Technical Paper: Regularized Gauss-Newton method using compressed Jacobian matrix for controlled source electromagnetic data inversion

Society: SEG
Paper Number:
Presentation Date: 2009
 

In this paper we propose an approach to improve the efficiency of the regularized Gauss-Newton inversion algorithm by using an adaptive cross approximation (ACA) technique. We apply the ACA technique to decompose the Jacobian matrix into two smaller rectangular matrices. In this way we improve the efficiency of the Gauss-Newton method on both memory requirements and CPU time. The improvement increases when we deal with large data sets with a large number of transmitters, receivers and frequencies. To demonstrate the improvements introduced by this method we present results of both synthetic and field data inversions for controlled source electromagnetic surveys.

Request More Information