Developing a spatiotemporal NO2 model to assess extreme exposures in the Rome region

Extreme air pollution is a major risk factor for population health, and reducing its frequency and intensity has proven benefits for quality of life and life expectancy. To this end, developing updated and robust quantitative methods that can capture the impact of human behavior is essential. Therefore, this can provide public decision-makers with tools to allocate health resources based on health risks. Extreme NO2 concentrations are analyzed in the Lazio region between 2017 and 2022. Following a data-driven approach, key temporal patterns are first explored, such as seasonality, workday effects, and autoregressive structure. Local quantile regression models are then fitted at each monitoring station to capture station-specific behaviors and spatial heterogeneity. Model selection was implemented taking advantage of the quantile models’ specific metrics. Finally, the latest developments toward a comprehensive Bayesian hierarchical spatiotemporal model are presented. Future work includes scaling the model to the national level, integrating preferential sampling within the MCMC estimation, and leveraging the inverse chemical relationship between NO2 and O3 to develop multivariate quantile regression models.