Titolo della tesi: Data Science for vehicle emissions mitigation: data, models and simulations
The use of digital data offers promising opportunities for shaping future cities.
By tracking the vehicles' movements with high precision in space and time, GPS data offer unprecedented chances for studying vehicular mobility and, consequently, the emissions it causes.
This thesis exploits novel data and models to study vehicle emissions in urban areas, their patterns, the impact of possible reduction strategies, their connection with the different mobility behaviours and routing strategies adopted, and their connection with the urban environment.
We find strong evidence of inequalities in vehicle emissions distributions, with a few vehicles and roads holding a disproportionate amount of emissions. While this fact is already reported in the literature, the present work is the first to deepen the statistical properties of these unequal emissions distributions, which can be well described by heavy-tailed models such as a truncated power law.
We show that this information is important because intervening on the tail of such distributions (e.g., on the most polluting vehicles) makes the overall vehicle emissions reduction super-linear, while non-informed random choices reduce them linearly.
Another novelty of this thesis is represented by a first step toward a deep investigation of how the different navigation strategies adopted by the vehicles shape the resulting emissions. By exploiting a framework based on a traffic simulator and an emissions model, we find that different routing strategies have different impacts: the overall CO$_2$ emissions and the inequality in their distribution across the roads are minimised when nearly half of the vehicles follow an app’s routing suggestion and the other half do not.
Our results pose interesting challenges to address on the path toward designing truly sustainable, system-aware, routing systems and open new questions about the overall impact of individual-oriented navigation systems.
In conclusion, we present a case study showing the potentiality of a spatial approach when modelling the presence of vehicle emissions on the roads of a city in relation to the urban environment. Our experiments give useful insights into the preeminent role played by the presence of main arterial roads and the population and road network densities in determining the average level of emissions on the roads of Rome’s neighbourhoods.
Given the analyses and results presented in this thesis, we believe that our work may represent a reliable example of how to exploit new urban data analytics to transform our cities and improve their future air quality and, thus, their citizens’ life.