Spatial attenuation of noise and PM2.5 in roadside pocket parks

Pocket parks are a prevalent green infrastructure form in high-density cities, yet quantitative evidence of their buffering efficacy is scarce. This study selected 12 roadside pocket parks in Shenyang, China, collecting 544 paired, synchronous noise and PM _2.5 measurements along 5–50 m distance gradients. Traffic-noise attenuation was significant: the core slope was − 1.21 dB per 10 m (95% CI: −1.61 to − 0.83), with linearity supported within the observed range. Each additional 1,000 m² of park area improved the slope by 0.21 dB per 10 m, whereas two small parks exhibited high probability of buffering failure. ROC analysis placed the quietness cutoff at 30–40 m, though compliance beyond 40 m remained low (14–24%). By contrast, the PM _2.5 distance effect was only marginally significant and model performance unstable. These findings support a "selective environmental barrier" perspective: pocket parks provide quantifiable noise buffering but should not be positioned as air-quality refuges.