How bike-friendly is your street? Modelling bikeability using open data

Description

An increase in the number of people that use the bicycle instead of motorised modes is associated with positive impacts on physical and mental health, climate, safety, and economy, among others. Consequently, stakeholders around the world acknowledge the need to promote cycling as utilitarian mode of transport. This requires a bikeable street network, i.e. a street network that is convenient and safe for cycling. We identified the need to lower the barriers for decision makers in public administration and the private planning sector to assess the bikeability of street networks, without requiring GIS or data science expertise. A comparable, automated assessment of the street network may foster better informed and targeted improvements of infrastructure. Therefore, we have developed a model and open-source toolbox for bikeability assessment which utilizes exclusively open and widely available data sources such as OpenStreetMap. We now intend to conduct a validation of the model using an interactive approach.The proposed bikeability model infers various quality indicators from network data by mapping attribute values to a universal numerical scale that represents suitability for cycling. These indicators range from characteristics of the street itself, such as the presence of separate bike lanes, the condition of the street surface, the gradient, and the speed limit, to characteristics of the street environment, such as greenness and noise. All quality indicators are then combined into a single quantitative bikeability index for each segment by computing a weighted average. For both stages of modelling -the numerical mapping of attribute values to quality indicators as well as the weighting of each indicator -the parametrisation is essential for generating viable results. Our proposed bikeability model comes with a set of parameters that is derived from previous research on cyclists' perception of safety and comfort, as well as on the analysis of crash reports. Thereby, we model bikeability as representation of commonly perceived infrastructure suitability for cycling. We are aware that preferences may be subjective and even depend on purposes and framing conditions for individual trips. Therefore, the open-source workflow allows easy customization of all parameters to adjust the model accordingly. Still, the generic assessment of bikeability as it is commonly perceived by cyclists provides valuable
input for mobility planning and decision support. Our proposed model and its open-source implementation have been successfully utilised as basis for numerous applications such as bicycle routing, gap detection, assessment of cycling accessibility, and agent-based modelling of cyclist mobility. We see great potential in not only presenting but interactively assessing our proposed model together with the experts attending the 7thAnnual Meeting of the Cycling Research Board. We envision to conduct a live-survey with acquiring expert ratings on bikeability for exemplary infrastructure situations, followed by an on-the-fly analysis and comparison to the model results. Various street configurations illustrated through photographs will be assessed during the expert ratings in a fun and interactive way. Consequently, we aim for presenting live validation results for our model within the session. Our intention is to use the collected expert feedback to further improve and extend the bikeability model. As we share the software and its source code under MIT license, the validated and / or improved model will be available to everyone and can be used as a basis for further analyses and applications. You find the NetAScore software at https://github.com/plus-mobilitylab/netascore.

Period	25 Oct 2023
Event title	Cycling Research Board Annual Meeting 2023
Event type	Conference
Location	Wuppertal, Germany, North Rhine-WestphaliaShow on map
Degree of Recognition	International