Mm. Deal et G. Nolet, Slab temperature and thickness from seismic tomography 2. Izu-Bonin, Japan, and Kuril subduction zones, J GEO R-SOL, 104(B12), 1999, pp. 28803-28812
Delay times from teleseismic and local P wave arrivals are used to invert f
or a high-resolution three-dimensional velocity model beneath the northwest
Pacific. The model shows high-velocity slabs with average velocity anomali
es of the order of 3-4%. Assuming the positive velocity deviations in the s
ubducting lithosphere are to first order due to a temperature anomaly, the
results of a theoretical slab temperature profile based on the diffusion eq
uation are converted to a. synthetic slab velocity model. Temperature varia
tions between the ambient mantle and the interior of the slab are converted
to P wave velocity perturbations using dV(p)/dT approximate to 4.8 x 10(-4
) km s(-1) degrees C-1. A nonlinear optimization scheme compares the tomogr
ams obtained via tomography to the theoretically predicted models in order
to determine the optimal values for slab thickness and mantle potential tem
perature. Using 1180 +/- 100 degrees C, as the potential temperature, thick
ness estimates of 88 +/- 8 km, 85 +/- 8 km, and 84 +/- 8 km are obtained fo
r the Izu-Bonin, Japan, and Kuril slabs, respectively. A correlation exists
between slab thickness and age, which is strong if mantle temperature vari
ations along the slab strike can be ruled out. In the process of estimating
slab thickness the predicted slab velocity model is used as a filter to en
hance the initial minimum-noon tomographic result. The initial tomogram is
modified to closely resemble the synthetic slab tomogram by using only null
-space components. The use of the null-space components guarantees that the
enhanced solution will satisfy the original seismic delay times. The enhan
ced slab images show very continuous and narrow slabs compared to the initi
al tomographic results.