Abstract Identifying air pollutant emissions has played a key role in improving air quality and hence the health of billions of people around the world. Central to this effort are the development of emission inventories and the mapping of air pollution using satellite remote sensing. The TROPOspheric Monitoring Instrument (TROPOMI) has been providing high resolution vertical column densities of nitrogen dioxide since late October 2018. Using the flux divergence method and a Gaussian Mixture Model, we identify peak emission hotspots over four cities in South Asia: Dhaka, Kolkata, Delhi and Lahore. We analyze data from November 2018 to March 2021 and focus on the three dry seasons (November to March) for which retrievals are available. The retrievals are shown to have sufficient spatial resolution to identify individual point and area sources. We further analyze the length scale and eccentricities of the hotspots to better characterize the emission sources. The TROPOMI emission estimates are compared with the EDGAR global emission inventory and the REAS regional inventory. This reveals areas of agreement but also significant discrepancies that should enable improvements and refinements of the inventories in the future. For example, urban emissions are underestimated while power generation emissions are overestimated. Some areas of light manufacturing cause significant signatures in TROPOMI retrievals but are mostly missing from the inventories. The spatial resolution of the TROPOMI instrument is now sufficient to provide detailed feedback to developers of emission inventories as well as to inform policy decisions at the urban to regional scale.