¤Jurado, Carmen Lo¤ pez, Mar|¤a Jose¤ Blanco, and all staff ofthe IGN Volcanological Observatory are thanked fortheir help in collecting the samples. We thank PEVOLCAand IGN for allowing us to use the samples for this study.We also thank three anonymous referees for their con-structive reviews.

FUNDINGThis research has been funded by CSIC, the ECVUELCO project, and MINECO grant CGL2011-16144-E.

REFERENCESAblay, G. J., Ernst, G. G. J., Mart|¤, J. & Sparks, R. S. J. (1995). The 2

ka subplinian eruption of Montan‹ a Blanca, Tenerife. Bulletin of

Volcanology 57, 337^355.Adam, J. & Green, T. (2006). Trace element partitioning between

mica- and amphibole-bearing garnet lherzolite and hydrous basa-nitic melt: 1. Experimental results and the investigation of controlson partitioning behavior. Contributions to Mineralogy and Petrology

152, 1^17.Allen, S. R., Fiske, R. S. & Cashman, K. (2008). Quenching of steam-

charged pumice: Implications for submarine pyroclastic volcanism.Earth and Planetary Science Letters 274, 40^49.

Ancochea, E., Barrera, J. L., Bellido, F., Benito, R., Bra« ndle, J. L.,Cebria¤ , J. M., Coello, J., Cubas, C. R., De la Nuez, J., Go¤ mez, J.A., Herna¤ n, F., Herrera, R., Huertas, M. J., Lo¤ pez-Ruiz, J.,Marti, J., Mun‹ oz, M. & Sagredo, J. (2004). Canarias y el vulca-nismo neo¤ geno peninsular. In: Vera, J. A. (ed.) Geolog|¤ a de Espan‹ a.Madrid: Instituto Geolo¤ gico y Minero de Espan‹ a (IGME) ySociedad Geolo¤ gica de Espan‹ a (SGE), pp. 635^682.

Anguita, F. & Herna¤ n, F. (1975). A propagating fracture model versusa hot-spot origin for the Canary Islands. Earth and Planetary ScienceLetters 27, 11^19.

Anguita, F. & Herna¤ n, F. (2000). The Canary Islands: a unifyingmodel. Journal of Volcanology and Geothermal Research 103, 1^26.

Aran‹ a, V. & Ortiz, R (1991). The Canary Islands: Tectonics,Magmatism and Geodynamic Framework. In: Kampunzu, A. B.& Lubala, R. T. (eds) Magmatism in Extensional Structural Settings,

The Phanerozoic African Plate. Berlin: Springer-Verlag, pp. 209^249.Arzi, A. A. (1978). Critical phenomena in the rheology of partially

melted rocks.Tectonophysics 44, 173^184.Asimow, P. D. & Ghiorso, M. S. (1998). Algorithmic modifications ex-

tending MELTS to calculate subsolidus phase relations. AmericanMineralogist 83, 1127^1131.

Baedecker, P. A. (1987). Methods for geological analysis. US Geological

Survey Bulletin 1770, 187.Bea, F. (1996). Residence of REE,Y,Th and U in granites and crustal

protoliths: implications for the chemistry of crustal melts. Journalof Petrology 37, 521^552.

Blundy, J. & Cashman, K. (2008). Petrologic reconstruction of mag-matic system variables and processes. In: Putirka, K. D. &Tepley, F. J., III (eds) Minerals, Inclusions and Volcanic Processes.

Mineralogical Society of America and Geochemical Society, Reviews in

Mineralogy and Geochemistry 69, 179^239.Bosshard, E. & MacFarlane, D. J. (1970). Crustal structure of the west-

ern Canary Island from seismic refraction and gravity data.Journal of Geophysical Research 75, 4901^4918.

