Geostationary Satellite Observation-Informed Modeling of Ozone Dynamics along the U.S.–Mexico Border

Geostationary atmospheric sensing provides hourly observations of urban air pollution, yet its potential to resolve fine-scale exposure dynamics and health implications remains underexplored. Here, we integrate hourly measurements from NASA’s Tropospheric Emissions Monitoring of Pollution (TEMPO) mission with machine learning to generate 1 km ground-level ozone (O 3 ) fields across urban corridors of the U.S.–Mexico border, a region characterized by complex transboundary transport and sparse regulatory monitoring. The resulting predictions capture 1 km hourly O 3 dynamics that are unresolved by conventional monitoring networks, with strong agreement against ground observations in random 10-fold cross-validation ( R 2 = 0.89, RMSE = 5.57 ppb) and independent maximum daily 8 h average (MDA8) evaluation ( R 2 = 0.93). We identify persistent intraurban ozone hotspots and pronounced temporal variability, with annual MDA8 O 3 concentrations reaching 42–50 ppb in metropolitan areas such as El Paso and Ciudad Juárez. In an exploratory analysis linking these exposure fields to sleep health in a population-based border cohort, we found each 1 ppb increase in annual MDA8 O 3 was associated with a 0.28 point increase in the Individual Pittsburgh Sleep Quality Index (PSQI) sleep medication use score among male participants (95% CI: 0.12–0.45, p < 0.001), an association that remained robust across different O 3 exposure metrics including annual daily mean O 3 and multiple sensitivity analyses adjusting for hypertension, obesity, and diabetes. These findings demonstrate how geostationary satellite systems enable high-resolution exposure characterization, reveal pollution gradients, and advance the understanding of behavioral health responses to atmospheric oxidants in border regions.