Assessing the uncertainty of ambient-seismic-source-distribution estimation with matched field processing and full waveform inversion

Ambient seismic sources (e.g. ocean waves) generate ambient seismic waves, and thus in turn, one can use these waves to infer the source distribution and study the source properties. Many studies focus on the source distribution estimation result, but few discuss the uncertainty of the estimation result, even though the uncertainty is significant and should be taken into account in the interpretation. We propose to compute the uncertainty of the estimated source distribution using singular value decomposition. We focus on two commonly used estimation methods: matched field processing and full waveform inversion. We demonstrate the uncertainty of the two methods by assessing the associated point spread functions. We determine that the full-waveform inversion method possesses higher resolution than matched field processing given enough independent data.