This project demonstrates the quantitative relationship between weather patterns and surface traffic conditions. The aviation and maritime industries use weather measurements and predictions as a normal part of operations, and this can be extended to surface transportation. Data from two data mines on the University of Washington campus were combined to evaluate the quantitative relationship between freeway speed reduction and rain fall rate as measured by Doppler radar. The University of Washington’s Atmospheric Science department maintains an archive of Nexrad radar data, and the Electrical Engineering department maintains a data mine of 20-second averaged inductance loop data. The radar data were converted into rainfall rate, and the speed data from the inductance loop speed traps were converted into a deviation from normal performance measure. The deviation from normal and the rainfall rate were used to construct an impulse response function that can be applied to radar measurements to predict traffic speed reduction. This research has the potential to accomplish (1) prediction of non-recurring traffic congestion and (2) prediction of conditions under which incidents or accidents can have a significant impact on the freeway system. This linkage of weather to traffic may be one of the only non-recurring congestion phenomena that can be accurately predicted. This project created algorithms and implementations to correlate weather with traffic congestion. Furthermore, it may provide a means for traffic management to determine where and when to proactively place resources to clear incidents. (Author/publisher)
Abstract