Developing LCZ morphological attenuation-based parameterization prediction models for roadside block-scale PM2.5 concentration

Abstract

The PM_2.5 concentration in roadside blocks is greatly reduced through the attenuation effects caused by block morphology. A series of block-scale rapid prediction models (LCPMs) are established based on the Local Climate Zone (LCZ) morphological parameterization scheme. Separate LCPMs are developed for upwind and downwind areas using a dual parameterization scheme, including an inherent attenuation coefficient (ζ) that represents the main diffusion obstruction capacity of each LCZ type, and additional attenuation coefficients (α, β, γ, δ, ε, θ) that modulate the inherent attenuation based on key morphological parameters. These models are resolved and validated using both mobile surveys and fixed station measurements conducted in a case region of Guangzhou under three seasons. Results show that the inherent attenuation coefficient (ζ) ranges from 0.228 to 0.959 of different seasons, and the additional coefficients (α, β, γ, δ, ε, θ) range from −0.85 to 0.96, indicating substantial differences in attenuation effects of different LCZs. Validation metrics show better fitting performance across all seasons (R²: 0.74–0.95). Based on the proposed LCPMs, mapping distributions of PM_2.5 concentration are obtained and display seasonal differences with spatially-average values varying from 27.43 μg/m³ to 59.41 μg/m³. In particular, high-density LCZs exhibit the highest PM_2.5 concentrations during peak traffic hours, with values exceeding 90 μg/m³ in winter. Then kernel density estimation, hotspot detection and spatial auto-correlation mapping results all emphasize the significant spatial heterogeneity. Finally, air quality problem areas are identified according to PM_2.5 concentration threshold and appropriate street tree planting strategies are applied to relieve air pollution.