Many communities live or work in near-road environments, exposing them to traffic-related air pollution (TRAP) which can cause adverse health effects. Roadside vegetation barriers can mitigate TRAP and improve the local air quality for those communities, by promoting pollutant deposition and dispersion. However, there is a lack of proper guidelines for local communities and urban planners on how to implement urban green designs (vegetation layout, species, and dimensions) to obtain optimal pollutant reduction. Generating urban green designs is challenging due to the influence of site-specific conditions, such as the local wind speed, location of the community with respect to the highway, and vegetation properties on the barriers’ effectiveness to mitigate TRAP. Therefore, evaluating urban green designs based on local conditions is necessary. We used computational fluid dynamics (CFD) to assess novel urban green designs under various urban conditions to provide general guidelines. Those include combining vegetation barriers with low-cost impermeable solid structures (LISS), e.g. thin wooden or glass fences, to enhance their pollutant reduction capabilities. We also investigated how the growth of roadside vegetation barriers influenced TRAP mitigation in near-road environments by assessing them at different growth stages. Based on our findings, we provided general recommendations on the ideal maturity of the barrier and how to maintain it. We also developed tools to help provide timely urban green design recommendations for local communities based on their local conditions. Urban planner and local communities lack the expertise and resources to use CFD to assess vegetation barriers. Hence, we developed a multi-region Gaussian plume dispersion physics-based model, and a machine learning data-driven model, to obtain downwind pollutant concentrations for roadside vegetation barriers. Both models can be integrated into an easy-to-use tool for local communities to utilize in order to obtain the optimal barrier design based on their local user input such as available space to grow vegetation, local wind speed, and species constraints. This tool would empower local communities as they strive to obtain site-specific urban green design recommendations.