A downscaling strategy from FWI to microscale reservoir properties from high-resolution images
Résumé
Extracting detailed earth information from an ensemble of seismic traces is a challenge facing full-waveform inversion. So far, success on synthetic and real data has been accomplished primarily for the twin purposes of complex structural imaging and geologic interpretation. An ongoing issue for the seismic-imaging community, in addition to building high-resolution images, is the reliable extraction of acoustic and shear velocities, anisotropic parameters, quality factors, and density. Such extractions, performed at the seismic resolution scale, should help greatly with quantitative interpretation and estimation of rock properties. A step toward this goal is described here. A generic rock-physics model is assumed, which upscales microscale rock-physics properties to mesoscale (effective-medium) poroelastic quantities to be recovered from macroscale estimates of seismic attributes. It is shown on simple synthetic examples that quantitative multiparameter reconstruction, when it is possible, can reduce ambiguities in mesoscale parameter estimation dramatically, using a semiglobal search. Successful estimation of these effective-medium quantities will narrow the range of possible rock-physics estimations to be considered for seismic imaging target zones. For example, estimating the P-wave quality factor along with P-wave velocity from full-waveform inversion is shown to change the estimation of mesoscale parameters significantly, assuming that the upscaling of the rock-physics model and the recovered macroscale parameters are well constrained. In addition, shear-wave information is shown to be crucial for pressure-saturation discrimination. The inferred information at the reservoir level, resulting from full-waveform inversion and subsequent mesoscale estimation, can be useful for reservoir characterization.