The atmospheric CO₂ concentrations are mainly influenced by regional sinks/sources and atmospheric transport processes,thus observations in urban area contain essential information of anthropogenic CO₂ emissions.To investigate the effect of COVID-19 on atmospheric CO₂ concentration and its anthropogenic emissions,this study chose Nanchang city as the study area and used a priori emission inventory with WRF-STILT (Stochastic Time-Inverted Lagrangian Transport) atmospheric transport model to simulate hourly CO₂ concentrations from January 24th to April 30th,2020.In accordance with the government measures to control COVID-19 epidemic,the whole study period was divided into two periods of Level 1 period (from January 24th to March 11th) and Level 2 period (from March 12th to April 30th).Results indicate the model can well capture hourly variations of CO₂ concentration,but it overestimated nighttime concentrations due to the negligence of emission source height.During Level 1 period,the observed and simulated afternoon (12:00-18:00) CO₂ mole fractions were 433.63×10^-6 and 438.22×10^-6,respectively,in which the anthropogenic emissions were 21.9% overestimated by simulation compared with observations.While during Level 2 period,the observation and simulation were very close as 432.06×10^-6 and 432.24×10^-6.The above comparisons indicate that the CO₂ emissions can be represented by a priori CO₂ emission inventory in Level 2 period,but was overestimated by 21.9% in Level 1 period,and the discrepancy was mainly due to government measures to control COVID-19 pandemic during this period.Besides,the average biological NEE enhancements were generally lower than 2×10^-6,indicating a small contribution compared with anthropogenic emissions.The higher PBLH (Planetary Boundary Layer Height) in Level 2 period also offset the enhancement in CO₂ emissions,which was also the main reason for the close observations during two periods.Our findings can provide scientific method supports for greenhouse gas emission inversions at urban scale.