Caricchi, L., Burlini, L., Ulmer, P., Gerya, T., Vassalli, M. &Papale, P. (2007). Non-Newtonian rheology of crystal-bearingmagmas and implications for magma ascent dynamics. Earth and

Planetary Science Letters 264, 402^419.Carracedo, J. C., Day, S., Guillou, H., Rodr|¤guez Badiola, E.,

Canas, J. A. & Pe¤ rez Torrado, J. F. (1998). Hotspot volcanism closeto a passive continental margin: the Canary Islands. Geological

Magazine 135, 591^604.Chakraborty, S. (2010). Diffusion coefficients in olivine, wadsleyite

and ringwoodite. In: Zhang, Y. & Cherniak, D. J. (eds) Diffusion



in Minerals and Melts. Mineralogical Society of America and Geochemical

Society, Reviews in Mineralogy and Geochemistry 72, 603^639.Chouet, B. H. (1996). Long-period volcano seismicity: its source and

use in eruption forecasting. Nature 380, 309^316.Coogan, L., Hain, A., Stahl, A. S. & Chakraborty, S. (2005).

Experimental determination of the diffusion coefficient for calciumin olivine between 9008C and 15008C. Geochimica et Cosmochimica

Acta 69, 3683^3694.Costa, F. & Dungan, M. (2005). Short time scales of magmatic assimi-

lation from diffusion modeling of multiple elements in olivine.Geology 33, 837^840.

Costa, F., Dohmen, R. & Chakraborty, S. (2008). Time scales of mag-matic processes from modeling the zoning patterns of crystals. In:Putirka, K. D. & Tepley, F. J., III (eds) Minerals, Inclusions and

Volcanic Processes. Mineralogical Society of America and Geochemical

Society, Reviews in Mineralogy and Geochemistry 69, 545^594.Day, J. M. D., Pearson, D. G., Macpherson, C. G., Lowry, D. &

Carracedo, J. C. (2010). Evidence for distinct proportions of sub-ducted oceanic crust and lithosphere in HIMU-type mantle be-neath El Hierro and La Palma, Canary Islands. Geochimica et

Cosmochimica Acta 74, 6565^6589.Dohmen, R. & Chakraborty, S. (2007). Fe^Mg diffusion in olivine II:

Point defect chemistry, change of diffusion mechanisms and amodel for calculation of diffusion coefficients in natural olivine.Physics and Chemistry of Minerals 34, 409^430.

Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer inmagmatic processes. IV. A revised and internally consistentthermodynamic model for the interpolation and extrapolation ofliquid^solid equilibria in magmatic systems at elevated tempera-tures and pressures. Contributions to Mineralogy and Petrology 119,197^212.

Giordano, D., Polacci, M., Papale, P. & Caricchi, L. (2010).Rheological control on the dynamics of explosive activity in the2000 summit eruption of Mt. Etna. Solid Earth 1, 61^69.

Guillou, H., Carracedo, J. C., Pe¤ rez-Torrado, F. & Rodr|¤guezBadiola, E. (1996). K^Ar ages and magnetic stratigraphy of a hot-spot-induced, fast grown oceanic island: El Hierro, CanaryIslands. Journal of Volcanology and Geothermal Research 73, 141^155.

Hermann,W. & Berry, R. F. (2002). MINSQça least squares spread-sheet method for calculating mineral proportions from whole rockmajor element analyses. Geochemistry: Exploration, Environment,

Analysis 2, 361^368.Hoernle, K. & Schmicke, H. U. (1993). The role of partial melting in

the 15-Ma geochemical evolution of Gran Canaria: a blob modelfor the Canarian hotspot. Geotechnology 10, 19^32.

Hoernle, K., Zhang,Y. S. & Graham, D. (1995). Seismic and geochem-ical evidence for large-scale mantle upwelling beneath the easternAtlantic and western and central Europe. Nature 374, 34^39.

Jellinek, A. M. & Bercovici, D. (2011). Seismic tremors and magmawagging during explosive volcanism. Nature 470, 522^525.

Kahl, M., Chakraborty, S., Costa, F. & Pompilio, M. (2011). Dynamicplumbing system beneath volcanoes revealed by kinetic modeling,and the connection to monitoring data: An example from Mt.Etna. Earth and Planetary Science Letters 308, 11^22.

