A Metamodeling Technique for Exploring the Correlation between Mobility and Environmental Factors at Signalized Intersections

Given the limitations of built-in emission estimation modules within current traffic simulations and signal optimization tools, this study aims to explore how the environmental impacts of transport are related to mobility measurements at signalized intersections based on high-fidelity simulation. The metamodeling-based method - involving experimental design, high-fidelity simulations, and multivariate regression analysis - is developed in this paper. The high-fidelity simulations - from microscopic traffic modeling and emerging emission estimator - provide the flexibility of utilizing various intersection types, vehicle types, and other characteristics such as drivers' behaviors, fuel types and meteorological factors. The multivariate multiple regression analysis provides large gains in expected prediction accuracy by taking the correlations between the response variables into account. The case studies demonstrate the operability of the proposed methodology and also set up the base for extensive application of simulation optimization to sustainable traffic operations and management.