Quantifying the limitations to the performance of passive correlation processing methods applied to the extraction of tomographic information
Abstract
Recent experiments have demonstrated that correlation of signals measured over spatially separated sensors in a diffuse, random wave field yields information about coherent (deterministic) wave propagation between the sensors. The measured coherence can be used to tomographically extract parameters related to the propagation environment. Being a stochastic measurement process, the coherent information of the sample correlation function is obscured by random (incoherent) fluctuations that average down as the sample observation duration increases. As the measurement stochastically converges with increasing sample duration to reveal the underlying coherent temporal structure associated with the spatial correlations, the signal-to-noise ratio (SNR) between the measured coherent power and the power of the random fluctuations likewise increases. While it is known that the SNR decreases with sensor separation, Cramer-Rao (CR) analysis indicates that the stochastic convergence rate for the extraction of tomographic information from measured wave coherence increases with sensor separation. Thus, while the SNR itself may be maximal for closely spaced sensors, in terms of performance in the tomographic context, it is preferable to use large sensor separation. In cases involving attenuation, there is an optimal distance of maximal tomographic performance. The CR prediction is verified through simulation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.S31C2255W
- Keywords:
-
- 7200 SEISMOLOGY