Transcript Slide 1

Convergent Margin Volcanism
Three topics
1. MORs versus ARCs, a fruitful comparison
2. What is the global population of arcs like?
a.
I add a wrinkle I have been trying to become comfortable
enough with to publish. Volcano spacing decreases as
plate convergence rate increases.
3. Central America is interesting
a.
Vents b. Links
c. Ba/La (Windows) d. Galapagos
Read everything first (slides and notes) and then select
specific slides (by number) to discuss
1
1. Why can’t the arcs* be more like the ridges?
• Whenever I think of some possible new tectonicvolcanic/geochemical relationship for Central
America, I check the RIDGE site and/or review the
extensive literature on Mid-ocean ridges. The global
set of convergent plate margins (CPMs) or arcs seems
to be more complicated than the ridges, or do the arc
groups just not talk to each other enough?
• *arcs (sensu lato - because many convergent plate
margins do not have an arc shape)
2
Spreading rates versus convergence rates:
Narrower distribution for convergence rates
MOR
CPM
Frequency
30
25
20
15
10
5
0
0
20
40
60
"Vc
80
100
120
140
160
(Km/Ma)
10 Km/Ma =10 mm/yr = 1 cm/yr
3
Structures depending on rates
• The MOR morphology, structure and gravity field has
an interesting dependence on spreading rate. Slow
spreading (mid Atlantic) has rugged topography and
an axial graben. Fast spreading (EPR) has smooth
topography and an axial high or crest.
• At ARCs there is nothing like the MOR systematics
with rate. There is some dependence of volcano
spacing and convergence (see below). Oblique
subduction may eventually define some global
patterns.
4
Magma chemistry and crustal thickness
• MOR depths/crustal thickness reflect magma
chemistry. The thicker the crust, the higher the degree
of melting and the lower the Na2O content (Klein and
Langmuir and a whole host of papers)
• ARC crust may affect magma chemistry in a similar
way but the community does not seem impressed
(Plank and Langmuir proposed this using Central
America as an example that works pretty well, but the
community resisted this idea.) I think it is a
reasonable idea
5
Age/history
• MOR - what history? The axis is zero age.
Plate geometry causes ridges to form and
jump. Hotspots influence ridge locations and
ridge geochemistry.
• ARC - history is vital on both plates (e.g.
Hotspot chains on subducting plate commonly
indent CPMs and/or shut off volcanism for a
period of time).
6
2. What is the global population of arcs like?
• There are relatively few global compilations of
arc properties. The recent G-Cubed paper by
Syracuse and Abers is a good start. It refers to
Jarrad (1986?) who made a global compilation
of arc parameters. Another useful paper is
d’Bremond d’Ars et al. 1995 in JGR. They
looked globally at volcano spacing and found
it random, not periodic.
7
Spacing of volcanic centers at arcs decreases
as plate convergence rate increases
Poisson Spacing (Km)
80
70
Uyeda and Kanamori
(1979) classification
60
Island arc: no active
back-arc spreading
50
Continental arc
40
Island arc: active
back-arc spreading
30
20
10
0
50
100
150
Plate convergence rate normal to arc (mm/yr)
8
Michael J. Carr IGC
G10.07 August 22, 2004
Why examine this question?
Because volcano spacings () differ significantly
Central America
Northern Sumatra
 = 23 Km
 = 65 Km
500 Km
500 Km
9
Aleutian volcanoes have spacings intermediate between Central
America and northern Sumatra
Aleutians
 = 40 Km
500 Km
10
Defining volcano spacings
• Use Central America
as a guide
• Ignore the back-arc
• Focus on the volcanic
front
• Define Volcanic
centers
• Use Smithsonian’s
GVP reference list
11
Ignore back-arc volcanoes and volcanoes like these cinder cones
12
Why ignore the little volcanoes?
Flux derived melts at volcanic front
Decompression melts in back-arc
13
A simple composite cone is a Center
Agua volcano in
Guatemala
14
A cross-arc alignment is a Center
Atitlán-Toliman-Cerro de Oro in Guatemala
15
Make decisions defining discrete centers
Volcanic center
Central America
Secondary cone in a center
Holocene activity doubtful
Back-arc cone
500 Km
Data are from Smithsonian's Global Volcanism Program
16
Use Poisson distribution to estimate spacing
• Calculate nearest neighbor spacing
• Create histogram using 10 Km or 20 Km bins
• Vary  in Poisson equation to fit histogram
Poisson is a discrete probability function
f(x, ) =
x -
e
x = 0,1,2,3,…
x!
