Source distribution estimation from direct Rayleigh waves in multicomponent crosscorrelations

If seismic sources are unevenly distributed, seismic interferometry (SI) should not be used to approximate Green's functions. An uneven source distribution often leads to artifacts in (tomographic) images derived from crosscorrelations of the wavefields. However, if the source distribution is known, one can extract useful information from the SI crosscorrelations. For the case when the source distribution is not known, we present an approach to estimate spatially distributed seismic sources from inversion of seismic wavefield correlations. We applied waveform inversion to the crosscorrelations to estimate the source strengths and distribution in space. We determine that vertical- plus radial-component crosscorrelations constrain the source inversion better than only vertical-component crosscorrelations, as has been previously used.