Hydrocarbon reservoirs commonly contain an array of fine-scale structures that are below the resolution of seismic images. These features may impact flow behavior and recovery, but their specific impacts may be obscured by the upscaling process for sector and field-scale reservoir simulations. It is therefore important to identify those situations in which subseismic structures can introduce significant departures from full-field flow predictions. Using exposures of Jurassic carbonate outcrops near the village of Amellago in the High Atlas Mountains of Morocco, we have developed a series of flow simulations to explore the interactions of a hierarchical fracture network with the rock matrix of carbonate ramp strata. Model geometries were constructed in CAD software using field interpretations and LiDAR1 data of an outcrop area that is 350 m long by 100 m high. The impact of water injection on oil recovery between an injector and producer pair was investigated. Simulations were performed by a single medium reservoir simulator using a single mesh to represent fracture planes as well as rock-matrix volumes. The effects of changing scenarios for rock permeability and porosity as well as facture permeability distributions were investigated. First-order results show that the best recovery was achieved by a model with a high permeability, homogeneous matrix combined with a heterogeneous fracture network. The worst recovery scenario was given by a model with low, homogeneous permeability and high fracture permeabilities. The results highlight the importance of the permeability contrasts between the matrix and the fractures for overall recovery and the very significant impact that fractures can have on recovery by creating shadow zones and providing critical connections between permeable layers. The presence of the hierarchical fracture network developed strong fingering even in homogeneous matrix cases and evolving velocity patterns reveal competing fluid pathways among matrix and fracture routes. Insights from these models can help to develop production strategies to improve recovery from fractured carbonate reservoirs and provide an initial platform from which to extend further evaluations of different populations of conductive and baffling structures, spatial variations in wettability and capillary pressures and well positions.