Physics-Imposed Resolution and Robustness Issues in Seismo-Acoustic Parameter Inversion

Abstract

Adaptive optimization schemes have become important tools for rapid matched field environment inversion. In spite of several successful applications to both synthetic and real data it has been found that these algorithms often fail to converge to the correct solution even for relatively simple environments. We use the theoretical Cramer-Rao resolution bounds to demonstrate that this lack of robustness is a result of the adaptive use of the local parameter sensitivity to adjust the search space. Thus, the waveguide physics yields highly inhomogeneous and coupled resolution properties within even a limited parameter space, which often forces adaptive optimization schemes to converge on a wrong local minimum of the cost function.