17
Volcano spacing in Central America
 = 23 Km
15
Poisson distribution
n=36, bin=10
 =2.3 or 23 Km
Frequency
10
5
0
0
10
20
30
40
50
60
70
80
90 100 Km
Volcano Spacings in 10 Km bins
18
Volcano spacing in Kuriles-Kamchatka
 = 17 Km
20
Poisson distribution
Frequency
15
n=62 bin=10
 =1.7 or 17 Km
10
5
Suggestion of a
second mode at
75 Km.
0
0
20
40
60
80
100 Km
Volcano Spacings in 10 Km bins
19
Volcano spacings determined here agree with those published by
d’Bremond d’Ars et al.1995
d'Ars et al 1995 spacing (Km)
80
70
60
50
40
30
Cascades - an outlier
because d’Ars used
Guffanti and Weaver’s
list not Smithsonian’s
20
10
0
0
10
20 30 40 50 60 70
Poisson Spacing (Km)
80
20
Negative correlation between plate convergence rate normal to
arc and volcano spacing
80
Poisson Spacing (Km)
70
n = 15
60
r = -0.82
Marianas
50
Ryukyus
40
Tonga
30
ignored in
regression
20
10
0
0
50
100
150 km
Plate convergence rate normal to arc (mm/yr)
21
Why a negative correlation?
1. Raleigh-Taylor gravitational instability and diapirs
μ2
μ1
If viscosity of lower layer, μ1 << μ2 then
h
wavelength, λ ~ h ( μ2/μ1)
1/3
- Whitehead and Luther (1975)
Higher convergence rate could increase the thickness of
the buoyant layer (h) or lowers its viscosity, μ1
Unlikely: a. effect of μ1 has to be > than effect of h
b. distributions of spacings are random
2. Multiple generations of cavity plumes – d’Bremond d’Ars et
al. (1995)
Higher convergence rate increases the rate of cavity
plume production, resulting in closer spacings
22
3. Central America is interesting.
a. The volcano distribution
• Stoiber and Carr 1973, after Sapper (1897) and
Dollfus and Montserrat (1868), showed that
the large volcanoes define several rightstepping lines or volcanic segments.
• What if you look at all the volcanoes? That is,
ignore size and just plot vent locations?
23
Volcanic segments based on “Centers”
24
Vents <600 ka in Central America
25
Vents younger than 600 ka with arcs
26
3b. To link Volcanology and geochemistry
We study the entire
volcanic chain. We
often plot our
volcanological and
geochemical data
against Distance
27
Regularities in the Distribution and
150
Geochemistry of Central American Volcanoes
100
Ba/La
50
0
=
Zr/Nb
70
60
50
40
30
20
10
0
El Salvador
Nicaragua
Guatemala El Salvador Nicaragua
Costa Rica
Costa Rica
Volcano volume Km3
400
300
200
100
0
0
1000 Km
28
Volcanic front consists of right stepping lines
Stoiber and Carr
(1973) suggested the
subducting slab was
segmented but the
Zr/Nb result of Bolge
(2006) requires a
smooth slab (e.g.
Syracuse and Abers,
Protti, etc) thus
volcanic segments
are an upper plate
phenomenon
29
Volume distribution along volcanic front
Guatemala El Salvador Nicaragua
Carr et al. (2007)
modified from Stoiber
and Carr (1973).
400
Atitlán Santa Ana
Volcano volume Km3
This mostly ignored
pattern can now be
linked to the volcanic
segmentation and
aspects of the
geochemistry.
Costa Rica
Irazú
300
Masaya
Tecapa
Barva
200
San
Cristóbal
Rincón
100
Mv
Arenal
0
0
Volcanic segments
500
1000
Distance Km
30
El Salvador
60
Costa Rica
40
30
20
Zr/Nb decreases along each
segment then steps up at the
beginning of the next
segment (except for Central
Costa Rica, where there is
no step in the volcanic line)
Yojoa-back-arc,
no slab signal
10
0
300
500
700
900
Distance along the arc (km)
1100
0
Depth to the slab (km)
El Salvador
Zr/Nb is similar to the sawtooth pattern of depths to
slab beneath volcanoes
(from Syracuse and Abers,
2006).
