Multi-scale geophysical characterization of microporous carbonate reservoirs utilizing machine learning techniques: An analog case study from an upper Jubaila formation outcrop, Saudi Arabia

Microporosity hosts a significant portion of the total hydrocarbon volume in Middle Eastern carbonate reservoirs. An improved understanding microcrystals' morphology that micropores, their impact on reservoir properties, and spatial distributions will contribute to more accurate quality prediction. This study proposes methodology utilizes machine learning approaches for microporosity characterization integrates multi-scale dataset ranging from micrometer-scale SEM images meter-scale seismic attributes build porosity models. We tested an outcrop Riyadh, Saudi Arabia, which exposes upper part Jubaila Formation equivalent lower Arab-D reservoir. acquired 35 m-long core 600 2D line behind outcrop. Laboratory-scale petrophysical measurements, including porosity, permeability, acoustic velocity, bulk grain density, x-ray diffraction, were performed over 106 horizontal plugs drilled core. quantitatively characterized microtextures micrite crystals using by performing Random Forest classifications trained image features. unsupervised classification Self-Organizing Map (SOM) all lab-measured properties data clustering. investigated potential correlations between SOM clusters with morphology, resulted predictable relationship following granularity sphericity microcrystals velocity. It was possible represent multiple lithofacies single log-linear porosity-permeability value differential effective medium (DEM) aspect ratio velocity-porosity space. The inter-relationship well then propagated reservoir-grid scale inverse impedance inverted data. developed this thus provides practical way integrate key sub-grid micro-and macro-heterogeneities into property