Heavy-duty trucks and buses represent one of the largest sources of greenhouse gas (GHG) emissions in the United States, yet the degree to which the GHG intensity of these vehicles varies is poorly understood. This dissertation improves the accuracy of GHG emission factors for heavy-duty trucks and buses by accounting for how these vehicles are utilized given the characteristics within infrastructure networks, and it provides details on how previous emissions inventories change in response to masking these features. The work focuses on the sensitivity of GHG emission factors to vehicle speed, vehicle productivity, and infrastructure topology.

To assess the influence of vehicle speed on GHG emission factor variability, we analyze the realtime activities of heavy-duty trucks on highways as well as buses in nine transit agencies across California. As expected, the results our case studies indicate that expected GHG emission rates are higher for vehicles operating within cities than along highways. For buses, the inclusion speed-corrected emission factors could improve the accuracy of previous emission inventories for transit agencies by upwards of 62%. In contrast, speed-corrected emission factors only marginally improve emission inventories for heavy-duty trucks across the state (<5% improvement). These findings suggest that speed-resolved emission factors should be used for transit buses, while a more narrow range of GHG emission factors for heavy-duty trucks could suffice.

Next, we assess the influence of variable vehicle loading factors (trucks: payloads, buses: passenger ridership) on GHG emission factor for heavy-duty trucks and buses. We present two case studies which examine (i) the fleet of heavy-duty vehicles in the United States and (ii) the performance of single bus network in the city of San Francisco. In each case study, we rely on highly resolved vehicle productivity data. Our results uncover systematic errors associated with assuming an average vehicle loading factors when estimating the GHG emission factors for heavy-duty vehicles. When this is the case, we show that emission factor estimation biases, described by Jensen's inequality, always result in larger-than-expected environmental impacts (3% - 400%) and depend strongly on the variability and skew of truck payloads and bus ridership.

Lastly, we examine the influence of network topology (e.g., how individual portions of an infrastructure network's constituent parts are interrelated or arranged) on GHG emission factors for heavy-duty trucks participating in intermodal activities in the United States. To understand this relationship, we built a national truck and rail logistics model that quantifies the GHG emissions from freight shipments between counties. Using this model, we find that both network topology and GHG emission factors for heavy-duty trucks and intermodal rail differ across commodity types. These two factors in combination cause the GHG reductions associated with shifting freight shipments from trucks (~120 g CO2,e per ton-km) to rail (~20 g CO2,e per ton-km) to vary across the United States. As a proof of concept for the application of this model, we identify the counties with the greatest potential to reduce GHG emissions by switching freight shipments from truck to intermodal rail for two commodity types: meat/seafood and paper articles.

Collectively, this work advances the scientific community's understanding of how GHG emissions from heavy-duty vehicles vary on more resolved spatial and temporal scales, thereby improving decision-making on a case-by-case basis. Future work on this topic should be directed to integrating these findings into broader emissions mitigation decision tools and also considering other important environmental impacts.