The movement ecology of pedestrian crowds

A quantitative understanding of human mobility is one of the most important scientific challenges of our times. In this thesis, we investigate pedestrian mobility at large crowded events, based on movement data derived from Wi-Fi detections of smart phones in the Johan Cruijff Arena in Amsterdam. Mass gathering events are unique occasions where people spend many consecutive hours of time, and where the only means of displacement is walking. The spatial and temporal scales of these events correspond to time spans of < 24 hours, and spatial areas < 1 square kilometer.
We propose solutions to various technical challenges presented by the data, and propose new approaches to the reconstruction and analysis of movement tracks. We test several existing hypotheses and mathematical models from the fields of human mobility, movement ecology, and epidemiology of infectious diseases. Our main contribution is the empirical observation and quantitative analysis of intermittent movement behaviour of visitors of large, crowded entertainment events. We look at the statistical properties of the displacements and investigate to what extent they resemble Lévy walks. Using state-space models, we infer different behavioural states of movement and rest from the movement tracks and find that the periods of rest have ‘bursty’ characteristics. We investigate the effect of the observed intermittency on the diffusion of a COVID-like infection through the crowd. Using simulations we study the effect of the intermittent movement behaviour and its interaction with a time-dependent infection probability.