Short-term measures on air quality improvement have been commonly conducted for major events held in China, while their effectiveness was insufficiently analyzed due to deficient capability of tracking the fast-changing emissions of cities. Here we combined a machine learning algorithm, multiple satellite measurements, and an air quality model, and deduced 7-day moving averages of NOX emissions for host and neighboring cities of 11 events held from 2010 to 2023 in Yangtze River Delta (YRD). We find the benefits of short-term controls on emissions for these events have been weakened over time, due to the gradually tightened long-term controls and to a more cautious strategy of air quality improvement for recent events. The main sector of emission abatement for events shifted from power to industry and transportation, reflecting the diverse progresses of regular controls for different sectors. As a legacy, short-term controls supported better design of long-term air quality policies.