Understanding plume migration andF trapping mechanisms within heterogeneous reservoirs is critical for optimizing carbon storage performance and ensuring long term containment. Predicting plume behaviour in heterogeneous reservoirs requires pore scale descriptors that preserve capillary physics and connectivity. We develop a lognormal Gaussian decomposition of microCT‑derived pore volume probability density functions (PDFs) for eight North Sea reservoir plugs. Each PDF is resolved into three subpopulations: micropores, mesopores and macropores characterised by modal weight , mean radius and spread Band cut‑offs are anchored in drainage capillary thresholds extracted from reference data (largest entry and percolation pressures), ensuring classes correspond to hydraulically meaningful activation pressures. Model fits are excellent (R² > 0.90). The grouped parameters show strong, physically consistent Spearman correlations with the laboratory petrophysics: permeability increases with and and decreases with . Consistent trends in sonic transit time, bulk density, and electrical resistivity corroborate these relationships. Image‑derived bands predict the shape of -water relative permeability curves and, in compositional -brine simulations (2,500-day injection plus 2,000 years shut‑in), reproduce plume stability, vertical rise, capillary trapping and dissolution signatures. A ternary projection of () provides an operational classification into micro-constrained, meso-dominated, and macro-connected pore systems. This classification anticipates the mobility ratio, fingering propensity, and sweep efficiency. The framework provides a compact, transport‑aware bridge from 3D pore geometry to reservoir scale behaviour, enabling quantitative screening, design and monitoring of geologic storage.