Time-lapse imaging of a localized weak change with multiply scattered waves using numerical-based sensitivity kernel

Multiply scattered seismic waves, due to their long path length within a finite volume, provide information that can be used to detect and image weak time-lapse changes within a medium. Such weak changes are usually not resolved with singly scattered waves. Previous use of multiply scattered waves for time-lapse monitoring assume statistical homogeneity of the scattering property of the scattering medium. This homogeneity is usually characterized by either a constant mean free path or a diffusion coefficient. In a realistic medium, however, this assumption of homogeneity likely breaks down. We demonstrate the capability of resolving a localized time-lapse velocity change within a three-layer 2-D scattering model using multiply scattered waves. The layers within the model have different scattering properties. The imaging algorithm requires numerically generating the sensitivity kernel that correctly represents the statistical heterogeneity of the scattering model. We localize the weak velocity change, but the resolution of the imaged change degrades with increasing coda traveltime. A comparison of the imaged changes from the numerical kernel and the kernel that assumes statistical homogenous model suggests that the numerical kernel provides a stronger constraint on the shape of the imaged velocity change.