Near road air quality is a growing concern for urban developers and transportation engineers as exposure to common vehicle emissions has been linked to numerous adverse health effects. Roadway design is being considered as one potential solution for mitigating exposure for those living and working nearby. This work examines the effectiveness of various roadway configurations, such as elevations and depressions, as well as the presence of solid barriers, such as those erected to reduce noise pollution, and vegetation barriers. Various experimental work has shown the potential benefits of these features. However, there is still a lack of mechanistic understanding of how they impact the air flow and pollutant transport in the near-road environment. In this work, we propose that Computational Fluid Dynamics (CFD) models can be utilized to further our understanding in this regard. To this end, we first use existing experimental data to validate our CFD model. Once validated, we can use the computational model to observe any number of other configurations. We find that flat terrain often has worse pollutant concentrations at grade than any of the other roadway configurations. Solid barriers near roadways in particular can reduce ground level concentrations by up to 80%. However, there are potential drawbacks, such as higher concentrations at higher elevations and higher on-road concentrations. Furthermore, vegetation barriers often have mixed results. They enhance particle deposition, but often lead to increased concentrations due to lower convective and turbulent transport. In the latter part of this work, we aim to use the knowledge gained from our computational simulations to help create a simple parameterized model to characterize pollutant transport near roadside barriers. CFD models are computationally very expensive and require extensive understanding in order to properly use. We propose a modification to Gaussian plume dispersion models to account for the impact of both solid and vegetative barriers. This model is shown to be only slightly less accurate than the CFD model while saving large amounts of computational time. In doing this we hope to make our findings more accessible to those who will need to utilize them for policy making decisions.