Development of a Land-Use Regression of Hourly Surface NO2 in preparation for GeoXO Atmospheric Composition Data

The Geostationary Extended Observations (GeoXO) satellite system is the intended successor to the GOES-R Series from NOAA, and it is planned to begin operating in the early 2030s. This next generation system will be outfitted with an Atmospheric Composition Instrument (ACX) that will provide hourly observations of tropospheric trace gases and aerosols including pollutants associated with poor health; the highest priority factors for air quality monitoring in this new system include some of the pollutants most hazardous to human health such as ozone (O₃), particulate matter (PM), and nitrogen dioxide (NO₂). GeoXO will be a geostationary satellite with multiple overpasses of the United States per day in contrast to the TROPOspheric Monitoring Instrument (TROPOMI) on board the sun-synchronous polar-orbiting Sentinel-5P satellite. This satellite – launched by the European Space Agency – overpasses each place on Earth 1-2 times per day around 1:30 PM.

In this project, we evaluate the influence that GeoXO remote sensing capabilities could have for assessing air pollution-related health impacts in the United States. We do this by comparing the health effects associated with predicted NO₂ exposure at TROPOMI overpass times to predicted NO₂ exposure during daylight hours. To conduct this comparison, we develop a land-use regression (LUR) model that predicts hourly surface-level NO₂ data per month in the United States using monthly oversampled TROPOMI NO₂ columns, static land-use data including roads, population density, built environment and elevation differential, and hourly meteorological reanalysis data of temperature, boundary layer height, precipitation, and total liquid water column amount. We fuse these variables using different regression techniques including both a lasso and multi-layer perceptron regression to predict monthly surface-level NO₂ during daylight hours.

We compare the predicted hourly surface-level NO₂ to NO₂ derived just at TROPOMI overpass time – approximately 1:30 PM – to quantify how geostationary observations could better reveal how populations are exposed to pollutants like NO₂ than exposures that are reliant on data from a single overpass time. We additionally investigate the health disparity introduced from this assumption by estimating the new pediatric asthma cases and premature mortality associated with hourly daylight NO₂ exposure versus NO₂ exposure from just a single overpass time. Additionally, we discuss how data from the new TEMPO instrument – that was launched by NASA in 2023 – will influence and improve this TROPOMI-derived LUR.