Likelihood and duration of flow breakdown: Modeling the effect of weather

The effect of rain on freeway flow breakdown behavior is investigated. Three aspects of flow breakdown are analyzed for rain versus no rain (clear) weather conditions. First, the probability of breakdown occurrence is examined by analyzing the distribution of prebreakdown flow rates observed immediately before the onset of traffic breakdown by using a survival analysis approach. At all study sections, a reduction with prebreakdown flow rates is observed under rain conditions compared with distributions under no rain and confirms higher breakdown likelihoods at lower flows. Log likelihood ratio tests confirm the statistical significance of differences in the prebreakdown flow rate distribution parameters under rain compared with clear conditions. Second, breakdown duration is examined by estimating a semiparametric Cox proportional hazard model. With a rain event indicator set as an independent variable, the effect of rain on breakdown duration is observed. Rain during a breakdown episode is found to increase its duration, whereas rain before breakdown does not appear to affect duration. Finally, prebreakdown and postbreakdown flow rates are compared. Overall, while a reduction in prebreakdown flow rates is observed because of rain, the flow drop between prebreakdown and postbreakdown is not much different between rain (3.9% to 12.0%) and no rain (7.8% to 12.7%) conditions.