INVESTIGATION OF ROADWAY GEOMETRIC AND TRAFFIC FLOW FACTORS FOR VEHICLE CRASHES USING SPATIOTEMPORAL INTERACTION

Gill, G.; Sakrani, T.; Cheng, W.; Zhou, J. · 2017 · DOAJ

DOI: 10.5194/isprs-archives-XLII-2-W7-1163-2017

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the critical need for accurate crash prediction models to support traffic safety management, including network screening and countermeasure prioritization. Motivated by the significant monetary and emotional costs of roadway crashes, the research investigates how roadway geometric and traffic flow factors influence vehicular crashes. The authors aim to improve upon traditional models by exploiting the spatial and temporal nature of crash data, thereby accounting for heterogeneities that standard approaches often overlook. The methodology employs a univariate spatiotemporal Poisson lognormal model with random effects, developed within a Full Bayesian framework. Data were sourced from the Highway Safety Information System (HSIS), covering 279 rural freeway segments in California over a five-year period (2007–2011). The model incorporates time-varying coefficients to capture yearly variations in explanatory variables and uses a conditional auto-regressive (CAR) model to account for spatial correlations among neighboring segments. Model estimation was performed using WinBUGS software with Markov Chain Monte Carlo (MCMC) algorithms, utilizing two chains of 45,000 iterations with a 5,000-sample burn-in period. Key variables included lane width, posted speed limit, pavement type, and Average Annual Daily Traffic (AADT). The results demonstrate significant correlations between specific geometric and traffic factors and crash frequency. Lane width generally showed an inverse relationship with crash risk, where wider lanes (9, 10, and 11 feet) decreased the likelihood of crashes due to increased lateral separation between vehicles. However, lanes wider than 12 feet exhibited a positive correlation with crashes, potentially because drivers perceive greater maneuvering safety and drive less cautiously. Posted speed limits were positively correlated with crash risk, as higher speeds reduce driver reaction time. Pavement type also significantly influenced safety; driving on bridge decks or asphalt concrete pavement less than seven inches thick heightened crash chances. AADT showed a weak correlation, likely due to lower traffic volumes on the rural roads studied. The time-varying coefficients confirmed that parameter estimates shifted annually, highlighting the importance of random parameter models. The significance of this work lies in its demonstration that spatiotemporal models provide more precise crash estimates by accounting for spatial and temporal variability. This improved accuracy aids safety agencies in identifying hotspots and allocating funds for effective countermeasures. The study concludes that incorporating complex correlation structures leads to less biased inferences and better safety performance outcomes. Future research is suggested to explore different explanatory variables, spatial levels, and crash severity classifications to further refine safety analysis.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• incidence prevalence
• pre crash contributing factors
• crash typology
• roadway lighting effects

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: crash risk outcomes