[Space-Time Estimations and Mapping of PM_2.5 Fine Particulates Based on Multi-source Data]

Area of Science:

• Environmental Science
• Atmospheric Science
• Geospatial Analysis

Context:

• Particulate Matter (PM2.5) pollution is a severe environmental and social issue in China.
• Public concern regarding PM2.5 levels is widespread.
• Accurate estimation of ground-level PM2.5 is crucial for public health and environmental policy.

Purpose:

• To estimate ground-level PM2.5 concentrations in eastern China during winter.
• To evaluate the spatial and temporal characteristics of PM2.5 exposure.
• To introduce and validate a new hybrid technique combining Bayesian Maximum Entropy (BME) and Geographically Weighted Regression (GWR).

Summary:

• A novel BME-GWR technique accurately estimated PM2.5 from satellite AOD, topography, meteorological data, and emissions.
• The hybrid model achieved a high cross-validation R² of 0.92 and low RMSE of 8.32 μg·m⁻³.
• High PM2.5 concentrations were observed in North China, the Yangtze River Delta, and Sichuan Basin, with a trend migrating inland from coastal areas.

Impact:

• The BME-GWR model significantly improved prediction accuracy compared to GWR alone.
• Identified key regions and temporal patterns of severe PM2.5 pollution in winter.
• Provides a robust methodology for PM2.5 exposure assessment in data-scarce regions.