HR: 0830h
AN: S21E-0359 [PDF]
TI: Correlation Length Scales of Isotopic Variations Along Mid-Ocean Ridges and Upper Mantle
Dynamics
AU: * Graham, D W
EM: dgraham@coas.oregonstate.edu
AF: Oregon State University, College of Oceanic and Atmospherc Sciences, Corvallis, OR 97331 United States
AU: Spera, F J
EM: spera@geol.ucsb.edu
AF: Univ of California Santa Barbara, Dept of Earth Science, Santa Barbara, Ca 93106 United States
AB:
How isotopic variations in basalts erupted at the Earth's surface are linked to convective mixing in the underlying mantle is
a central problem in geodynamics. The objective of this study is to quantify the length scales of upper mantle heterogeneity
through spatial statistical analysis of MORB. We define a characteristic length scale, the scale of segregation L, computed
from the spatial self-correlations for $^{3}$He/$^{4}$He, $^{87}$Sr/$^{86}$Sr, $^{143}$Nd/$^{144}$Nd and
$^{206,207,208}$Pb/$^{204}$Pb in "zero age" lavas from mid-ocean ridges. Our working hypothesis is that small scale
convection in the upper mantle controls dispersion of geochemical tracers. Differences in L between ocean basins may then be
quantitatively related to unsteadiness in this convection, due to thickening of the lithosphere, plume impingement, or
lateral temperature/compositional differences between continental and oceanic lithosphere induced by batholith formation.
The correlation coefficient R, and the separation distance r, are calculated for every i,j pair of points. R$_{i,j}$ is given
by the product of the deviations in isotope composition from the population mean, normalized to the population variance, and
R(r) is computed as an ensemble average. The total number of point pairs (N) for n sample locations is given by N=n(n-1)/2.
For the global MORB data set (n=1265 and 735 for Sr and He, respectively), N exceeds 10$^{5}$ (799480 and 269745,
respectively). A value of R(r) close to 1 indicates that an isotope ratio above (or below) the population average is likely
to be associated with an above (or below) average value at a distance r away. A value of R(r) close to zero implies a random
relationship, and a value close to -1 implies an anti-correlation. R(r) approaches unity at small r by definition, as points
close together are from the same "clump" of mantle. The value of r at which R first goes to zero is denoted as r*. On a
diagram of R(r) vs. r (the correlogram), the integral of R(r) from r=0 to r=r* is the scale of segregation L, a
characteristic length related to "clump" size. Within the Earth's interior compositionally distinct regions may vary in size
and shape and their boundaries may be diffuse. Lithosphere recycling at subduction zones and plume input from the lower
mantle or transition zone are two ways by which compositionally distinct materials are introduced to the upper mantle. We use
the L concept because it is a quantitative estimate of size which is precisely defined from spatially referenced geochemical
data.
Only a narrow range of L values is found within each ocean basin. In general, the isotope tracers have L values between
350-750 km for the N Atlantic, S Atlantic. and Indian Oceans. For the Pacific Ocean, L appears to be significantly less,
between 100-180 km, indicating that mantle clumps in this region are smaller, in agreement with inferences based on the
standard (non-spatial) statistical approach. Our results provide supporting evidence that dispersion of isotopic anomalies in
the upper mantle is primarily controlled by small-scale convection. On a geological time scale, mantle flow is certainly
unsteady. Small-scale convective cells, having their long axes aligned subparallel to the plate scale flow, are one simple
manifestation of this unsteadiness. In Rayleigh-B‚nard convection, the amplitude and frequency of cellular wobble controls
the extent of lateral mass transport (dispersion) between convective cells. Based on the spatial self-correlation of isotopic
tracers, this dispersion appears to be more efficient in the Pacific upper mantle than elsewhere.
DE: 1025 Composition of the mantle
DE: 3035 Midocean ridge processes
SC: Seismology [S]
MN: 2003 Fall Meeting