The development of low-cost sensors and novel calibration algorithms provides new hints to complement conventional ground-based observation sites to evaluate the spatial and temporal distribution of pollutants on hyperlocal scales (tens of meters). Here we use sensors deployed on a taxi fleet to explore the air quality in the road network of Nanjing over the course of a year (October 2019–September 2020). Based on GIS technology, we develop a grid analysis method to obtain 50 m resolution maps of major air pollutants (CO, NO 2 , and O 3 ). Through hotspot identification analysis, we find three main sources of air pollutants including traffic, industrial emissions, and cooking fumes. We find that CO and NO 2 concentrations show a pattern: highways > arterial roads > secondary roads > branch roads > residential streets, reflecting traffic volume. The O 3 concentrations in these five road types are in opposite order due to the titration effect of NO x . Combined the mobile measurements and the stationary station data, we diagnose that the contribution of traffic-related emissions to CO and NO 2 are 42.6 % and 26.3 %, respectively. Compared to the pre-COVID period, the concentrations of CO and NO 2 during the COVID-lockdown period decreased for 44.9 % and 47.1 %, respectively, and the contribution of traffic-related emissions to them both decreased by more than 50 %. With the end of the COVID-lockdown period, traffic emissions and air pollutant concentrations rebounded substantially, indicating that traffic emissions have a crucial impact on the variation of air pollutant levels in urban regions. This research demonstrates the sensing power of mobile monitoring for urban air pollution, which provides detailed information for source attribution, accurate traceability, and potential mitigation strategies at the urban micro-scale.