Petrologic Reconstruction of
Magma System Variables and
Processes
A Systematic Approach
Jon Blundy (University of Bristol)
Kathy Cashman (University of Oregon)
Step 1: Find a Volcano
Volcán Colima, Mexico
Da’ Ure, Ethiopia
Mount St. Helens, USA
Shiveluch, Kamchatka
Observational Petrography
Textures provide
constraints on eruptive
processes
1 mm
- transitions in lava
flow morphology
- onset of
fragmentation
- changes in
eruption style
Hawaiian lava flows
transitions from
pahoehoe to `a`a
morphologies
relate to...
Hawaiian lava flows
pl + px
ves
glass
ol
changes in
crystallinity
during flow
Quantitative Petrography
BSE
Al map
Fe map
Plag
Qui ck Ti me ™ an d a
Gra ph i cs d ec om pres so r
a re ne ed ed to s ee th i s pi c tu re.
Opx
•X-ray mapping (various scales)
•Mineral modes (phenocryst + groundmass)
•Crystal size distributions (multi-phase)
Example: nucleation-dominated
crystallization in lava channels
DOWN CHANNEL
VENT
Crystallinity increases by addition of small crystals
Example: nucleation-dominated
crystallization in lava channels
DOWN CHANNEL
VENT
16%
1%
CSD has a constant slope (size) and changing
intercept/slope (number)
LAVA LAKE SURFACE
Example: growth-dominated
crystallization in lava lakes
Crystallinity increases by crystal growth
LAVA LAKE SURFACE
Example: growth-dominated
crystallization in lava lakes
CSD has a variable slope (size) and constant
intercept/slope (number)
Mineral
Geothermometry
Two-pyroxene
1200
1100
1000
1100
900
Brey & Kohler (1990)
aad=62°C
800
800
900 1000 1100 1200 1300
Experimental temperature (°C)
1200
aad=36°C
Two-oxide
Calculated temperature (°C)
Calculated temperature (°C)
1300
Holland & Blundy (1994)
Thermometer B
Hb-Plag
1000
900
800
700
Calculated temperature (°C)
1100
600
1000
n=298
600
700
800
900 1000 1100
Experimental temperature (°C)
900
• Ensure phases are in textural
equilibrium, ideally touching
800
• Different mineral pairs have different
closure temperatures
700
Andersen & Lindsley (1985)
with Lindsley & Spencer (1982)
600
600
700
800
900 1000 1100
Experimental temperature (°C)
1200
• Accuracy << Precision
Degassing-induced reduction -
why fumaroles are more reduced than phenocrysts
-10
-11
-10.4
Mount St. Helens
Mount St. Helens
-10.8
-12
NNO
log10 fO2
log10 fO2
-13
-11.2
-14
-15
-18
-12
-12.4
-16
-17
-11.6
Cryptodome
May 18, 1980 (Plinian)
May 25, 1980 to June 1981
High-T fumaroles
-12.8
-13.2
-13.6
680 720 760 800 840 880 920 960
Temperature (°C)
NNO
Cryptodome
May 18, 1980 (Plin.)
May 25, 1980
Jun 12, 1980
Jul 22, 1980
Aug 7, 1980
Jun 1981
Jun 1984
May 1985
May 1986
820 840 860 880 900 920 940 960
Temperature (°C)
Fe3+ (melt) + S2- (melt) = Fe2+ (melt) + S4+(vapour)
7
pre-1980 (bulk)
1980-86 (bulk)
current (bulk)
1980-86 (MI-plag)
1980-86 (MI-hbl)
1980-86 (MI-opx)
1980-86 (MI-cpx)
1980-86 (gm)
current (MI-plag)
current (MI-hbl)
current (MI-opx)
6
5
wt% K2O
Redrafted from Hammer &
Rutherford (2002)
4
3
2
1
0
45
50
55
60
65
70
wt% SiO2
75
80
85
Glass composition as
a barometer
Blundy & Cashman (2001)
Data from Martel & Schmidt (2003)
Melt inclusions - clues to magma ascent
•Analysis of H2O and CO2 by SIMS, FTIR or microRaman
•Plagioclase-melt thermometryof embayments
•Incompatible elements in glass yield crystallinity
Solubility of H2O as a Barometer
600
Calculated pressure (MPa)
500
Mangan et al
Gardner 07
Gardner et al 99
Larsen & Gardner
Scaillet et al
400
1:1
300
200
VolatileCalc
Newman & Lowenstern (2002)
400
Calculated pressure (MPa)
Humphreys et al
Martel et al
Pichavant et al
Gardner et al 95
Martel & Schmidt
Mangan & Sisson
1:1
300
200
100
100
Liu et al. (2005)
0
0
0
100
200
300
400
Experimental pressure (MPa)
0
100
200
300
400
Experimental pressure (MPa)
•pH2O - aad ≈ 23 MPa
•CO2 is much less soluble and can contribute significantly to Ptotal
How do melt inclusions form?
