Convergent Margin VolcanismThree topics • MORs versus ARCs, a fruitful comparison • What is the global population of arcs like? • I add a wrinkle I have been trying to become comfortable enough with to publish. Volcano spacing decreases as plate convergence rate increases. • Central America is interesting • 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. Why can’t the arcs* be more like the ridges? • Whenever I think of some possible new tectonic-volcanic/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)
MOR CPM Frequency 30 25 20 15 10 5 0 0 20 40 60 80 100 120 140 160 "Vc (Km/Ma) 10 Km/Ma =10 mm/yr = 1 cm/yr Spreading rates versus convergence rates:Narrower distribution for convergence rates
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.
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
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).
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.
70 Island arc: no active back-arc spreading 60 Continental arc 50 Poisson Spacing (Km) Island arc: active back-arc spreading 40 30 20 10 0 50 100 150 Plate convergence rate normal to arc (mm/yr) Spacing of volcanic centers at arcs decreases as plate convergence rate increases 80 Uyeda and Kanamori (1979) classification Michael J. Carr IGC G10.07 August 22, 2004
Why examine this question?Because volcano spacings () differ significantly Central America N Northern Sumatra N = 23 Km = 65 Km 500 Km 500 Km
N Aleutian volcanoes have spacings intermediate between Central America and northern Sumatra Aleutians = 40 Km 500 Km
Use Central America as a guide Ignore the back-arc Focus on the volcanic front Define Volcanic centers Use Smithsonian’s GVP reference list Defining volcano spacings
Why ignore the little volcanoes? Flux derived melts at volcanic frontDecompression melts in back-arc
A simple composite cone is a Center Agua volcano in Guatemala
A cross-arc alignment is a Center Atitlán-Toliman-Cerro de Oro in Guatemala
Central America Volcanic center N Secondary cone in a center Holocene activity doubtful Back-arc cone 500 Km Make decisions defining discrete centers Data are from Smithsonian's Global Volcanism Program
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 xe- x = 0,1,2,3,… f(x, ) = x!
15 Poisson distribution n=36, bin=10 =2.3 or 23 Km 10 Frequency 5 0 0 10 20 30 40 50 60 70 80 90 100 Km Volcano Spacings in 10 Km bins Volcano spacing in Central America= 23 Km
20 Poisson distribution n=62 bin=10 =1.7 or 17 Km 15 Frequency 10 5 0 0 20 40 60 80 100 Km Volcano Spacings in 10 Km bins Volcano spacing in Kuriles-Kamchatka= 17 Km Suggestion of a second mode at 75 Km.
Volcano spacings determined here agree with those published by d’Bremond d’Ars et al.1995 80 70 60 45º 50 d'Ars et al 1995 spacing (Km) 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 80 Poisson Spacing (Km)
Negative correlation between plate convergence rate normal to arc and volcano spacing 80 70 n = 15 r = -0.82 60 Marianas Ryukyus Tonga ignored in regression 50 Poisson Spacing (Km) 40 30 20 10 0 0 50 100 150 km Plate convergence rate normal to arc (mm/yr)
Why a negative correlation? 1. Raleigh-Taylor gravitational instability and diapirs μ2 If viscosity of lower layer, μ1 << μ2 then h wavelength, λ ~ h(μ2/μ1)1/3 - Whitehead and Luther (1975) μ1 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
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 right-stepping lines or volcanic segments. • What if you look at all the volcanoes? That is, ignore size and just plot vent locations?
