Abstract: Accurate quantification of land-atmosphere exchange is crucial for the simulation of regional scale carbon flux and CO₂ concentrations. In this study, a coupling model, WRF-GHG (Weather Research and Forecasting Model with Greenhouse Gases Module) is employed to simulate regional net ecosystem exchange (NEE) and atmospheric CO₂ concentrations over the Yangtze River Delta (YRD) from July 28 to August 2, 2010. In the modeling system, several "tagged" species are defined to trace the contribution of various sources to atmospheric CO₂ concentrations. The simulated NEE fluxes and CO₂ concentrations are evaluated against in-situ measurements. The results show that VPRM can reproduce spatial-temporal variation patterns of NEE over different types of vegetation. Overall, the diurnal patterns in CO₂ concentration compared quite well with field measurements, indicating that the model captures the major features in the diurnal variations of CO₂. However, the model underestimates CO₂ concentrations by 5~15 ppm (10^-6). This is most likely due to an underestimation of anthropogenic emissions, the uncertainties of parameters defined in VPRM, and meteorological inputs. Local meteorological conditions, such as land-lake breeze, exert an important impact on CO₂ concentrations. While lake-atmospheric interactions over Lake Taihu and vegetation-atmospheric interactions over mountain areas in northern Zhejiang province play a dominate role in carbon sink, urban-related anthropogenic emissions act as a major source of carbon. The WRF-GHG modeling system demonstrated its capability of simulating local and regional variations in CO₂ fluxes and concentrations.