Daniel L Goldberg, Madankui Tao, Gaige Kerr, Siqi Ma, Daniel Tong, Arlene M. Fiore, Angela Dickens, Zac Adelman, Susan Anenberg

Satellite datasets are increasingly used to evaluate NOx emissions inventories. Such studies often require the use of a chemical transport model or a complex statistical framework to account for meteorological factors that can complicate the comparison. Here, we apply a novel method to compare inventory-based emissions directly to Tropospheric Monitoring Instrument (TROPOMI) NO2 data without a chemical transport model by accounting for the two most important meteorological factors: wind speed and direction. We oversample the satellite data over multiple years and filter to use data on stagnant wind days only, and then use this satellite average to evaluate the spatial representativeness of the 1 × 1 km2 inventory-based Neighborhood Emission Mapping Operation (NEMO). In nine out of ten US cities, spatial r2-values between NEMO NOX emissions and TROPOMI NO2 exceeded 0.75. This suggests that the 108 spatial surrogates used by NEMO to spatially disaggregate NOx emissions from the U.S. county-level (5-200 km length scale) to the neighborhood level (~1 km length scale) are generally appropriate. However, this analysis also suggests some areas for improvement. Areas with dense warehouse operations appear to underestimate NOx emissions. Conversely, we find that NOx emissions in wealthy communities often appear to be overestimated using standard surrogates to disaggregate the inventory. Globally, we find spatial agreement between the EDGAR NOX inventory and NO2 observed by TROPOMI, which indicates that most NOx sources are properly included in the inventory. The inventory and satellite measurements agreed most closely in the U.S (