Nicaragua
50
Zr/Nb
Zr/Nb or Nb depletion
correlates with volcanic
segmentation (Bolge, 2005)
70
Nicaragua
Costa Rica
50
100
150
200
300
500
700
900
Distance along the arc (km)
1100
31
Volcanic segments are oblique to gently curved
axis that connects the large volcanoes
Axis of volcanic
productivity,
similar to
contours of
seismic zone; 150
km in Nicaragua,
90 km contour in
Costa Rica
QSC
32
Within the same segment, magma paths vary, let Zr/Nb = slab signal
NW
SE
Variable reactive
path lengths
Caribbean
Plate
Decompression
melt
Cocos
Plate
Zoned
region of
Upper plate stress field
controls where the wedge
is tapped
Lower output with
short path, higher
slab signal
Maximum output,
taps everything
flux melt
Sed melt
Water
Lower output with
long path, lower 33
slab signal
A plausible model of Zr/Nb variation: basalt reacts
with mantle during ascent
80
AFC model
Part.Coefs. for cpx
Cosigüina - short path
60
R=1
Massimilant/Mmagma=2
to DM
Zr/Nb 40
to EM
DM
Mantle
compositions
20
0
Momotombo-long path
EM
0
50
100
Ba/La
150
200
34
New insights on volcanic segmentation
• Zr/Nb saw-tooth requires the smooth slab
imaged in modern seismicity studies
• Volcanic segments are upper plate structures
• A volcano’s size depends on its location
relative to melt zone
• Nb depletion is sensitive to depth to the slab
• Need to know: What causes the segments?
35
3c. What causes the regional variation in Slab
signal (Ba/La)?
Guatemala
| El Salvador | Nicaragua
| Costa Rica
150
100
Ba/La
50
0
0
500
DSDP 495
Distance
1000 Km
DSDP 1039
36
Incoming sedimentary sections are similar
but substantial unmeasured variation may exist
e
37
DSDP 495 sediment and MORB
Low
variance
High
variance
maximum in
carbonate
maximum in
hemipelagic
Hemipelagic
0
Carbonate
200
300
400
Morb
Depth in meters
100
500
.1
1
10
U/Th
10
100
Ba/La
--------Regional--------
100 1000 10000
Ba/Th
.01
.1
U/La
1
---------Local--------38
See regional variation if sediments are similar
See local variation if sediments differ
Regional Variation
Local Variation
Ba/Th
Ba/La
CS
HS
CS
10000
VF
low-Ti
100
20%
HS
1000
N.C.R.
W. Nic.
Yohoa
10
EM
El Sal.
Yohoa
100
EM
DM
DM
1
U/Th
TWO DIFFERENT WINDOWS!!
10
.01
.1
U/La
Note parallel arrays in local variation
39
La carries the regional signal,
not Ba
SiO2< 55wt. %
0.4
1/La
Guatemala El Salvador Nicaragua
Costa Rica
0.2
150
0.0
100
Ba/La
50
SiO2< 55wt. %
1000
Ba
0
500
Distance
1000 Km
Black crosses are estimated
mantle contributions
500
0
0
500
Distance
1000 Km
40
Eiler et al. 2005, strong evidence for a serpentine
component in Nicaragua from 18O data
carbonate sed
serpentine
41
Irazú-Turrialba volcanic center Costa Rica
Turrialba
569±6 ka
Irazú
136±5 ka
594±16 ka
42
855±6 ka pre Irazú
Interplay of geology and geochronology improved both age and
volume estimates
43
Extrusive volcanic flux: all segments the
same within error
44
Subducted contribution of flux is total flux minus mantle
contribution
45
For subduction contribution
Ba estimate is robust! La is not!
Balava = 100
Bamantle = 4
Basubducted=96%
Lalava = 14
Lamantle = 8
Lasubducted= 43%
Masaya
100
10
7.5% melt of DM source
1
Cs Rb Ba Th U Nb Ta K La Ce Pb Pr Sr P Nd Zr Sm Eu Ti Dy Y Yb Lu
Masaya volcano, Nicaragua
mantle contribution: 7.5% melt of DM
46
Constant flux for highly enriched elements
(Cs, Ba, K, Pb, Sr)
Segment\Element
NW Nicaragua
SE Nicaragua
Guanacaste
Cordillera Central
Cs
Rb
Ba
Th
Element flux in units of 10
0.84
20 899 0.90
1.04
25 1076 1.60
0.73
27 892 1.75
1.01
45 755 6.88
U K2O
4
La
Pb
Sr
Kg/m/Ma
1.00 1.17 4.2
1.75 1.40 7.2
1.14 1.42 9.6
2.25 1.70 22.5
3.71
4.48
3.55
5.21
566
392
554
523
Very weak model of mantle contribution
If a variable flux of subducted fluids occurs, then highly
enriched elements, like Ba, should decrease from NW to
SE. They do not.
La increases from NW to SE but has high error.
47
The Galapagos is one of the sources
48
Himu
High-μ
49
END
50