A METHOD FOR RECOGNIZING RAINFALL-SENSITIVE URBAN ROADS BASED ON TRAJECTORY DATA
Keywords: Rainfall, Trajectory Data, Road Traffic, Mann-Kendall, Sensitivity Analysis, Urban Flooding
Abstract. Rainfall has a substantial impact on urban traffic. Due to climate change and urbanization, rain and associated flooding will become an increasing hazard to urban traffic. Therefore, it is significant to detect road traffic anomalies and analyze their evolution. This study focuses on rainfall-sensitive roads, which have significant changes in speed and an apparent trend of change during multiple rainfalls. Firstly, the trajectory data are converted into road traffic time-series data, and the box plot is applied to determine whether there are significant outliers in roads speed. Secondly, the Mann-Kendall Trend Test is used to judge whether there is a noticeable trend change in the rainfall process. The proposed method is applied to Nanjing City in China. Based on the trajectory data from June to August 2019, the percentage of urban roads with significant changes in rainfall events is between 30% and 50%, with 217 roads that reveal significant changes every time. A total of 24 rainfall-sensitive roads are extracted by the trend test. This study will provide additional assistance for crucial road monitoring of urban rainstorm hazards.