What If We Use Integer Speed for Emission Modeling: Impacts of VSP Binning Methods on Sensitivity to Speed Precision

초록

This study evaluates the sensitivity of three Vehicle Specific Power (VSP)-based emission modeling methods (OpMode, standard VSP bins, and speed-specific VSP bins) to speed data precision. Using NGSIM trajectory data from US-101 freeway segments, two speed scenarios were analyzed: speeds rounded to one decimal place and to integers. CO2 emissions were derived from China VI light-duty gasoline vehicle tests, with binning methods as the sole variable. Emission factors were computed for trips (and micro-trips) and compared using a quasi difference-in-differences approach. Results reveal that all three methods are significantly impacted by speed data precision, particularly in lower speed ranges (20-30 km/h) where frequent stop-and-go traffic amplifies sensitivity. Among the methods, speed-specific VSP bins, leveraging 60-second micro-trips, produced the most detailed emission factor-speed relationships but showed the highest sensitivity to speed precision changes. This highlights its potential for nuanced modeling but underscores its dependence on high-quality data.