Safety effects of traffic, geometry and weather in short time intervals.

摘要

Highway safety is a complex interaction of human vehicles, roadway geometry, and an ambient environment. To understand the relationship between the contributing factors for crashes, all possible sources of information at a disaggregate level is necessary. This disaggregate level of information is collected from different sources and requires extensive efforts to manage and store the data, process the different variables, prepare samples for the models and interpret the results. In this thesis, a traditional econometric model, a logit model for crash likelihood and injury severity, was developed using data from early 2006 to mid-2008 data. An integrated dataset, consisting of crash, traffic, road geometry, and weather in observations of short time intervals and short segment lengths, provided a broad perspective to capture change of traffic, geometry, and weather in the model outcome. The model results are discussed in the light of factors such as traffic, geometry, weather and time of the day and are interpreted from the overall safety aspect on an interstate system over a two and one-half year period. The crash likelihood models capture the time of day, such as midnight to early morning, morning peak to midday, and afternoon, traffic, such as the volume of traffic in each lane in an hour and the percentage of heavy vehicles and weather conditions, such as temperature and visibility. A severity likelihood model indicated joint effects for single and multiple vehicles collisions for time of day, weather conditions, and roadway geometry. The developed econometric models can be used as a systematic analytical tool by transportation professionals to assess safety countermeasures and traffic management plan by safety researchers.