To improve the accuracy and applicability of MOVES, this study incorporated local driving cycles into MOVES to calculate vehicle emissions. Based on massive vehicle activity data collected in Xi’an, 20 driving cycles for light-duty vehicles were developed by using the index of VSP distribution. The prediction accuracies of this approach could be improved by 37.16%, 9.94%, 24.24%, 3.57%, 18.76%, 18.45%, 5.86%, 3.44%, and 8.68% for NOx, THC, CO, CO2, PM2.5, PM10, VOCs, SO2 and NH3 when compared with the method only using MOVES. Further, calculation errors were within 15% against the results by the method of “MOVES + Real driving data”. However, the prediction errors were still high when compared with measured emission data. Thus, many reasons were discussed and several suggestions were proposed for further studies. This research approach could be generalized to other vehicle categories and the developed cycles could also be applied to other cities in China. Keywords: Vehicle emission; Emission model; Driving cycle; Average speed

CICTP