Objective Taking the green belts of Jingkai highway as sampling site to analyze the traffic noise attenuation characteristics in horizontal space and to study the impact of width and community structure of green belts on traffic noise attenuation, so as to provide some references for optimizing the vegetation structure of green belts as well as programming and construction of green belts.

Method Three typical forest types were chosen, and 9 sample plots (20m×60m) were established. At each plot, the monitoring sites were set up in different width gradient perpendicular to the direction of the highway. Based on continuous monitoring, the difference among various forest types and width gradient were analyzed by using the one-way ANOVA. Pearson correlation coefficient was used to analyze the relationship between community structural indexes and noise attenuation.

Result All the three types of green belts show certain functions of noise reduction, and the functions of Cedrus deodara pure forest are the best. The optimal width of green belts to reduce noise should be 0~10 m in the selected greening modes, with the increasing of width, the ability of noise reduction gradually declines. The ability of noise reduction is composed of multiple community structural indexes, while an indivial structure index could not represent the noise reduction ability. The sectional area of green belt in same height is an important indicator affecting the noise reduction ability.

Conclusion The green belts of highway could play an important role in noise reduction. Different forest types of green belts have different ability of noise reduction. When establishing green belts, it is better to chose the species with low branching points, more branches of fine and dense. In the case of insufficient urban green space, other technical measures should be assisted to completely eliminate the influence of traffic noise on residents' life, .