📝 SummarXiv

Abstract

Understanding human mobility during emergencies is critical for strengthening urban resilience and guiding emergency management. This study examines transitions between returners, who repeatedly visit a limited set of locations, and explorers, who travel across broader destinations, over a 15-day emergency period in a densely populated metropolitan region using the YJMob100K dataset. High-resolution spatial data reveal intra-urban behavioral dynamics often masked at coarser scales. Beyond static comparisons, we analyze how mobility evolves over time, with varying emergency durations, across weekdays and weekends, and relative to neighborhood boundaries, linking the analysis to the 15-minute city framework. Results show that at least two weeks of data are required to detect meaningful behavioral shifts. During prolonged emergencies, individuals resume visits to non-essential locations more slowly than under normal conditions. Explorers markedly reduce long distance travel, while weekends and holidays consistently exhibit returner-like, short distance patterns. Residents of low Points of Interest (POI) density neighborhoods often travel to POI rich areas, highlighting spatial disparities. Strengthening local accessibility may improve urban resilience during crises. Full reproducibility is supported through the project website: https://github.com/wissamkontar