Urban viaducts influence air pollution due to their elevated emission sources and complex wind fields. This study investigates the three-dimensional distribution of a typical traffic pollutant (i.e., PM_2.5) alongside a viaduct in Guangzhou, China, by using a movable measurement platform. Based on the measured data, the study explores the statistical relationship between the pollutant and the influencing factors. A hybrid method with validation using the measured data is developed to model the dynamic variations in the PM_2.5 distribution. The distribution of PM_2.5 is studied in an asymmetric street canyon with a viaduct under different wind speeds (i.e., 0.7 m/s, 4 m/s, 10 m/s), and the influence of canyon structures with different height–width ratios on the PM_2.5 dispersion is evaluated. The study finally proposes a series of strategies for reducing roadside pollution along the viaduct. The well-validated hybrid model is also used to determine the optimal pollution reduction strategy for a specific scenario. This work not only reveals the dynamic distribution law and key influencing factors of traffic pollutants in urban viaducts but also provides an important practical basis for traffic infrastructure construction and roadside building design to effectively alleviate air pollution from traffic emission sources.