Network screening and analysis of pedestrian and bicyclist crashes on Florida arterials using a corridor-level approach

Abstract

In this paper, a corridor-level approach is used to network screen and analyze pedestrian and bicyclist crashes. This approach uses less data than site-level analyses while also considering the relationship between intersections and roadway segments. 548 roadway corridors covering over 1000 centerline miles (1609 km) were identified on urban and suburban arterial roads in seven Florida counties based on context classification and lane count. From 2017 to 2021, these corridors experienced 3773 pedestrian crashes and 2599 bicyclist crashes, with about 88 % of these crashes resulting in fatalities or injuries. Three negative binomial regression models were developed to predict pedestrian crashes only, bicyclist crashes only, and both pedestrian and bicyclist crashes together (combined crashes model). Significant predictors from the models included traffic volume, speed limit, area type, intersection-related variables, and modality-related variables. Using the combined crashes model, a 0.75-mile (1.21-km) corridor was identified as the corridor with highest potential for crash frequency reduction. Examination of this corridor suggested that bicycle lanes, improved lighting, and midblock crossings could be effective countermeasures to reduce pedestrian and bicyclist crashes. Based on several performance metrics, the developed approach provided an accurate and statistically reliable way to model crashes in corridors. This corridor-level approach can help agencies expedite network screening and identify locations where many pedestrian and bicyclist crashes are likely to occur so they can take proactive actions to prevent these crashes and help keep these vulnerable road users safe.