Geospatial and statistical methods to model intracity truck crashes

In recent years, there has been a renewed interest in statistical ranking criteria to identify hot spots on road networks. These criteria potentially represent high crash risk zones for further engineering evaluation and safety improvement. Many studies also focused on the development of crash estimation models to quantify the safety effects of geometric, traffic, and environmental factors on expected number of total, fatal, injury, and/or property damage crashes at specific locations. However, freight safety, specifically truck safety, was meagerly addressed. Trucks and long-combination vehicles (LCVs) that carry approximately 70% freight have significant potential in triggering crash occurrences on roads, mostly severe crashes. Truck transportation is therefore attracting more and more attention due to its effect on safety and operational performance as well as rapid industrial growth.;Most of the past research on truck safety focused on intercity or Interstate truck trips. Intracity truck safety related studies or research was hardly pursued. The major research objectives of this dissertation are: 1) to develop a geospatial method to identify high truck crash zones, 2) to evaluate the use of different ranking methods for prioritization and allocation of resources, 3) to investigate the relations between intracity truck crash occurrences and various predictor variables (on- and off-network characteristics) to provide greater insights regarding crash occurrence and effective countermeasures, and 4) to develop truck crash prediction models.;The prioritization of high truck crash zones was performed by identifying truck crash hot spots and ranking them based on several parameters. Geospatial methods along with statistical methods were deployed to understand the relationships between geometric road conditions, land use characteristics, demographic, and socio-economic characteristics and truck crashes. Truck crash estimation models were then developed using selected on- and off- network characteristics data. To assess the suitability of these models, several goodness-of-fit statistics were computed. The geospatial methods and development of truck crash estimation models are illustrated using data for the city of Charlotte, North Carolina for the year 2008. It was found that on-off network characteristics, socio-economic characteristics and demographic characteristics that are within the 0.5-mile proximity have a vital influence on truck crash occurrence.;The findings from the research are expected to provide information and methods on identifying truck crash zones and the likelihood of a truck crash occurrence due to intracity trips and its relationship with on- and off-network characteristics of a region. Furthermore, this research is expected to aid significantly in the process of selecting meaningful countermeasures to improve safety of users on roads.