An open population displacement risk model built on physical and socioeconomic drivers of displacement

Every year millions of people are displaced by extreme events around the world. The factors that cause someone to leave their home during a disaster are complex and interacting, and they are different between countries, cultures and socioeconomic groups.

However, the data on events and displacement can be noisy and uncertain, and building any kind of global model of disaster displacement is a challenge, although a necessary one. In this work we use theory from migration and displacement studies, both quantitative and qualitative, to constrain and guide the design of improved global displacement risk models for earthquakes and tropical cyclones. The model describes population displacement as a process driven by regionally-varying socioeconomic factors, not just loss of physical housing.

This work builds on an existing global probabilistic displacement risk model built by our consortium. We identify the most relevant drivers of displacement by modelling historic displacement events and selecting from a larger set of socioeconomic drivers of vulnerability. Our dimensional reduction process optimises explanatory power while ensuring that we stay consistent with theoretical frameworks of population displacement. Our modelling uses the CLIMADA platform and IDMC displacement data and we plan to expand to additional hazards.

Our work that informs strategic risk assessments for international aid organisations, global early warning systems, and provides a robust framework for individual countries and actors to train models with their own data and context. All our work is open source and we invite and support you to adapt this work for your own needs.