A global framework to estimate urban spatial cycling patterns based on crowdsourced data

TL;DR

This paper introduces a global framework that uses openly accessible Strava heatmaps, refined with population and POI data, to estimate and validate urban cycling patterns across cities worldwide.

Contribution

It presents a novel, low-effort method to accurately estimate urban cycling patterns using crowdsourced Strava data, validated against official cycle counts.

Findings

High correlation between Strava heatmaps and cycle counts in most cities.

POI weighting outperforms population weighting in estimating cycling patterns.

Method performs better in European cities and areas with higher cycling modal share.

Abstract

Cycling is a cornerstone of a sustainable mobility transition in cities. Cycling research depends on the data available, but it has been difficult to produce or access these data in comparable ways. Sports tracking platforms like Strava have been transformative in mass-tracking cycling patterns and data sharing through applications and data competitions. Nevertheless, access to data has remained limited. Here, we present a framework that draws on the openly accessible Strava Global Heatmap to estimate spatial patterns of relative cycling intensity on an urban scale. To refine the raw heatmap outputs, we weighted them with population and point of interest (POI) counts within varying buffers. The cycling patterns were validated in a global context, comparing the heatmap values with cycle count data from 29 cities. Both population and POI weighting delivered high correlations in most cases between the heatmap and the cycle counts, POI weighting performing better overall. The strongest associations between Strava heatmap and cycle counts were observed in European cities and along the North American east coast, with p>0.7 for all, and p>0.8 for most cities. Additionally, the performance of our approach improved with higher cycling modal share at the city level. We demonstrate that a POI-weighted Strava heatmap can accurately represent urban cycling patterns and provide estimates of categorical cycling volumes. Our approach can be applied with relatively low effort to support the planning for urban cycling if official counts are sparse. Furthermore, it can enable the use of consistent cycling data for large-scale urban cycling analyses.