Driving environment, including road surface conditions and traffic states, often changes over time and influences 18 crash probability considerably. It becomes stretched for traditional crash frequency models developed in large 19 temporal scales to capture the time-varying characteristics of these factors, which may cause substantial loss 20 of critical driving environmental information on crash prediction. Crash predictionmodelswith refined temporal 21 data (hourly records) are developed to characterize the time-varying nature of these contributing factors. Unbal- 22 anced panel datamixed logitmodels are developed to analyze hourly crash likelihood of highway segments. The Q238 refined temporal driving environmental data, including road surface and traffic condition, obtained from the 24 RoadWeather Information System (RWIS), are incorporated into the models. Model estimation results indicate 25 that the traffic speed, traffic volume, curvature and chemically wet road surface indicator are better modeled as 26 randomparameters. The estimation results of themixed logitmodels based on unbalanced panel data showthat Q279 there are a number of factors related to crash likelihood on I-25. Specifically, weekend indicator, November indi- 28 cator, low speed limit and long remaining service life of rutting indicator are found to increase crash likelihood, Q2910 while 5-amindicator and Number ofmerging ramps per lane permile are found to decrease crash likelihood. The 30 study underscores and confirms the unique and significant impacts on crash imposed by the real-time weather, 31 road surface and traffic conditions.With the unbalanced panel data structure, the rich information fromreal-time 32 driving environmental big data can be well incorporated.
