Uncertainty-aware reservoir characterization using probabilistic seismic–petrophysical integration

Accurate reservoir characterization is critical for hydrocarbon exploration and production; however, conventional deterministic approaches often fail to account for uncertainty arising from seismic noise, limited well-log coverage, and interpretation ambiguity. This study proposes a probabilistic seismic–petrophysical integration framework to quantify and propagate uncertainty in reservoir property estimation. The methodology integrates 3D seismic attributes, seismic inversion results, and well-log data using Bayesian neural networks and Monte Carlo simulation to generate probabilistic predictions of key reservoir parameters such as porosity, permeability, and hydrocarbon saturation. Unlike traditional workflows, the proposed approach provides confidence intervals and probability maps, enabling risk-aware decision-making. Results demonstrate that incorporating uncertainty improves reservoir prediction reliability and reduces exploration risk. The framework represents a significant advancement toward intelligent, data-driven reservoir modeling for modern hydrocarbon systems.