Urban road waterlogging occurs frequently during heavy rainstorms. Effectively identifying urban road waterlogging can help people plan their travel reasonably and thus reduce losses. By comparing the precipitation and floating car data in the waterlogging state with those in the normal state, an automaic road waterlogging detection algorithm using precipitation and floating car speed as dual thresholds is illustrated. Thresholds are chosen considering whether there are significant differences between waterlogging and normal and their values are determined by the lower confidence limits of historical data in a normal state considering crosses of peak period, off-peak period, arterial road, and secondary road. Then a case study is conducted on Shenzhen City on June 13, 2017, based on the detection algorithm. Result shows the algorithm performs satisfactorily with a 68%-90% detection rate and a 1.5%-2% false alarm rate. Therefore, we conclude that this FCD-based algorithm could aid in waterlogging detection.
展开▼
