Basin simulations, reservoir simulations, laboratory measurements and field measurements are crucial details needed for making good operational decisions in frontier areas. Seismic reservoir characterization is the task that combines engineering, geological and geophysical data. Basin simulation gives the geoscientist the opportunity to incorporate sophisticated modeling into their predictions of subsurface properties. This simulation technique normally uses a regional seismic interpretation as an endpoint for a compaction, temperature, pressure or mineralogical forward model that has engineering and geophysical calibrations. Reservoir characterization work often produces multiple interpretations, using various techniques, of the same volume of the earth. How should these interpretations be combined? Which interpretations should carry more influence? 

The technological challenge of using basin simulation output with traditional seismic inversion is that the exact location of facies is not accurate. Therefore, the derived static low frequency model constructed using rock physics transforms leads to an inversion product with unphysical artifacts at worst and at best, a reiteration of the basin model with slight property variations from the seismic amplitude input conspicuously overlying. 

We present an inversion that utilizes a Bayesian framework to iteratively constructs a facies and impedance model using prior estimates of facies distribution and impedance uncertainty. This framework allows the spatial variability of properties from the basin model to be included in the inversion without introducing localized artifacts. The benefit of using a Bayesian framework in deterministic inversion at seismic resolution is that priors may be considered in order to disqualify unphysical or unlikely yet acceptable solutions from the non-unique solution space. In this application, the prior is constructed using facies specific porosity compaction trends, cement profiles based on temperature and timing and pore pressures, transformed with rock physics models to elastic properties. With these facies property volumes, we produce unique probability density functions at every seismic sample. Given the seismic input and additional priors, the inversion produces a most probable facies volume and impedances (Vp-VsDensity). The resulting properties are thus an integration of a complex basin simulation model with a deterministic seismic inversion.