[1] We present the results from a series of numerical simulations to explore systematic k heterogeneity effects on both CO 2 trapping mechanisms and buoyancy-driven CO 2 migration. For this purpose, we generated various permutations of two-dimensional numerical models of subsurface porous media: homogeneous, random, homogenous with a low-permeability (k) lens, and isotropically/anisotropically correlated k fields. For heterogeneous cases, we used a sequential Gaussian simulation technique to generate ten realizations in each model permutation. In each simulation, the amounts of mobile, residually, and aqueously trapped CO 2 were calculated, and the spatial distributions of the CO 2 plumes were quantified using first and second spatial moments. Simulation results from both homogeneous and random k fields suggest that the amount of residually trapped CO 2 increases as the mean effective k increases. These results imply that the overall velocity distribution, which governs the sweeping area of the supercritical-phase CO 2 plume, is a critical factor for controlling residual CO 2 trapping. However, as overall velocity (or effective k field) increases, we predict that the CO 2 plume potentially reaches the caprock more quickly. In addition, results also show that the decrease of variance in ln k increases the amount of residually trapped CO 2 . In simulations of anisotropically correlated k fields, the vertical CO 2 migration distance due to buoyancy shortens as the horizontal correlation length becomes greater. In addition, as the horizontal correlation length becomes greater, residual CO 2 trapping increases and mobile CO 2 decreases because the CO 2 plume spreads farther laterally (i.e., it sweeps a larger area). In summary, results of these analyses suggest that heterogeneous k fields with greater anisotropic correlation ratios potentially maximize residual CO 2 trapping and minimize buoyancy-driven CO 2 migration. Our findings also suggest that when heterogeneous k fields have a certain structure such as a low-k lens or other hydraulic barriers (e.g., faults), the amount of residually trapped CO 2 may increase and depend more on the geometry of geological structures than the magnitude of effective k.