Developing Operating Mode Distributions from Sparse Trajectories for Emission Estimation

Abstract

To estimate local emissions accurately, it is necessary to develop site-specific operating mode (OpMode) distributions from local vehicle activity data at the 1-second interval (field continuous data/field data). Nowadays, large amounts of vehicle activity data are collected for navigation or traffic state monitoring. However, these data are not at the 1-second interval, they are at 2-second interval and above, and named the sparse vehicle activity data/sparse data in the paper. These data cannot be used to develop local OpMode distributions. Thus, the purpose of this study is to investigate how to use the sparse data at different time intervals, from 2 to 60 seconds, to develop OpMode distributions and emission factors (EFs) accurately. Analyzing the variations of acceleration rate, three trajectory reconstruction methods were developed to interpolate the sparse data into 1-second interval data: Cubic Spline Interpolation (CSI), Interpolation with Random Acceleration (IRA), and Cubic Spline + Interpolation with Random Acceleration (CSI+IRA). An application method of the sparse data was proposed by comparing the EFs of the field continuous data and those of the sparse data before and after the interpolation. This application method included three scenarios to accurately estimate CO₂, CO, HC, and NO_X emissions: (a) the sparse data at 2- to 3-second intervals can be directly used; (b) the sparse data at 4- to 5-second intervals can be used after interpolation by CSI; and (c) the sparse data from 6- to 60-second intervals can be used after interpolation by CSI + IRA.