Multi-crowdsourced Data Fusion for Modeling Link-level Traffic Resilience to Adverse Weather Events
Songhua Hu,Kailai Wang,Lingyao Li,Yingrui Zhao,Zhenbing He,Yunpeng Zhang
DOI: https://doi.org/10.1016/j.ijdrr.2024.104754
IF: 4.842
2024-08-16
International Journal of Disaster Risk Reduction
Abstract:Climate change leads to more frequent and intense weather events, posing escalating risks to road traffic. Crowdsourced data present new opportunities to monitor and investigate road traffic changes during adverse weather. This study utilizes two types of crowdsourced data: passively collected location data (PCLD) from mobile devices and actively collected user reports (ACUR) from the WAZE app, to examine the impact of floods, winter storms, and fog on road traffic. Three metrics derived from PCLD—speed change, event duration, and area under the curve (AUC)—along with an additional metric from ACUR, user-perceived severity, are utilized to assess link-level traffic resilience. All metrics are correlated with road characteristics such as functional class, number of lanes, divider types, road geometry, and historical traffic, to analyze underlying factors and train predictors. Using the Dallas–Fort Worth (DFW) metroplex in 2022 as a case, the study finds that overall, winter storms have the most substantial impact on road traffic, followed by floods and fog. Travel demand metrics, such as POI visits and traffic volume, experience a more significant decrease than traffic speed. The impact of weather also varies by road characteristics, with higher-class roads with greater volume and higher speed experiencing more substantial changes. Road geometry, such as minimal altitude and slopes, also significantly influences road traffic resilience under floods but not during winter storms. Moreover, the study reveals that ACUR may not accurately reflect the actual impact during extreme weather events, as few users are able to travel outside to actively act as "sensors". In sum, this study evaluates a set of crowdsourced data for assessing link-level traffic resilience, designs various resilience metrics, and examines how they vary across road attributes and weather events, which are essential for guiding disaster preparedness, response, and recovery in road transportation systems.
geosciences, multidisciplinary,meteorology & atmospheric sciences,water resources