A net-zero strategy for counterbalancing anthropogenic carbon dioxide (CO 2 ) emissions has become a key aspect in achieving the global decarbonization initiative by 2050. In particular, carbon geosequestration and enhanced oil recovery (EOR) in hydrocarbon reservoirs is a cost-effective strategy for attaining net-zero emissions. However, EOR and the ability to capture residual CO 2 are greatly affected by the presence or absence of oil layers along with the pore size distribution and fluid displacement, which drastically impact the reservoir-scale flow. Thus, the present study makes systematic use of the versatile in situ two-dimensional (2D) nuclear magnetic resonance (NMR) T 1 −T 2 imaging technique to visualize the fluid occupancy of the pore network in dolomite rock and uses the T 1 /T 2 ratios to assess the physicochemical properties thereof, including the microscopic wettability with respect to the pore-space fluids, after each process step. Specifically, the T 2 relaxation time is measured to demonstrate the displacement processes and to evaluate the trapping behavior and associated rock/fluid interactions at the pore level, which are closely correlated with the fluid flow behavior. Thus, in the tested water-wet dolomite, the strongest-wetting phase (i.e., water) is shown to occupy the smallest pores, where it is held by strong capillary forces, while the nonwetting CO 2 phase occupies the large pores and the intermediate-wetting oil reside in the intermediate-sized pores. As a result, 16% of the residual CO 2 trapped in large pores, while 71% of the oil is recovered, which is comparable to previous results obtained for analogue two-phase flow in water-wet dolomite samples, although some details differ and the displacements are noticeably complex. These findings have significant impacts on net-zero emissions, the budgets and storage capacities of CO 2 -EOR project schemes, and CO 2 geosequestration. In the latter context, the present findings also greatly increase the geo-storage integrity over reservoir-scale implementations.