Keuppers, U., Nichols, A. R. L., Zanon,V., Potuzak, M. & Pacheco, J.M. R. (2012). Lava balloonsçpeculiar products of basaltic submar-ine eruptions. Bulletin of Volcanology 74, 1379^1393, doi:10.1007/s00445-012-0597-x.

Kluu« gel, A., Hansteen, T. H. & Galipp, K. (2005). Magma storageand underplating beneath Cumbre Vieja volcano, La Palma(Canary Islands). Earth and Planetary Science Letters 236, 211^ 226.

Kretz, R. (1983). Symbols for rock-forming minerals. American

Mineralogist 68, 277^279.

Lahr, J. C., Chouet, B. A., Stephens, C. D., Power, J. A. & Page, R. A.(1994). Earthquake classification, location, and error analysis in avolcanic environment: Implications for the magmatic system ofthe 1989^1990 eruptions at Redoubt Volcano. Journal of Volcanologyand Geothermal Research 62, 137^151.

Lavalle¤ e,Y.,Meredith,P.G.,Dingwell,D.B.,Hess,K.U.,Wassermann, J.,Cordonnier, B., Gerik, A. & Kruhl, J. H. (2008). Seismogenic lavasandexplosive eruption forecasting.Nature 453, 507^510.

Le Pichon, X. & Fox, P. J. (1971). Marginal offsets, fracture zones, andthe early opening of the North Atlantic. Journal of Geophysical

Research 76, 6294^6308.Longo, A., Papale, P., Vassalli, M., Saccorotti, G., Montagna, C.,

Cassioli, A., Giudice, S. & Boschi, E. (2012). Magma convectionand mixing dynamics as a source of ultra-long-period oscillations.Bulletin of Volcanology 74, 873^880.

Lo¤ pez, C., Blanco, M. J., Abella, R. et al. (2012). Monitoring the unrestof El Hierro (Canary Islands) before the onset of the 2011 submar-ine eruption. Geophysical Research Letters 39, L13303, doi:10.1029/2012GL051846.

Lustrino, M. & Wilson, M. (2007). The circum-Mediterranean anoro-genic Cenozoic igneous province. Earth-Science Reviews 81, 1^65.

Maaloe, S. (1985). Igneous Petrology. Berlin: Springer, 374 p.Marsh, B. D. (1981). On the crystallinity, probability of occurrence and

rheology of lava and magma. Contributions to Mineralogy and

Petrology 78, 85^98.Martel, C., Pichavant, M., Bourdier, J. L., Traineau, H., Holtz, F. &

Scaillet, B. (1998). Magma storage conditions and control of erup-tion regime in silicic volcanoes: Experimental evidence from Mt.Pele¤ e. Earth and Planetary Science Letters 156, 89^99.

Mart|¤, J., Pinel, V., Lo¤ pez, C., Geyer, A., Blanco, M. J., Abella, R.,Ta¤ rraga, M., Castro, A. & Rodr|¤guez, C. (2013). Causes and mech-anisms of the 2011^2012 El Hierro (Canary Islands) submarineeruption. Journal of Geophysical Research 118, doi:10.1002/jgrb.50087.

McDade, P., Wood, B. J., Van Westrenen, W., Brooker, R.,Gudmundsson, G., Soulard, H., Najorka, J. & Blundy, J. (2002).Pressure corrections for a selection of piston-cylinder cell assem-blies. Mineralogical Magazine 66, 1021^1028.

McNutt, S. R. (2005). Volcano seismology. Annual Review of Earth and

Planetary Sciences 33, 461^491.Meletlidis, S., Di Roberto, A., Pompilio, M., Bertagnini, A.,

Iribarren, I., Felpeto, A., Torres, P. A. & D’Oriano, C. (2012).Xenopumices from the 2011^2012 submarine eruption of El Hierro(Canary Islands, Spain): Constraints on the plumbing system andmagma ascent. Geophysical Research Letters 39, L17302, doi:10.1029/2012GL052675.

Neumann, E., Wulff-Pedersen, E.W., Simonsen, S. L., Pearson, N. J.,Mart|¤, J. & Mitjavila, J. (1999). Evidence for fractional crystalliza-tion of periodically refilled magma chambers in Tenerife, CanaryIslands. Journal of Petrology 40, 1089^1123.

