Transcript ppt

IBM’s Second Arc
Jim Gill, UCSC
• SubFac has transformative potential...
• But beware of consensus:
Some of what Rummie said he knew
wasn’t true.
Logic of IBM Focus Site:
Compare and Contrast I vs M
• Now:
– Izu shallower slab, Marianas
steeper
– Marianas backarc spreading; Izu
rifting
– Izu subducts more loess; Mariana
more OIB-type vclastic sediment
• Before:
– SPVB: more melting in Izu backarc
– WPB: north has old arcs; south was
oceanic crust
Thanks in part to MARGINS…
• Lots of data are in databases
* Georoc: CIP 1017 entries; Izu Arc 3367
* Petdb: Mariana Trough 1463
* IFREE/Ganseki: ~400
• ~10% are ‘high quality comprehensive data’ for
single samples (Majors, ICPMS TE, Sr-NdPb±Hf,Useries)
* Izu Arc/Backarc: ~300
* Mariana CIP+NSP: ~100
• Glass/Melt inclusion data including volatiles
* Through space (Kent, Shaw, Kelley)
* Through time (Straub)
Pre-MARGINS
Transformative Consensus
A. Three components contribute to magmas
• Mantle wedge: I-DMM
– Best seen in HREE, HFSE
• AOC ‘Fluid’ (Pearce ‘shallow component’)
– Best seen in As, Sb, B, U, Ba, Pb
• Sediment ‘Melt’ (Pearce ‘deep component’)
– Best seen in 10Be, Th, LREE
B. Fluid and Melt are added separately
(ternary mixing is the norm)
Why Transformative?
• Technical breakthrough: HQCD show elementisotope systematics that were interpreted as
implicating sediment melt.
• If sediment melts, so does wet-AOC.
• Slab melt is everywhere.
• If slab melt is everywhere, the old numerical
models are wrong.
• Informs testable models (e.g., Kimura, Baker)
• The time scale of slab dehydration, flux melting,
melt ascent, and differentiation is human.
However, remember Rummie.
• The mantle is not uniformly D-IMM. The VF vs
RA difference is not just more sediment in RA.
• Ba tracks melt as well as fluid; Th is fluid mobile;
much Pb comes from the mantle in the RA;
accessory minerals matter.
• U-Th disequilibria require fast processes but may
not require two stages.
• Therefore, we still don’t yet understand the steady
state forcing functions.
• Non-steady-state processes matter and we
understand them even less.
• How does all this affect crustal evolution?
• SubFac needs post-MARGINS transformation.
IBM Vocabulary
• Mariana (Along strike)
– CIP
– NSP
• Izu (Across strike)
– Volcanic Front
– BAK (extensional zone)
– WS(reararc seamount
chains)
WS BAK
VF
How well does the current 3-component consensus
explain an iconic figure?
F
M
DMM
Mariana CIP
F
M
DMM+M
DMM
Izu VF: Higher Ba/La but lower Ba/Yb;
lower La/Yb in Second Arc; no slab melt
Izu WS: 3-12 Ma; 1-2% H2O in MI (Kent)
Izu BAK: <2 Ma; 0.5-1.5% H2O in MI;
mostly decompression melts
Little overlap
of I and M
Elliott et al 1997’s Tranformative Science:
or, Pearce’s talk plus isotopes
Analytically challenging, especially between labs.
However, also works with Ce/Ce*.
CIP: Elliott Plus Woodhead
Izu VF: flat trend for isotopically depleted mantle:
Th not from sed melt; maybe AOC fluid?
Izu WS: another flat trend; not just sediment melt;
Isotopically enriched mantle?
Izu BAK: purest DMM ± AOC melts
Analytical improvements; I≠M now; Izu VF≠BAK≠WS;
RA more ‘Indian’ going north; CIP between sediment and AOC
Lots of mantle Pb in RA (from sulfide?)
50% mantle Pb
See Tollstrup poster: I≠M; Izu BAK is least “Indian”
because of AOC melt?; BAK most like First Arc
Rummie summary questions
• Is there initial along-strike variability in the mantle as well as in
recent slab components? (Can we distinguish a ‘deus ex sedimentus’
from enriched mantle without 10Be?) How much of this mantle
variation is inherited by the crust?
• How much of the across-strike variability of arc and backarc magmas
is due to the percent of flux melting, the nature of the flux, or the
nature of the mantle?
• What aspects of that ‘nature of the flux’ reflects its solute content
(fluid vs melt) vs the composition of its source (AOC, sediment) vs
the refractory mineralogy of the source (rutile, zircon, monazite,
phengite)? How does this relate to variations in P-T?
• Why are the rear arc melts so similar throughout Mariana, Izu, NEJ
when volcanic front melts are so different?
• What caused the differences between Izu and Mariana to increase
after first backarc spreading?
• Can intra-crustal differentiation create “bulk continental crust”
without recycling significant continent-derived sediment?
More consensus: There are consistent spatial and
temporal variations in components
• DMM mantle is variably depleted
– More depleted in VF than RA (trenchward advection even w/o spreading)
– More depleted in Izu after SPV backarc spreading
• Slab fluid is ubiquitous
– Fluid effect decreases from VF to RA because of less %fluid, distillation
of fluid source, less depleted mantle, more sediment in RA
• Sediment-rich (few%) slab melt is more localized
– More in Marianas: least in central CIP (Maug to Guguan)
– Less in Izu: absent from VF ± BAK
– Little slab sediment melting in Izu since 3 Ma
• Quaternary volcanoes track along-strike variations in sediment
206Pb/204Pb, Th/La
All IBM
Add Kasuga RA (~NSP)