Project description

Intersection counts are ubiquitous measures in transportation planning research and practice. In particular, they are frequently normalized by area to calculate intersection density, the most common measure of compact street network design in planning practice for sustainable transport, active travel, and (alongside connectivity) networks resilient to perturbation. However, due to the nature of typical street network data (centerlines) and the typical tools used to count intersections (desktop GIS), traditional methods of counting intersections overcount by 16% on average, but unevenly so in different kinds of places and as high as by 33% in urban areas in certain countries such as Spain and Australia. Worst of all, the nature of this bias means that individual neighborhoods with the most car-centric complex intersections are the most misrepresented as being more compact and fine-grained than they are in reality. Researchers and practitioners rely on intersection counts and densities as foundational inputs and require accurate measurements to represent the real-world truly and plan resilient and sustainable transportation infrastructure accordingly. This project will help solve the longstanding problem of intersection count bias by 1) developing new algorithms to automatically and correctly calculate intersection counts and densities anywhere in the world, 2) distributing them in an easily reusable tool for others to utilize, 3) conducting a worldwide empirical assessment of traditional intersection counting methods' bias to quantify the importance of measurement bias, and 4) assessing this bias's impact on resilience simulations. In transportation planning, innumerable downstream models and reports rely on intersection counts as input data. A full accounting of this bias and better methods to overcome it are necessary for data-driven, evidence-based planning for sustainable transportation networks that support active and resilient travel.

Research seminar highlights video