Petry, C., Chakraborty, S. & Palme, H. (2004). Experimental determin-ation of Ni diffusion coefficients in olivine and their dependence ontemperature, composition, oxygen fugacity, and crystallographicorientation.Geochimica etCosmochimicaActa 68,4179^4188.

Pichavant, M., Martel, C., Bourdier, J. L. & Scaillet, B. (2002).Physical conditions, structure, and dynamics of a zoned magmachamber: Mount Pele¤ e (Martinique, Lesser Antilles Arc). Journalof Geophysical Research 107, 2^30.

Pichavant, M., Di Carlo, I., Le Gac, Y., Rotolo, S. G. & Scaillet, B.(2009). Experimental constraints on the deep magmafeeding system at Stromboli Volcano, Italy. Journal of Petrology 50,601^624.

Putirka, K. D. (2008). Thermometers and barometers for volcanic sys-tems. In: Putirka, K. D. & Tepley, F. J., III (eds) Minerals,



Inclusions and Volcanic Processes. Mineralogical Society of America and

Geochemical Society, Reviews in Mineralogy and Geochemistry 69, 61^120.Roeder, P. L. & Emslie, R. F. (1970). Olivine^liquid equilibrium.

Contributions to Mineralogy and Petrology 29, 275^289.Saunders, K., Blundy, J., Dohmen, R. & Cashman, K. (2012). Linking

petrology and seismology at an active volcano. Science 336,1023^1027.

Schmincke, H. -U. (1982). Volcanic and geochemical evolution of theCanary Islands. In: von Rad, U., Hinz, K., Sarnthein, M. &Siebold, E. (eds) Geology of the Northwest African Continental Margin.Berlin: Springer-Verlag, pp. 273^306.

Schmicke, H. -U. & Sumita, M. (1998). Volcanic evolution of GranCanaria reconstructed from apron sediemnts: synthesis of VICAPproject drilling. In: Weaver, P. P. E., Schmicke, H.-U., Firth, J. V. &Duffield,W. (eds) Proceedings of Ocean Drilling Program, Scientific Results,

157. College Station,TX: Ocean Drilling Program, pp. 443^469.Sigmarsson, O., Laporte, D., Devouard, B., Devidal, J. L.,

Carpentier, M. & Mart|¤, J. (2013). Formation of U-depleted rhyolitefrom a basanite at El Hierro, Canary Islands. Contributions to

Mineralogy and Petrology 165, 601^622.

Sigmundsson, F., Hreinsdo¤ ttir, S., Hooper, A., A¤ rnado¤ ttir, T.,Pedersen, R., Roberts, M. J., O¤ skarsson, N., Auriac, A.,Decriem, J., Einarsson, P., Geirsson, H., Hensch, M.,O¤ feigsson, B. G., Sturkell, E., Sveinbjo« rnsson, H. & Feigl, K. L.(2010). Intrusion triggering of the 2010 Eyjafjallajo« kull explosiveeruption. Nature 468, 426^432.

Sparks, R. S. J. (2003). Forecasting volcanic eruptions. Earth and

Planetary Science Letters 210, 1^15.Stroncik, N. A., Klu« gel, A. & Hansteen, T. H. (2009). The mag-

matic plumbing system beneath El Hierro (Canary Islands):Constraints from phenocrysts and naturally quenched basalticglasses in submarine rocks. Contributions to Mineralogy and Petrology

157, 593^607.Vigneresse, J. L., Barbey, P. & Cuney, M. (1996). Rheological transi-

tions during partial melting and crystallization with application tofelsic magma segregation and transfer. Journal of Petrology 37,1579^1600.

Watts, A. B. (1994). Crustal structure, gravity anomalies and flexure oflithosphere in the vicinity of the Canary Islands. Geophysical

Journal International 119, 648^ 666.



Correlation of Magma Evolution and Geophysical Monitoring during ...

Documents

Transcript of Correlation of Magma Evolution and Geophysical Monitoring during ...