Hackathon: Scaling data triangulation for disaster response

Building on previous work to quantify conflict-induced displacement in Ukraine, the Geographic Data Science Lab and the International Organization for Migration’s Displacement Tracking Matrix (IOM DTM) co-organised a hackathon in Berlin in May 2025. This event was part of a broader series of workshops led by Dr Elisabetta Pietrostefani, supported by the Policy Support Fund at the University of Liverpool, and funded by Research England. The hackathon aimed to test how triangulated data methodologies could be adapted to support humanitarian response across a wider range of disaster contexts.

The event explored how rapid triangulation of traditional and digital non-traditional data sources could be scaled to support humanitarian response across a wider range of crises—including those driven by conflict, climate-related disasters, and epidemics.

Supported by Snowflake Inc., the hackathon leveraged cloud-native infrastructure to manage and analyse diverse datasets. Snowflake provided tools and demonstrations that enabled participants to efficiently process and visualize complex data, facilitating deeper insights into disaster impacts.

The hackathon focused on the 2022 Pakistan floods, which affected over 33 million people. Participants worked with a rich mix of datasets, including IOM’s Community Needs Identification (CNI) survey and Meta’s Facebook mobility data, alongside flood severity metrics, demographic indicators, and climate data. This multi-layered approach allowed for a comprehensive analysis of displacement patterns and humanitarian needs.

Key findings highlighted the complementary strengths and limitations of traditional and digital non-traditional data. While traditional surveys offer detailed community-level insights, they are often delayed and geographically limited. Digital data, though broader and timelier, may lack demographic detail and be less representative in areas with low social media use. Integrating these sources required careful geographic alignment and thoughtful aggregation to ensure actionable outputs.

Participants tackled four problem statements, ranging from rapid response indicator development to predictive mobility modelling. The hackathon underscored the importance of flexible, multi-scale data systems, clear visual communication, and transparency around data limitations—critical elements for effective, data-driven humanitarian action.

Full details, including datasets and methodologies, are available at: https://pietrostefani.github.io/pop-displacement-disaster/dataaccess.html