HR: 10:35h
AN: S42A-02 [Abstracts]
TI: Joint inversion of receiver functions and Rayleigh wave dispersion for crustal and upper mantle structures: a comparison of a nonlinear genetic approach and a linear inversion
AU: * Ma, Y
EM: mayl@rice.edu
AF: Rice University, Department of Earth Sciences
6100 Main Street, Houston, TX 77005, United States
AU: Niu, F
EM: niu@rice.edu
AF: Rice University, Department of Earth Sciences
6100 Main Street, Houston, TX 77005, United States
AU: Zhou, H
EM: zhouhl@gucas.ac.cn
AF: Graduate University of Chinese Academy of Sciences, 19(A) Yuquan Road,
Shijingshan District, Beijing, 100049, China
AU: Chen, Y
EM: ychen01@harris.com
AF: Harris Corporation, 1227 S Patrick Dr., Suite 110A, Satellite Beach, FL 32937, United
States
AB:
To reduce the non-uniqueness in receiver function inversion, surface wave dispersion data have been inverted
simultaneously for crustal and upper mantle structure using either a linear iterative approach or nonlinear global
search algorithms. The former, however, has been implemented with a fixed Vp/Vs ratio while the latter was
exercised only in determining crustal structure. Here we developed a parallelized inversion scheme utilizing the
niching genetic algorithm to jointly invert receiver function and Rayleigh wave group velocity dispersion data to
determine crustal and upper mantle structure down to approximately 200 km depth. Layer thickness and Vp/Vs
ratio were treated as free and fixed parameters for the crustal and upper mantle layers, respectively. To better
constrain Moho depth and Vp/Vs ratio in the crust, we included the H-κ stacking amplitude of high quality
receiver functions in the object function. We applied the inversion to 7 CDSN stations in northeast China. For each
station, the PREM model with a total of 40 layers (9 crustal layers + 31 upper mantle layers) was used as the
initial model. We have searched a model space that consists of 500 generations that are made by a total of
500,000 individual models. Our preliminary results reveal a low velocity layer at 60-80 km depth beneath 3
stations located in the Songliao Basin which are not seen from the uplifted surrounding areas. We also perform a
linear iterative inversion with the same data set. In general, the two inversions yield very similar velocity structure,
especially at shallow depth such as crustal velocity and the crust mantle transition. There are, however, some
significant differences between velocity models derived from the two methods, for example, the linear inversion
always shows a low velocity zone at approximately 160-180 km depth, which are not seen by the nonlinear
inversion models. Our synthetic tests indicate that such a lower velocity layer is likely an artifact from fitting the
dispersion curve in the linear inversion.
DE: 7203 Body waves
DE: 7205 Continental crust (1219)
DE: 7290 Computational seismology
SC: Seismology [S]
MN: 2007 Fall Meeting