**7. Conclusions**

Several problems must be addressed to apply the micro-level emission estimation method at the regional or national level by using the vehicle trajectory data collected through GPS data. The biggest problem is that it is not possible to collect the vehicle trajectory data of all vehicles running on the traffic network. The second problem is related to the task of extracting necessary data from the collected vehicle trajectory, which requires a considerable amount of data processing and operation time in calculating the micro-level emissions of individual vehicles and aggregating results by road section.

This study proposed a countermeasure to solve these problems. In this study, a micro-level emission estimation method using the massive vehicle trajectory data collected from vehicle navigation, DTG, and mobile devices was developed, which can be applicable at the regional or national level. The vehicle trajectories from collected GPS data were classified as link ID and time period to estimate the emissions and emission factors for each link at a specific time period. Vehicle trajectories for a link at a time period were divided into several groups through cluster analysis, in which the ratios of each OpMode used in MOVES were used as cluster variables for clustering similar vehicle trajectories. The choice of cluster variables is the biggest difference from the other methods for clustering vehicle trajectories. The derived values of each cluster center from clustering analysis, the OpMode distribution, can be used for calculating micro-level emissions. The center of each cluster denotes the representative vehicle trajectory for each cluster. The emissions of the cluster center can be calculated easily by using the values of the cluster center. The weighted averages of emissions of all vehicles are obtained by applying the cluster size as a weight, which represents the emissions per vehicle on the corresponding road section. They can be used as micro-level link emission factors to estimate emissions of regionalor national-level traffic networks. When vehicle type-specific traffic volume is provided, the emissions from all the vehicles on the traffic network will be easily calculated by multiplying by the micro-level link emission factors. This is the main purpose of developing the proposed method.

The proposed method is not free from computational difficulty because the operating distribution of each vehicle trajectory must be calculated to estimate link-based micro-level emission factors. Moreover, more data must be collected and analyzed in order to increase the representativeness. This requires more storage space and computing power. Fortunately, not only can the calculation procedures of the proposed method be automated but also high-performance machines can be utilized for the calculation. Thus, it is expected that the issues of storage space and computing power related to the proposed method can be addressed.

The confirmation procedure explained in the previous section is still required. However, if the proposed method were automated to accumulate data, such as navigation data, DTG data, and mobile data, for each traffic network link and update the link-based micro-level emission factors, only having traffic volume by vehicle type at the analysis time period would enable local or nationwide micro-level emission estimation to be performed efficiently.

**Author Contributions:** Conceptualization, H.H. and G.L.; Data curation, J.H.K. and H.S.; Formal analysis, H.H. and H.S.; Writing—original draft, H.H., J.H.K., and G.L.; supervision, G.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a grant(20TLRP-B148676-03) from Transportation & Logistics Research Program (TLRP) Program funded by Ministry of Land, Infrastructure and Transport of Korean governmen<sup>t</sup> and also supported by supported by the National Research Foundation of Korea gran<sup>t</sup> funded by the Republic of Korea governmen<sup>t</sup> (MSIT) (No. 2018R1C1B6006330).

**Conflicts of Interest:** The authors declare no conflict of interest.
