Fluid Saturation Model Inversion Using Misfit Classification

The patent implements dynamic fluid saturation models that adapt to complex geological conditions by allowing saturation values to vary spatially and temporally. The system uses iterative inversion techniques that dynamically adjust model parameters based on seismic data mismatches, enabling accurate characterization of variable hydrocarbon contacts and uncertain time-to-depth relationships rather than assuming static, uniform saturation distributions

The patent transforms the inversion problem by changing parameters from direct saturation estimation to misfit-based classification. It introduces artificial rock types as intermediate parameters that capture complex fluid distributions, and uses probabilistic saturation values that evolve through iterative refinement. This parameter transformation enables the model to handle complex geologies by representing uncertainty and variability in a mathematically tractable form

The patent performs preliminary classification of rock types and fluid saturations before full inversion by analyzing misfit patterns in the seismic data. This preliminary action identifies likely fluid types and saturation ranges, which then constrain the subsequent iterative inversion process. By pre-characterizing the subsurface using misfit-based classification, the system reduces the search space for iterative optimization, thereby maintaining accuracy while improving computational efficiency

The patent implements a two-stage inversion approach where a coarse, low-resolution model is generated first using misfit-based classification, followed by refined inversion only in regions of interest or high-uncertainty areas. This partial action strategy applies computationally intensive iterative methods selectively rather than uniformly across the entire subsurface volume, thereby reducing overall computational burden while preserving accuracy where it matters most