Melt composition is locked-in at the
moment of occlusion.
Use ion-microprobe data to calculate:
• P (from H2O and CO2)
•F (from incompatible trace elements)
•T (from plag-melt thermometry)
Decompression-driven versus cooling-driven
crystallisation
7
6
II. Isobaric
H O-undersaturated
2
I. Isobaric
H O-saturated
wt% H2O
5
2
4
3
IV. Syn-eruptive
degassing
III. Decompression
H O-saturated
2
2
1
0
68
70
72
74
wt% SiO2 (anhyd.)
76
78
Inclusion populations record ascent trajectories. Time and pressure
Decompression-driven crystallisation
Apr-10 (plag)
Crypto (plag)
Crypto (opx)
May-18 (plag)
May-18 (hbl)
May-25 (plag)
Jun-12 (plag)
Jul-22 (plag)
Aug-7 (plag)
Aug-7 (opx)
Oct-16 (plag)
Oct-16 (hbl)
Dec-27 (plag)
Dec-27 (cpx)
Jun 1981 (plag)
May-18 (gm)
1980 (gm)
7
6
wt% H2O
5
4
3
ruptured
inclusions
2
1
0
66
1980-81
68
70
72
74
wt% SiO2(n)
76
78
80
Much faster than crystallisation driven by cooling
Linking thermometry and barometry
-10
250
pH2O (MPa)
200
150
100
-11
log 10 fO2
Cryptodome
18-May-80 (Plin)
25-May-80
12-Jun-80
22-Jul-80
7-Aug-80
16-Oct-80
27-Dec-80
18-Jun-81
19-Mar-82
-12
Cryptodome
18-May-80 (Plin)
25-May-80
12-Jun-80
22-Jul-80
7-Aug-80
18-Jun-81
amph-out
-13
NNO
50
-14
0
820
940
900
860
Temperature (°C)
980
820
860
900
940
Temperature (°C)
980
Blundy et al. (Nature, 2006)
Magma heating by decompression crystallisation at Mount St. Helens
Latent heat release ≈ 2.5 °C/%
Volatile Systematics
350
1200
Late Oranui
300
250 MPa
Early Bishop
150 MPa
Mid Oranui
Middle Bishop
1000
Early Oranui
250 MPa
Late Bishop
250
ppm CO2
ppm CO2
800
200
50 MPa
150
350 MPa
600
150 MPa
400
100
50
200
0
0
0
1
2
3
4
wt% H2O
5
6
7
2
3
4
5
wt% H2O
•Simple degassing trends are rarely observed
•Melt inclusion record is not a static snapshot
6
7
Simple degassing-crystallisation scenarios
0
600
50
500
100
ppm CO2
Pressure (MPa)
400
150
200
decompression only
isobaric vap. sat'd
isobaric vap. undersat'd
slow decomp. xtlln
fast decomp xtlln
250
200
100
300
350
0
0
10
20
30
40
50
wt% crystallised
60
70
1
8
600
7
500
2
3
4
5
wt% H2O
6
7
8
60
70
parents
6
400
ppm CO 2
parents
wt% H2O
parents
300
5
4
300
200
3
100
2
1
0
0
10
20
30
40
50
wt% crystallised
60
70
0
10
20
30
40
50
wt% crystallised
Basaltic systems - Mt. Etna
0
Data from Spilliaert et al. (2006)
4
H2O (wt%)
Pressure (MPa)
100
200
300
400
500
3
2
1
0
0
10
20
30
40
50
60
QuickTime™ and a0
TIFF (Uncompressed) decompressor
are needed to see this picture.
Crystallinity (%)
5000
2002
60
Crystallinity (%)
4000
CO2 (ppm)
CO2 (ppm)
40
5000
2001
4000
20
3000
2000
3000
2000
1000
1000
0
0
0
0
1
2
H2O (wt%)
3
4
20
40
Crystallinity (%)
Decompression crystallisation, but data are scattered
60
A model magma chamber
roof
0.6
0.03
150
0.5
200
wall
0.4
interior
Pressure (MPa)
Xg=0.02
0.01
0.3
250
0.2
Xc=0.1
volatile saturated
300
unsat'd
floor
750
800
850
Temperature (°C)
900
Spatial variation of
intensive
parameters and
fractions of crystals
and gas
Melt inclusions
potentially sample
this complexity
700
8
325 MPa
275 MPa
225 MPa
175 MPa
125 MPa
600
field of melt
inclusions
7
parent
wt% H2O
ppm CO2
500
400
300
field of melt
inclusions
200
Melt
Inclusion
Diversity
6
5
4
parent
100
0
3
4
5
6
wt% H2O
7
3
8
0
10
20
30 40 50 60
wt% crystallised
70
80
Can
inclusion
populations
be used to
map out
storage
regions?
