The methodology to identify cycling network discontinuities is proposed in Chapter four of this dissertation. The proposed methodology is automated to provide a novel tool for studies to extract and include discontinuity measures in their analysis, which can be adopted by urban planners and city officials to identify discontinuity hotspot locations for further improvement as well as in the planning stage of a cycling network to eliminate discontinuity locations. The proposed approach is applied at two levels, 1) within the city: three Montréal boroughs (Appendix C); and 2) among regions: four North American areas (Chapter four), to compare and rank the discontinuity of the cycling networks. The borough level analysis in Montréal extracted seven discontinuity indicators based on the available data: end of cycling facility, change in cycling facility type, number of intersections along cycling network, change in number of road lanes, change in road class, change in traffic volume and bus stop locations. The city-level analysis among four North American cities extracted two discontinuity indicators based on the available data from open data portals: end of cycling facility and change in cycling facility type. Although 16 discontinuity indicators are proposed in Table 4-2, due to the lack of available information, only the two mentioned discontinuity indicators can be automatically extracted. This highlights one of the gaps in this field, where despite the significant attention and high importance placed on cycling by planners, city officials and researchers, the availability of data related to this field is limited. One of the basic forms of data is the geo-referenced cycling network that includes information on the number and direction of facility lanes, width of cycling facility, definitions and consistent assignments of cycling facility types, the location of cycling facility on the road: yet some areas lack even basic information such as cycling facility type and their direction of travel. For example, the city of Amsterdam, known as one of the most cycling friendly cities, provides open access to the city’s cycling network; however, the cycling facility types are defined based on functional class (commuter facility type and leisure facility type) or categorised as origin-destination class, all of which can include shared road segments, shared space with pedestrian, physically separate sections and bike lanes. Other information that is not easily accessible is road characteristics such as the number of lanes, the lane width, the road class, the location of bus stops, the parking locations, and the type of signalization. Traffic volume which includes cyclist and pedestrian flows are even more difficult to access. Road related information is usually easier to find given the years of experience in auto-based data registry (from insurance and automotive companies), and cities with a more
significant interest in promoting walking and cycling may have more robust and detailed data that can be accessed. However, the varying level of quality and detail of open access data limits the opportunity to perform studies and evaluations for analysis and comparison purposes. This limitation emphasizes the importance of recording and sharing high quality data by different governmental levels and transportation agencies. The automated method proposed in this dissertation is made available under an open source license so that cycling network discontinuity indicators can easily be extracted with available data (Nabavi Niaki et al., 2018).
One of the discontinuity indicators proposed in Chapter four is change in road lighting. The lack of nighttime road lighting audit methodologies motivated the development of easily applicable road illuminance data collection procedures in this dissertation for maintenance and safety analysis purposes. The initial intersection-based data collection method (presented in Appendix B) relies on walking across four legs of an intersection starting around 15 meters before the intersection and ending roughly 15 meters after the intersection. Since the GPS sensor is not accurate at the small intersection scale, especially in dense areas with high-rise buildings, the start time at each crossing is recorded to identify the exact set of illuminance measurements for each intersection leg. This methodology was applied to 158 intersections in Montréal for safety analysis. The methodology was then improved to eliminate the time it took to walk across intersections and record starting times at the beginning of each leg crossing. This improved methodology collects illuminance data at road link and intersections by attaching the illuminance sensor to a bike or scooter to travel along roads. This method allows for safety analysis at both intersection and road link levels, with faster data collection and processing times. The collected illuminance data can be summarized per road link or intersection by aggregating the illuminance measures within a buffer around the intersection or road link. The road lighting discontinuity indicator is calculated as the illuminance uniformity (ratio of average illuminance to minimum illuminance). The average illuminance and uniformity per location can be checked against road lighting standards to identify locations where road lighting is below standards. In Montréal, out of the 1442 downtown road links where illuminance data was collected, 48 % of links and 59 % of the 158 intersections had sub-standard lighting based on a medium pedestrian activity level (selected based on the assumed nighttime urban area pedestrian volume).