Improving GIS-based models for bicycling speed estimations

• Published in: 46th European Transport Conference, ETC 2018, Dublin, Ireland Vol. 42; pp. 85 - 99
• Main Authors: Manum, Bendik, Arnesen, Petter, Nordstrom, Tobias, Gil, Jorge
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Elsevier B.V 2019
• Subjects: Bicycle; bicycle speed estimations; bicycle speed model; Cycling; GIS; GPS; Liveable Cities; Modelling; Urban Mobility; GIS; bicycle speed model; GPS; bicycle speed estimations
• ISSN: 2352-1465; 2352-1457; 2352-1465
• DOI: 10.1016/j.trpro.2019.12.009

For reasons ranging from public health to sustainable urban development, traffic planning as well as urban design aim to increase bicycling at the cost of fossil fuel-based transport. Despite this increasing interest in bicycling, most planning practices handle bicycling schematically, applying methods that rely on fixed speed templates, paying little attention to how bicycling speeds vary depending on bicycle routes and the contexts of routes. Since travel time is essential for travel mode and route choice, more refined methods for predicting bicycling speeds should be highly useful. This paper presents a model that combines variables from two recent bikeability modelling studies. One is an urban form-based study that identified a set of variables significant for bicycling speeds. The other, based on so-called Markov-modelling for grasping speed dependence between contiguous road segments, estimates speed based on horizontal and vertical geometry of routes, and provides very detailed and continuous speed profiles along entire routes. The new combined model is estimated using GPS tracking of real bicycle trips in combination with GIS-based data of the bicycle route networks. The covariates included in the model are route geometry, intersection impedance, type of bicycle-route, kind of surface, and density of entrances to buildings along route. The latter is a proxy for slower bicycling due to vibrant urban context. The new model results in more detailed and realistic speed estimations than the previous models. This paper presents the model and some results from applying the model on bicycle routes in Gothenburg.