Abstract
BACKGROUND: Machine-learning algorithms are becoming popular techniques to predict ambient air PM(2.5) concentrations at high spatial resolutions (1 × 1 km) using satellite-based aerosol optical depth (AOD). Most machine-learning models have aimed to predict 24 h-averaged PM(2.5) concentrations (mean PM(2.5)) in high-income regions. Over Mexico, none have been developed to predict subdaily peak levels, such as the maximum daily 1-h concentration (max PM(2.5)). OBJECTIVE: Our goal was to develop a machine-learning model to predict mean PM(2.5) and max PM(2.5) concentrations in the Mexico City Metropolitan Area from 2004 through 2019. METHODS: We present a new modeling approach based on extreme gradient boosting (XGBoost) and inverse-distance weighting that uses AOD, meteorology, and land-use variables. We also investigated applications of our mean PM(2.5) predictions that can aid local authorities in air-quality management and public-health surveillance, such as the co-occurrence of high PM(2.5) and heat, compliance with local air-quality standards, and the relationship of PM(2.5) exposure with social marginalization. RESULTS: Our models for mean and max PM(2.5) exhibited good performance, with overall cross-validated mean absolute errors (MAE) of 3.68 and 9.20 μg/m(3), respectively, compared to mean absolute deviations from the median (MAD) of 8.55 and 15.64 μg/m(3). In 2010, everybody in the study region was exposed to unhealthy levels of PM(2.5). Hotter days had greater PM(2.5) concentrations. Finally, we found similar exposure to PM(2.5) across levels of social marginalization. SIGNIFICANCE: Machine learning algorithms can be used to predict highly spatiotemporally resolved PM(2.5) concentrations even in regions with sparse monitoring. IMPACT: Our PM(2.5) predictions can aid local authorities in air-quality management and public-health surveillance, and they can advance epidemiological research in Central Mexico with state-of-the-art exposure assessment methods.