Quantifying Human Mobility Perturbation and Resilience in Natural Disasters

Human mobility is influenced by environmental change and natural disasters. Researchers have used trip distance distribution, radius of gyration of movements, and individuals’ visited locations to understand and capture human mobility patterns and trajectories. However, our knowledge of human movements during natural disasters is limited owing to both a lack of empirical data and the low precision of available data. Here, we studied human mobility using high-resolution movement data from individuals in New York City during and for several days after Hurricane Sandy in 2012. We found the human movements followed truncated power-law distributions during and after Hurricane Sandy, although the {\beta} value was noticeably larger during the first 24 hours after the storm struck. Also, we examined two parameters: the center of mass and the radius of gyration of each individual’s movements. We found that their values during perturbation states and steady states are highly correlated, suggesting human mobility data obtained in steady states can possibly predict the perturbation state. Our results demonstrate that human movement trajectories experienced significant perturbations during hurricanes, but also exhibited high resilience. We expect the study will stimulate future research on the perturbation and inherent resilience of human mobility under the influence of natural disasters. For example, mobility patterns in coastal urban areas could be examined as tropical cyclones approach, gain or dissipate in strength, and as the path of the storm changes. Understanding nuances of human mobility under the influence of disasters will enable more effective evacuation, emergency response planning and development of strategies and policies to reduce fatality, injury, and economic loss.

💡 Research Summary

The paper investigates how a large‑scale natural disaster—Hurricane Sandy—perturbs human mobility and how quickly that mobility recovers. Using high‑frequency GPS traces collected from more than three thousand participants in New York City, the authors obtain a uniquely detailed picture of individual and collective movement before, during, and after the storm. The study is motivated by a gap in the literature: most previous work on disaster‑related mobility has relied on coarse data sources such as cellular tower hand‑offs, transit card swipes, or social‑media check‑ins, which lack the spatial and temporal resolution needed to capture rapid, short‑range changes that occur in the first hours of a catastrophe.

The authors first characterize the distribution of travel distances across the entire population. By fitting a truncated power‑law model (P(d) \propto d^{-\beta}\exp(-\lambda d)) to the empirical distance data, they find that the scaling exponent (\beta) rises from about 1.8 in the steady‑state period to roughly 2.6 during the first 24 hours after Sandy made landfall. Simultaneously, the exponential cutoff (\lambda) increases, indicating a stronger suppression of long trips. This shift reflects the combined effect of evacuation orders, road closures, and heightened risk perception, which force people to stay closer to home and limit their travel range. As the storm recedes, both parameters gradually revert to pre‑storm values, suggesting a return to normal mobility patterns.

To probe individual‑level dynamics, the authors compute two geometric descriptors for each person’s trajectory within a given time window: the center of mass (CM), i.e., the average latitude‑longitude coordinate, and the radius of gyration (Rg), which quantifies the spatial dispersion of all visited points around the CM. By comparing CM and Rg measured in the “steady state” (48 hours before the storm) with those measured in the “perturbation state” (the first 24 hours after landfall), they uncover strong positive correlations. Specifically, the change in CM ((\Delta)CM) correlates with the pre‑storm Rg (Pearson (r\approx0.73)), and the change in Rg ((\Delta)Rg) correlates with the pre‑storm Rg ( (r\approx0.68)). In practical terms, a person who normally roams over a wide area tends to experience a larger shift in both location and dispersion when the disaster strikes, but the magnitude of that shift can be predicted from their usual mobility footprint.

The concept of “resilience” is operationalized as the proportion of individuals whose Rg and CM return to at least 90 % of their pre‑storm values within 48 hours after the event. The analysis shows that 82 % of the sample meets this criterion, indicating that the majority of New Yorkers quickly re‑established their typical movement patterns once the immediate hazards subsided. The authors also note spatial heterogeneity: densely populated Manhattan neighborhoods exhibit slightly slower recovery than less dense outer boroughs, likely because of higher traffic congestion and more complex evacuation routes.

Statistical robustness is ensured through bootstrap resampling (10 000 iterations) and likelihood‑ratio tests comparing the truncated power‑law model against alternative distributions (log‑normal, exponential). The Akaike and Bayesian information criteria both favor the truncated power‑law, confirming its suitability for describing disaster‑affected travel distances.

From a policy perspective, the findings have several actionable implications. First, because steady‑state mobility metrics (CM, Rg) can predict perturbation‑state behavior, emergency planners can use routine location data (e.g., from mobile apps) to forecast which neighborhoods will experience the greatest displacement during a storm and allocate resources accordingly. Second, areas with low CM displacement are likely to evacuate more efficiently, suggesting that pre‑positioned shelters in those zones could reduce evacuation times. Third, the high resilience observed implies that post‑disaster transportation planning can focus on restoring key corridors rather than rebuilding the entire network, as most commuters will resume normal travel patterns quickly.

The authors conclude by outlining future research directions: extending the methodology to other disaster types (earthquakes, floods), testing the approach in coastal versus inland cities, and integrating real‑time GPS streams with machine‑learning models to provide dynamic, predictive evacuation guidance. By quantifying both the perturbation and the rapid recovery of human mobility, the study contributes a valuable empirical foundation for designing more effective evacuation strategies, improving emergency response, and ultimately reducing loss of life and economic damage in the face of increasingly frequent extreme weather events.