Acoustic imaging using unknown random sources

We investigate the feasibility of imaging localized velocity contrasts within a nonattenuating acoustic medium using volume-distributed random point sources. We propose a simple, two-step processing flow that utilizes the linear sampling method to invert for the target locations directly from the recorded waveforms. We present several proof-of-concept experiments using Monte Carlo simulations to generate independent realizations of band limited "white noise" sources, which are randomly distributed in both time and space. Despite the unknown and random character of the illumination on the imaging targets, we show that it is possible to image strong velocity contrasts directly from multiply scattered coda waves in the recorded data. We benchmark the images obtained from the random-source experiments with those obtained by a standard application of the linear sampling method to analogous controlled-source experiments.