Meta user movement data for analysing disaster-related mobility: sensitivity, bias, and usability

This study provides the first systematic evaluation of the Meta Movement Distribution Maps (MDM) dataset for analysing hazard-related mobility. Using a full year of data for Türkiye (2023), we assess MDM’s sensitivity to both routine disruptions—such as weekends, holidays, and school breaks—and acute hazard events, including a medium-scale storm and a localized wildfire. MDM, which aggregates anonymized Facebook users’ geolocation data, effectively captures shifts in immobility, short-range (0–10 km), and medium-range (10–100 km) mobility categories. Statistical analyses reveal significant associations between mobility patterns and socio-demographic factors, with population size, age, and literacy emerging as the strongest predictors. The July storm generated clear regional-scale mobility anomalies, while the August wildfire, though smaller in scale, still produced detectable short-range displacements. Limitations remain, notably demographic bias toward urban and literate users and the absence of information on the motivations behind movements. Despite these constraints, MDM offers a valuable, transparent, and near-real-time proxy for monitoring mobility during natural hazards. Its integration with complementary data sources could enhance disaster preparedness, targeted aid delivery, and recovery planning.