Reservoir Property Model Seismic Inversion Error Assessment

The patent introduces an intermediary quality metric that mediates between seismic inversion products and reservoir management decisions. This metric acts as a filter or gatekeeper, determining which inversion results are reliable enough to influence decisions. The quality metric is computed by comparing inversion results across multiple realizations and thresholding the variance, thereby introducing a mediating layer that resolves the low correlation issue by selectively accepting or rejecting inversion outputs based on their computed quality.

The patent applies preliminary action by computing the quality metric before the reservoir property model is finalized or used for decision-making. The quality assessment is performed in advance on the inversion products, allowing the modeling workflow to proceed efficiently while ensuring that only quality-assessed results are used for optimization. This preliminary quality check prevents unreliable models from being built, thereby maintaining both productivity and reliability.

The patent applies local quality by allowing different modeling approaches (deterministic or stochastic) to be applied to different regions of the reservoir based on locally computed quality metrics. High-quality regions use deterministic modeling for consistency, while low-quality regions use stochastic modeling to account for uncertainty. This spatially varying approach resolves the contradiction by making the modeling strategy adaptive to local conditions rather than uniformly applied.

The patent applies partial action by performing cross-validation in a simplified or reduced form rather than exhaustive cross-validation. The quality metric computation uses a practical approximation that provides sufficient accuracy for reservoir management decisions without the full computational burden of complete cross-validation. This partial validation approach maintains acceptable precision while significantly reducing computational time.