The spatial and temporal disaggregation models of high-accuracy vehicle emission inventory

A high-accuracy gridding vehicle emission inventory is not only the foundation for developing refined emission control strategies but a necessary input to air quality model as well. An accurate approach to the spatiotemporal disaggregation is the key step to improving the accuracy of gridding emission inventories. The existing spatial disaggregation method considers relatively fewer impact factors, lacking adequate correlation analysis among impact factors. Additionally, the existing temporal disaggregation method does not correspond with the actual travel behavior of residents. This paper proposes a multi-factor spatial disaggregation model by principal component analysis (PCAM), based on a correlation analysis of the main impact factors. Further, a new temporal disaggregation model is proposed based on the congestion delay index combined with the traffic flow fundamental model (CDITF). The results from a case study in Jinan show that the square of correlation coefficients (RSQ) between the model- disaggregated NO₂ emissions based on PCAM and the monitored NO₂ concentration increased by 34.4% compared to the traditional disaggregation model based on the standard road length, and the RSQ for CO increased by 13%; the NMD and NME of the simulation results based on CMAQ model compared to standard road length model decrease by approximately 33.7% and 35.5%, respectively. The trend of the monthly, daily, and hourly variations of NO₂ and CO emissions disaggregated by the proposed temporal disaggregation model is quite consistent with that of the monitored concentration data. The PCAM method and the CDITF proposed in this paper are more in line with the actual situation using the cumulative emissions on road sections. The vehicle emissions in Jinan are found to be concentrated in the center of each district and county and near high-grade roads. The disaggregation results in areas with large road slopes are more realistic for considering road slope factors. The trend of the monthly, daily, and hourly variations of NO₂ and CO emissions disaggregated by the proposed temporal disaggregation model is quite consistent with that of the monitored concentration data, however, the monitored concentration data presents a certain degree of time lag. The proposed spatiotemporal disaggregation model in this paper improves the accuracy of gridding vehicle emission inventory, which is of a great significance for developing precise control strategies of vehicle emissions and improving the urban air quality in general.