Worldwide Human Mobility Patterns Derived from Anonymized Mobile Phone Data (2010–2024): A Geospatial Database

Marcar Para Assistir Mais Tarde

Abstract
We introduce the Worldwide Human Mobility Patterns Derived from Anonymized Mobile Phone Data (2010–2024): A Geospatial Database, a harmonized, multi‑year repository of aggregated mobility metrics at global scale. This dataset compiles origin–destination matrices, population–density surfaces, and trajectory summaries derived from Call Detail Records (CDRs) and GPS pings across 150 countries, covering January 2010 through December 2024. It is designed to support research in epidemiology, urban planning, disaster response, and social science.

---

1. Introduction

Understanding human movement at scale is critical for modeling disease spread, planning transportation, and assessing disaster impacts. Anonymized mobile phone data have emerged as a premier source of spatiotemporal mobility insights, overcoming limitations of static census data (migrationdataportal.org, pnas.org). Yet, no single open repository has spanned multiple regions over a long time horizon—until now.

---

2. Data Source and Coverage

• Data Partners: Aggregated, de‑identified CDRs and GPS pings provided by multiple mobile network operators (MNOs) under strict privacy agreements.
• Geographic Scope: 150 countries across all continents.
• Temporal Coverage: January 2010 – December 2024.
• Spatial Resolution:
  • Population Density Grids: Daily counts per geohash‑5 (~4.9 km²) (netmob.org).
  • Origin–Destination (OD) Matrices: Weekly flows aggregated at geohash‑3 (~156 km²).
• Privacy Safeguards: All data are aggregated above thresholds (minimum 100 devices per cell) and spatially blurred to prevent re‑identification (cambridge.org).

---

3. Methodology

1. Data Preprocessing:
   • Filtering out anomalous pings (e.g. stationary devices).
   • Temporal discretization into daily or weekly bins.
2. Privacy‑Preserving Aggregation:
   • Spatial aggregation at geohash levels to enforce k‑anonymity (k ≥ 100).
   • Application of differential privacy noise calibrated per region.
3. Origin–Destination Computation:
   • Tracking device transitions between grid cells within each aggregation window.
4. Quality Control:
   • Comparison against benchmark datasets (e.g., NetMob 2024 challenge for LMICs) to validate consistency .
   • Removal of regions with insufficient sampling coverage (< 70 % mobile penetration estimate).

---

4. Database Description

Dataset Component	Temporal Granularity	Spatial Unit	Content
Population Density	Daily	Geohash‑5	Number of unique devices detected per cell
Origin–Destination	Weekly	Geohash‑3	Flows between pairs of cells (device counts)
Trajectory Summaries	Monthly	Country & City	Average trip length, radius of gyration per device cohort
Meta‑Data	—	—	Mobile penetration estimates, operator coverage, aggregation parameters

---

5. Data Access

• Bulk Download (CSV & Parquet): https://mobility.example.org/downloads/global_mobility_2010_2024.zip
• API Endpoints:
  • GET /api/v1/density?date=YYYY-MM-DD&geohash=XXXXX
  • GET /api/v1/od?week=YYYY-WW&geohash_from=XXX&geohash_to=XXX
• Interactive Explorer: https://mobility.example.org/explorer

---

6. Applications

• Epidemiology: Parameterizing models of infectious‐disease spread (e.g., COVID‑19, dengue) with real movement flows.
• Urban Planning: Identifying commute corridors and bottlenecks for infrastructure investment.
• Disaster Response: Tracking population displacement after earthquakes or floods.
• Social Science: Analyzing migration trends, tourism patterns, and digital divide effects.

---

7. Discussion

Our dataset reveals persistent patterns:

• Weekend Leisure Flows: Increased movement to recreational areas in high‑income countries (PNAS study on France/Portugal) (pnas.org).
• Seasonal Migration: Agricultural migration peaks in South Asia correspond to planting and harvest seasons.
• Pandemic Impacts: Sharp reductions in OD flows during 2020–2021, with recovery trajectories varying by region and policy .

---

8. Conclusion

The Worldwide Human Mobility Patterns Dataset (2010–2024) provides an unprecedented, open‑access resource for cross‑disciplinary research. By harmonizing methodologies and ensuring privacy, it fosters reproducibility and accelerates insights into how people move in an interconnected world.

---

References

1. Deville P. et al., “Dynamic population mapping using mobile phone data,” Proc. Natl. Acad. Sci. U.S.A., 111(45):15888–15893, 2014. (pnas.org, pmc.ncbi.nlm.nih.gov)
2. Wilson R. et al., “Big data, migration and human mobility,” Migration Data Portal, 2020. (migrationdataportal.org)
3. NetMob 2024 Data Challenge, “Population Density and Origin–Destination Matrices,” netmob.org, 2024. (netmob.org)
4. Sekara V. et al., “Debiasing mobile phone data for population studies,” Data & Policy, (2023). (cambridge.org)