Development of a hybrid modelling approach for the generation of an urban on-road transportation emission inventory

The development of accurate emission inventories at an urban scale is of utmost importance for cities in light of climate change commitments and the need to identify the emission reduction potential of various strategies. Emission inventories for on-road transportation are sensitive to the network models used to generate traffic activity data. For large networks (cities or regions), average-speed models have been relied upon extensively in research and practice, primarily due to their computational attractiveness. Nevertheless, these models are myopic to traffic states and driving cycles and therefore lack in accuracy. The aim of this study is to improve the quality of regional on-road emission inventories without resorting to computationally-intensive traffic microsimulation of an entire region. For this purpose, macroscopic, mesoscopic, and microscopic emission models are applied and compared, using average speed, average speed and its standard deviation, and instantaneous speeds. We also propose a hybrid approach called the CLustEr-based Validated Emission Re-calculation (CLEVER), which bridges between the microscopic and mesoscopic approaches. CLEVER defines unsupervised traffic conditions using a combination of mesoscopic traffic characteristics for selected road segments, and identifies a representative emission factor (EF) for each condition based on the microscopic driving cycle of the sample. Regional emissions can then be estimated by classifying segments in the regional network into these conditions, and applying corresponding EFs. The results of the CLEVER method are compared with the results of microsimulation and of mesoscopic approaches revealing a robust methodology that improves the emission inventory while reducing computational burden.