Towards the development of air quality digital twin

Nowadays, digital twin models are being implemented by cities to manage multiple factors through a single tool, enabling real-time assessment and prediction of policy impacts. This includes monitoring atmospheric conditions, particularly pollutant concentrations, through hyper-realistic models known as air quality digital twins. Since 2023, the city of Bologna has been developing its own digital twin project. This thesis contributes to it by evaluating the compatibility between state-of-the-art air quality models—those extracting roughness from bulk morphometric parameters—and nearly continuous aerodynamic roughness values derived from DEMs within the digital twin framework. The main goal is to assess the impact on air quality predictions of using very high-resolution geometrical data obtained from aerial orthophotos, acquired during the digital twin’s development. To this end, the ADMS-Urban dispersion model was run in several configurations, differing in methodology and spatial resolution of urban geometry inputs. Although prior studies show that changes in spatial resolution affect aerodynamic roughness and, subsequently, pollutant dispersion, this work stands out due to the refined definition of the alternative geometry. It compares pollutant concentrations and roughness values derived from bulk parameters versus high-definition Lidar data over a short summer 2023 period, when intensive field measurements and aerial scans were conducted. Roughness based on Lidar data is systematically higher, often leading to lower pollutant concentrations, mainly due to increased resolution. Yet, the high-resolution models don’t consistently outperform traditional models; improvements appear only for some pollutants and stations. While this method isn't ready for operational use, the observed spatial variability influence encourages further integration between these two approaches.