Cycling is a healthy and green transportation mode which can alleviate traffic congestion and reduce air pollution. To promote cycling, more bicycle infrastructures are being implemented/planned at the statewide and national levels. Several factors, including annual average daily bicycle volume (AADB) and cycling injury risk, are needed for consideration during the planning procedure. Both the cost effective crowdsourced bicycle data and the traditional manual or automated count data are useful and essential for estimating the bicycle volume and calculating the cyclist injury risk.
This project will utilize both manual and automated bicycle count collected at continuous count stations (CCS) and the data collected by using a smartphone application (i.e., Strava). First, CCS are deployed which can be used to extrapolate short duration counts (SDC). Upon installation, the CCS undergoes a validation to determine a correction factor. This factor accounts for systematic equipment error including a false positive bicycle count that is actually attributed to a motor vehicle, natural occlusion of a pedestrian sensor when a pair of pedestrians walk side-by-side, among other possible errors. The research team will determine how frequently routine validation should be performed to account for potential machine drift over time and at what level the correction factor should be applied. Second, the programmatic cost associated with calibrating systems at varying frequencies over time will be determined. Third, after correcting the automated bicycle count data from the CCS, the crowdsourced data collected from Strava application is used. Based on both the manual count data and the corrected continuous count data, as well as the crowdsourced bicycle data, a bicycle volume prediction model will be developed. A bicycle ridership map will be generated based on the modeling results. Fourth, the bicycle crash data will be collected and combined with the existed crowdsourced bicycle data utilizing the ArcGIS software. Locations that are associated with high crash frequency will be identified. An injury risk model will be developed to analyze the factors that affect the cyclist injury frequency. Risk analysis will be conducted based on the modeling results.