Overview
The global challenge of internal displacement, exacerbated by climate-induced natural hazards, demands more timely and granular data than traditional survey methods alone can provide. In 2024, floods affected an estimated 40 million people worldwide — making them the most common disaster type globally — yet decision-makers in the critical first days after an event often face a near-total absence of reliable displacement figures.
FloodTraces addresses this gap by developing a replicable data pipeline that integrates digital trace data — GPS mobility records, social media signals, satellite imagery, and Cloudflare connectivity estimates — with traditional sources such as IOM’s Displacement Tracking Matrix surveys. The goal is not to replace established methods, but to triangulate and augment them: producing estimates that are faster, spatially finer, and available earlier than field surveys alone can deliver.
The project makes three complementary contributions:
Timely insights when traditional data are scarce.
A replicable pipeline that delivers actionable displacement information in the critical first weeks after a flood, before survey operations are fully operational.
Triangulation and validation of traditional sources.
Cross-referencing GPS signals and other digital traces with IOM DTM data to strengthen the reliability and coverage of displacement assessments.
Anticipatory action in recurrent flood contexts.
Spatially and temporally granular data to inform preparedness in areas with a known history of repeated flooding.
Case Studies
As a proof of concept, the pipeline is applied to three flood events:
| Country | Event | Analysis Window |
|---|---|---|
| Indonesia | West Sumatra floods & landslides | February–March 2024 |
| Colombia | Floods & conflict (flood signal hard to isolate) | October–November 2024 |
| Pakistan | Seasonal monsoon floods | 2022 event; historical analysis |
Each case study produces a Digital Traces Situation Report (SitRep) — a concise, policy-oriented document showing displacement estimates derived from the GPS pipeline, key validation against OCHA and IOM benchmarks, and a brief summary of data limitations. The Indonesia and Colombia SitReps are presented at the Day 1 policy workshop; the Pakistan case forms the basis for the hackathon.
Partners
This work is led by the Geographic Data Science Lab at the University of Liverpool, and supported by Imago: Data Service for Imagery conducted in close partnership with:
- FCDO (Foreign, Commonwealth & Development Office) — policy partner and advisory role, including the Humanitarian and Stabilisation Operations Team (HSOT). FCDO is not a funder of this work.
- IOM Displacement Tracking Matrix (DTM) — operational data partner and validation benchmark