In recent years, cycling has become an increasingly popular transportation mode around the world. In comparison with other popular modes of transportation, cycling is economical and energy efficient. While many studies have been conducted to analyze bicycle safety, most were limited to bicycle exposure and on-street data. This study tries to improve the current safety performance functions for bicycle crashes in urban corridors by utilizing crowdsource data from STRAVA and data on street speed management strategies.
Speed management strategies are any roadway alterations that cause a change in motorists’ driving behavior. In Florida, these speed management strategies are defined by the Florida Department of Transportation design manual. Considering the disproportionate representation of cyclists from the STRAVA data, adjustments were made to represent more accurately the cyclists based on the video detection data by developing a Tobit model. The adjusted STRAVA data was used for bicyclist exposure to analyze bicycle crashes on urban arterials. A Bayesian joint model was developed to identify the relationship between bicycle crash frequency and factors relating to speed management strategies. Other factors, such as vehicle traffic data, roadway information, socio-demographic characteristics, and land use data, were also considered in the model. The results suggest that the adjusted STRAVA data could be used as the exposure for bicycle crash analysis. The results also highlight the significant effects of speed management strategies, such as parking lots and surface pavement. These findings are expected to help engineers develop effective strategies to enhance safety for bicyclists.
The area of interest, Central Florida, includes context classifications from C1 (Natural) to C6 (Urban Core), as shown in Fig 1. The primary focus of the study will be urban arterials in Central Florida. Thus, C4 (Urban General) was selected because of the high level of interaction between cyclists and drivers. The study area also has a relatively high percentage of cyclists and has a high rate of bicycle crashes. Various types of datasets were used for analyzing bicycle safety on C4 roads. These datasets include; bicycle crash data, bicycle volume from STRAVA data, video detection data, and speed management strategies data. Other parameters, such as road attributes, traffic, land use, and socio-demographic variables, were also collected.
Fig 1. Roads categorized into different context classifications
STRAVA data was retrieved from January 2018 to July 2018 from the Florida Department of Transportation Unified Basemap Repository (UBR). Figure 2 shows the number of cyclist trips on various road segments in Orlando.
Fig 2. Total cyclist trips from STRAVA between January 2018 to July 2018
Video detection data provided the bicycle volumes at each intersection. The bicycle volumes were divided into directions, as shown in Figure 3.
Fig 3. Calculation of bicycle volumes between Intersection 1 and 2 using video detection.
The results for the Bayesian joint model are shown in Table 1. The Poisson-lognormal model had 10 significant explanatory variables within the 90% confidence level. The variables in the Poisson-lognormal model were categorized into 4 groups: traffic data, roadway information, speed management strategies, and land use.
Table 1. Bayesian joint model, including the Tobit Model and Poisson Lognormal model
For traffic data, the significant factors were traffic volume and bicycle volume. Both variables had positive coefficients, positively associated with bicycle crashes. Therefore, bicycle crashes increased significantly as the road segments began to experience higher volumes of vehicle and bicyclist traffic.
For roadway information, the significant factors were out-shoulder width, speed limit, access per mile, and median width. Both speed limit and access per mile were positively associated with bicycle crashes. Road segments that have a higher speed limit had a greater number of bicycle crashes. Similarly, road segments with more access also had a greater number of bicycle crashes. Out-shoulder width and median width were both negatively associated with bicycle crashes. As the out-shoulder and median width increase, bicycle crashes decrease. It is also important to note that median width had a greater magnitude, meaning it has a greater impact on reducing bicycle crashes.
For speed management strategies, the significant factors were on-street parking, two-way-let-turn lanes, and pavement type. On-street parking was found to have a positive association with bicycle crashes. As parking spaces per segment increased, bicycle crashes also increased. Two-way-left-turn lanes and asphalt pavement have a negative association with bicycle crashes. The more two-way-left-turn lanes there are on a road segment; the fewer bicycle crashes may occur.
The findings confirm the growing number of studies suggesting crowdsourced data could be used to estimate bicycle exposure for roadway segments. In addition, STRAVA data can be adjusted to estimate the bicycle volume better and provide a better model performance. This study contributes to transportation agencies by identifying an efficient way to accurately obtain bicycle volumes and identify roadway characteristics critical for enhancing bicycle safety. Transportation engineers and planners should focus on the relationships between the roadway characteristics to design urban roadways better.
For more information on this study, please visit https://journals.sagepub.com/doi/pdf/10.1177/03611981211036681
Comentários