Mobile Monitoring and Land Use Regression Modeling of Ultrafine Particle Spatial Distribution in Beijing, China,...

Ultrafine particles (UFP, <100 nm) may pose more significant health risks compared to larger particulate matter due to higher specific surface area and potential for deposition throughout the respiratory system. However, their spatial distribution remains poorly characterized, particularly in Asian megacities such as Beijing with high population density and diverse emission sources. We developed a high-resolution land use regression (LUR) model to predict UFP spatial distribution in Beijing during August 2022-July 2023, using stop-and-go mobile monitoring data from 164 sites. Time adjustments were validated and applied to adjust short-term observations to 1-year average concentrations. Four modeling approaches were compared to address high-dimensional covariates, with linear regularization showing the best performance (R² = 0.73 and 0.65; RMSE = 971 and 1050 pt/cm³ in the full model and 10-fold cross-validation). Key predictors included population density, impervious surfaces, bus stop proximity, and Chinese restaurant density, indicating significant contributions from human activity and vehicular emissions. Predicted concentrations ranged from 1.5 to 22.5 × 10³ pt/cm³, with hotspots concentrated in the urban core and along major roadways. This first quantitative assessment of UFP spatial variability in Beijing underscores the value of dense mobile monitoring and robust statistical modeling for UFP exposure assessment in densely populated urban areas.