Given the sparsity of traditional ground-based air quality monitoring, satellite observations are increasingly being used to study the unequally distributed impacts of air pollution. Exciting advances in satellite-based remote sensing of atmospheric chemistry, particularly from geostationary orbit, are expected to accelerate these efforts. It is imperative that we consider when, and to what extent, space-based retrievals are representative of the processes relevant to human exposures. Using a combination of in-situ and remote sensing observations complemented by high-resolution modeling, this presentation will highlight the interaction of air pollution and meteorology in complex urban systems and examine the implications for satellite-based measurements. We specifically leverage ground-based remote sensing measurements of nitrogen dioxide in Boston and Salt Lake City to study the link between total column abundance (as would be retrieved from satellites) and surface concentrations relevant to ambient exposures. Our work confirms that satellite observations are uniquely challenged under certain conditions and in specific settings. We show that gradients in pollution inferred by the total column abundance are not always representative of, and may even be in opposite directions to, gradients in pollution observed at the ground. Opportunities to address these challenges and advance the understanding of urban air quality with remote sensing will be discussed.