Crowd-sourced data for traffic flow estimate for environmental pollution decision tools

Background: Crowd-sourced data are available for traffic estimates since travel times can provide dynamic information to model road traffic. In fact, any traffic model needs a local input regardless of the approach used. These data potentially can drive traffic policy and decisions to decrease noise or air pollution and improve citizens’ health in cities. However, travel times are limited available and need to be tuned to the context to be used for deriving traffic flows. Methods: The study selects a set of roads where travel times are available, apply transportation functions and validate results in different contexts to estimate road traffic noise. Elaborated flows are compared with acquisitions of two other methods, i.e. microwave traffic counters and AI-powered cameras for traffic recognition. Results: Studies have shown the usability of crowdsourced data to estimate noise with accuracy similar to those of other methods, but only for main and interdistrict roads and day/evening periods in which flow is enough to influence behavior of drivers. Crowd- sourced data can be suitable to identify congestion and pollution on cities’ main roads network. Conclusions: Challenges are open in using such data to drive policies: a project is ongoing to visualize noise in digital twin such that noise can be potentially combined to other urban solutions and visualized in a useful tool for policy makers. Together with other tools like ITS, crowdsourced data and digital twins can enable dynamic decisions for health. Acknowledgments: This study is carried on within PRIN 2022 OUTFIT funded by Next Generation EU, Mission 4 Component 1 CUP B53D23012960006, CNR funded Bilateral project CNR – RA (Italy -Romania) 2023-2025 and GIOVANISI INTREPID co-funded by Tuscany regional program FSE+ 2021-2027.