TROPOMI Satellite Data Reshape NO2 Air Pollution Land-Use Regression Modeling Capabilities in the United States

Abstract

Nitrogen dioxide (NO₂) pollution is associated with adverse health effects, but its spatial variability between ground monitors is poorly characterized. NO₂ column observations from the Tropospheric Monitoring Instrument (TROPOMI) have unprecedented spatial resolution and high accuracy over the globe. Land-use regression (LUR) models predict surface-level NO₂ with relevance for epidemiological and environmental justice studies. We use TROPOMI NO₂ columns in a land use regression (LUR) model to improve surface NO₂ concentration estimates over the United States. The TROPOMI LUR predictions have improved correlation with ground monitors (Adj. R² = 0.72) and bias (Mean Bias, MB = 14.2%) compared with an existing LUR using less granular NO₂ data from a legacy satellite instrument (Adj. R² = 0.54 and MB = 49%; for North America). Removing TROPOMI NO₂ from the LUR decreased R² by 29.1%, 8.1 times the impact of removing road system information. These findings reveal that novel Earth observing satellites can enhance surface NO₂ surveillance by capturing pollution variation between monitors without relying heavily on other data sources.

Improved satellite air pollution observations from TROPOMI further support that surface NO₂ can be estimated without heavy reliance on other data.