Realistic representation of subsurface heterogeneity is essential to better understand and effectively predict the migration and trapping patterns of carbon dioxide (CO₂) during geological carbon sequestration (GCS). Many candidate aquifers for GCS have sedimentary architectures which reflect fluvial deposition, where coarser-grained facies with higher-permeability (e.g., sandstone) are juxtaposed within finer-grained facies with lower-permeability (e.g., shale). Because the subsurface is difficult to access and sample, geostatistical methods are often used to model the spatial distribution of geological facies across different scales. We use a transition-probability based approach to simulate heterogeneous systems with binary facies distributions and the resulting petrophysical properties at the field scale. The approach produces heterogeneity fields which honor observable and physical facies attributes (e.g., volumetric proportions, mean lengths, and juxtapositional tendencies). Further, we use the associated facies-dependent properties for both relative permeability and capillary pressure relations and their hysteretic behavior. Heterogeneous facies models are used to investigate the sensitivity of different trapping mechanisms (i.e., dissolution, residual trapping) as well as CO₂ plume dynamics to variability in (1) the spatial organization and connectivity of sedimentary facies types; (2) aquifer temperature; (3) CO₂ injection period; (4) perforation length; and (5) the level of impurity, represented here as methane (CH₄) present in injected CO₂ streams. Model results show that the magnitudes of residual and solubility trapping are reduced by increasing the percentage and degree of connectivity of high-permeability facies. An increase in aquifer temperature leads to a decrease in residual trapping and an increase in solubility trapping. Results also reveal that for a given volume of injected CO₂, shorter injection times yield higher total amounts of trapped CO₂. Similarly, wells perforated over a shorter thickness of the aquifer contribute to an increase in both residual and solubility trapping. We also find that increased CH₄ concentrations in the injected CO₂ streams decrease residual trapping while increasing solubility trapping. This effect is more pronounced at shallower depths, where the pressure and temperature of the aquifer are lower.