Methodology

Methodology

Our model assesses the risk of severe housing damage or destruction, as well as the loss of livelihoods, to estimate the likelihood of displacement. In essence, it focuses on the risk of medium- to long-term displacement and does not account for or model pre-emptive evacuations.

Using the similar approach of “catastrophe Modeling” CAT assessing economic losses associated with disasters (Average Annual Loss -AAL- and Probable Maximum Loss -PML-), we “humanised” the approach by looking, instead of the monitory value of residential building, how many people lives in it, to estimate the probability of people getting displaced. Our outputs are presented under two main metrics: Average Annual Displacement (AAD) and Probable Maximum Displacement (PMD) for each hazard type and downscaled at admin 1 resolution (for more information regarding our metrics, interact with the dashboard and read or methodology).

Displacement risk from the four hazards is estimated using three distinct risk modeling platforms, each of which estimates human displacement resulting from the interplay of the hazard, exposure, and vulnerability data. This consistency allows for a comparative analysis of outputs across models.

Displacement risk from riverine floods is assessed using the risk model developed by CIMA foundation, while displacement from tropical cyclones (winds) and coastal floods (storm surge and sea level rise) is calculated using the Python implementation of CLIMADA under the Weather and Climate Risks at ETH Zürich. Drought-related displacement is modeled using an earlier version of CLIMADA, implemented in Matlab, by https://unu.edu/ehs/our-work/risk-adaptation/climate-finance-insurance/climate-risk-analytics.

Hazards:

River flood hazard maps were generated using a climate-hydrology-inundation framework with bias-corrected CMIP6 projections from 15 global climate models. The Continuum hydrological model simulated river discharge, which was processed through the REFLEX inundation model to produce flood hazard maps. A synthetic 3,000-year event catalogue was created to improve risk estimates. (CIMA foundation)

Coastal flood hazard maps were developed based on storm surges and sea-level rise (SLR) projections, considering future scenarios for 2050 and 2100 under SSP1-2.6 and SSP5-8.5. (Disaster Analytics for Society Lab - Nanyang Technological University - NTU Singapore)

Drought hazard conditions were analysed using the Standardized Precipitation Index (SPI_12), which measures long-term precipitation anomalies. Data from the Cordex dataset at a 0.22° resolution were used, incorporating RCP2.6 and RCP8.5 scenarios. Drought intensity and frequency were assessed for return periods of 10 to 100 years. (UNU-EHS – Climate Risk Analytics)

Tropical cyclone wind hazards. We used synthetic tropical cyclone event sets from the MIT model, downscaled from ERA-5 reanalysis data for historical periods and from nine GCMs for future climate scenarios (SSP2-4.5 and SSP5-8.5) for 2041–2060 and 2081–2100. Wind-driven impacts were modelled using the Holland (2008) parametric wind model, with maximum sustained wind speeds serving as the hazard intensity variable. Storm surge effects were categorized under coastal flooding, while rainfall impacts were excluded. (MIT - Weather and Climate Risks at ETH Zürich)

Exposure

We use the Global Building Exposure Model at 1km resolution globally to assess how different hazards impact communities and infrastructure. This model helps us understand which buildings and populations are most at risk from disasters like cyclones, floods, and droughts. By integrating high-resolution data on population, land use, and economic activity, we can better estimate potential displacement and improve disaster response planning. (The GIRI global building exposure model (BEM)  - UNEP-GRID-Geneva-CDRI-IDMC).

Vulnerabilities:

In hazard risk modelling, vulnerability is represented through impact functions, which estimate structural damage and displacement risk from hazards like tropical cyclones, floods, and droughts. We use CAPRA impact functions to assess building damage based on wind speed and flood depth, while a separate function estimates livelihood loss due to flooding in agricultural areas. (CAPRA - Comprehensive Approach to Probabilistic Risk Assessment: international initiative for risk management effectiveness and Rossi and al. A new methodology for probabilistic flood displacement risk assessment)

For drought-related displacement, our model considers economic, social, and environmental factors to refine risk estimates at national and subnational levels. This approach ensures a comprehensive understanding of displacement drivers and informs more effective resilience planning.

Scientific paper:

A scientific paper presenting a detail methodology is under review and will be published in the 1st semester 20025 – “A natural hazard risk modelling approach to human displacement - frontiers & challenges”

Contributors to the new displacement risk model, iteration of summer 2024:

CIMA foundation: Daria Ottonelli, Tatiana Ghizzoni, Eva Trasforini, Roberto Rudari, Lauro Rossi

United Nations University: Negar Mohammadiamanab, David Daou, Magdalena Peter, Robert Oakes, Maxime Souvignet

Institute for Environmental Decisions (IED), ETH Zurich: Simona Meiler, Evelyn Mühlhofer, Samuel Lüthi, Kam Pui Man, David N. Bresch, Chahan Kropf

Nanyang Technological University - NTU Singapore: Kasmalkar Indraneel Gireendra, Sonali Manimaran, David Lallemant

Potsdam Institute for Climate Impact Research (PIK): Jacob Schewe, Sandra Zimmermann.

Internal Displacement Monitoring Centre (IDMC): Sylvain Ponserre, Thannaletchimy Housset