Distance 3b. To link Volcanology and geochemistry We study the entire volcanic chain. We often plot our volcanological and geochemical data against Distance
150 100 Ba/La 7 0 El Salvador Nicaragua Costa Rica 6 0 50 5 0 4 0 3 0 2 0 0 1 0 0 Guatemala El Salvador Nicaragua Costa Rica 400 300 Volcano volume Km3 200 100 0 0 1000 Km Regularities in the Distribution and Geochemistry of Central American Volcanoes = Zr/Nb
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
Guatemala El Salvador Nicaragua Costa Rica Guatemala El Salvador Nicaragua Costa Rica 400 400 Atitl Atitl á á n n Iraz Iraz ú ú Santa Ana Santa Ana 300 300 Masaya Masaya Tecapa Tecapa Volcano volume Km3 Barva 200 200 San San Crist Crist ó ó bal bal Rinc Rinc ó ó n n 100 100 Mv Arenal 0 0 0 0 500 500 1000 1000 Distance Km Volume distribution along volcanic front Carr et al. (2007) modified from Stoiber and Carr (1973). This mostly ignored pattern can now be linked to the volcanic segmentation and aspects of the geochemistry. Volcanic segments
7 0 El Salvador Nicaragua Costa Rica 6 0 5 0 b 4 0 N / r Z 3 0 2 0 1 0 0 3 0 0 5 0 0 7 0 0 9 0 0 1 1 0 0 Distance along the arc (km) Zr/Nb or Nb depletion correlates with volcanic segmentation (Bolge, 2005) 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 0 El Salvador Nicaragua Costa Rica ) m 5 0 Zr/Nb is similar to the saw-tooth pattern of depths to slab beneath volcanoes (from Syracuse and Abers, 2006). k ( b a l s e 1 0 0 h t o t h t p 1 5 0 e D 2 0 0 3 0 0 5 0 0 7 0 0 9 0 0 1 1 0 0 Distance along the arc (km)
QSC 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
Cocos Plate Sed melt Water Within the same segment, magma paths vary, let Zr/Nb = slab signal NW SE Upper plate stress field controls where the wedge is tapped Variable reactive path lengths Caribbean Plate Lower output with short path, higher slab signal Decompression melt Maximum output, taps everything Zoned region of flux melt Lower output with long path, lower slab signal
A plausible model of Zr/Nb variation: basalt reacts with mantle during ascent 80 AFC model Part.Coefs. for cpx R=1 Massimilant/Mmagma=2 Cosigüina - short path 60 to DM Zr/Nb 40 to EM DM Mantle compositions Momotombo-long path 20 EM 0 0 50 100 150 200 Ba/La
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?
Guatemala | El Salvador | Nicaragua | Costa Rica 150 100 Ba/La 50 0 0 500 1000 Km Distance 3c. What causes the regional variation in Slab signal (Ba/La)? DSDP 495 DSDP 1039
e Incoming sedimentary sections are similar but substantial unmeasured variation may exist
DSDP 495 sediment and MORB Low variance maximum in carbonate maximum in hemipelagic High variance ---------Local--------- --------Regional--------
See regional variation if sediments are similarSee local variation if sediments differ Note parallel arrays in local variation TWO DIFFERENT WINDOWS!!
Guatemala El Salvador Nicaragua Costa Rica 1 5 0 S i O < 55 w t . % 2 1 0 0 0 B a 1 0 0 B a / L a S i O < 55 w t . % 2 5 0 0 0 . 4 5 0 1 / L a 0 0 0 . 2 0 5 0 0 1 0 0 0 Km 5 0 0 1 0 0 0 Km D i s t a n c e D i s t a n c e 0 . 0 La carries the regional signal, not Ba Black crosses are estimated mantle contributions
serpentine Eiler et al. 2005, strong evidence for a serpentine component in Nicaragua from 18O data carbonate sed
Irazú-Turrialba volcanic center Costa Rica Turrialba 569±6 ka Irazú 136±5 ka 594±16 ka 855±6 ka pre Irazú
Interplay of geology and geochronology improved both age and volume estimates
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, Nicaraguamantle contribution: 7.5% melt of DM For subduction contribution Ba estimate is robust! La is not! Balava = 100 Bamantle = 4Basubducted=96% Lalava = 14Lamantle = 8 Lasubducted= 43%
Constant flux for highly enriched elements (Cs, Ba, K, Pb, Sr) 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.