700
8
600
field of melt
inclusions
7
parent
6
ppm CO 2
wt% H2O
500
5
400
field of melt
inclusions
300
200
parent
4
100
3
750
800
850
Temperature °C
900
0
10
20
30 40 50 60
wt% crystallised
70
80
Large silicic systems
Oruanui and Bishop Tuff
7
6
H2O (wt%)
5
1600
CO2 (ppm)
1400
4
3
1200
2
1000
1
800
0
Oruanui (early)
Oruanui (middle)
Oruanui (late)
Bishop (early)
Bishop (middle)
Bishop (lat e)
0
600
20
40
60
Cryst allinity (%)
400
1200
50 MPa
200
1000
0
1
2
3
4
H2O (wt%)
5
6
7
CO2 (ppm)
0
800
600
400
200
0
0
20
40
Crystallinit y (%)
Data from Wallace et al. (1999); Liu et al. (2006)
60
Trace Element Chemistry of
Minerals
Diffusion-moderated record of
open system processes
Smith et al (in press)
Experimental
Reconstruction
Crystallinity
200 MPa, 850 °C
Plag/Hbl
SiO2 (gl)
Al 2O3 (gl)
CaO (gl)
XAn (pl)
Al 2O3 (hb)
3.5
4.0
Data from Costa et al. (2004)
4.5
6.0
5.5
5.0
bulk wt% H2O
•Careful choice of starting materials
•Concept of reactive volume
•Matching of multiple petrological
parameters
6.5
7.0
Experimentallydetermined kinetic
parameters
Log plagioclase Nv (/mm3)
7
6
50 MPa
5
125 MPa
4
3
0
200
400
600
Time (hours)
Growth (Y) and Nucleation (I) rates
From Hammer & Rutherford (2002)
Plagioclase Number Density (Nv) as
as a barometer
Data from Couch (2003)
Links to Volcano Monitoring
Mount St Helens, USA
Volcan Colima, Mexico
Cadavers and Clinics
Calculating P
(and z) from
H2O and CO2
For comparison
to earthquake
depths
depth below sea-level (km)
21-Oct-86
17-Jun-84
7-Feb-83
19-Mar-82
18-Jun-81
27-Dec-80
16-Oct-80
7-Aug-80
12-Jun-80
18-May-80
-3
10-Apr-80
Petrology vs Seismology at Mount St. Helens
0
3
6
9
12
15
100
1000
Days since 17 March 1980
earthquakes
P = 1.287 pH O
tot
May-Dec 1980 = period of high gas flux, deep
melt inclusions and few earthquakes…
P
tot
2
= pH O + pCO
2
2
Petrologic
Gas Flux vs
Observed
2.0
0.3
3
Melt Inclusions
2
0.2
1
0.1
0
0
500
1000
1500
2000
Days since 17 March 1980
SO2 flux (kTe/day)
wt% SO2 in MI
Flux
observed SO 2 flux (kTe/day)
0.4
1.5
16-Oct-80
12-Jun-80
1.0
7-Aug-80
18-Jun-81
27-Dec-80
0.5
21-Oct-86
14-May-82
0
2500
19-Mar-82
18-Aug-82
7-Feb-83
0.0
0.0
24-May-85
0.5
1.0
1.5
calculated SO2 flux (kTe/day)
Plinian eruptions have excess gas; effusive eruptions do not
2.0
Calculating conduit dimensions
from melt inclusions
r
Mount St. Helens 1980-82
27 Dec 80
V
12 Jun 80
20
15
22 Jul 80
25
5
0
100
20 Mar 82
10
h
Chadwick et al. (1988)
25 May 80
Conduit radius (m)
30
18 Oct 80
7 Aug 80
35
200 300 400 500 600
Days since 17 March 1980
700
800
r
V
z
Essential, but hard-to-constrain, parameter for numerical
 modelling
Conclusions
• Petrology is a powerful tool for unravelling
conditions and processes
• Phase chemistry allied to quantitative
petrography
• Underpinned by experimental constraints
• Temporal and spatial variability
• Inputs to dynamical modelling
• Links to monitoring signals