A new methodology for probabilistic flood displacement risk assessment: the case of Fiji and Vanuatu

Abstract

This paper presents an enhanced probabilistic flood displacement risk assessment methodology. Several techniques have been proposed to estimate the number of people at risk of being displaced triggered due to climatic extremes. Among these methods, the probabilistic approach is promising for its quantitative nature and versatility at different scales. However, it has so far been limited to assessing loss of housing as the sole cause of displacement. The proposed methodology addresses this limitation by considering two additional elements beyond the traditional evaluation of housing loss: the likelihood of losing means of livelihood, directly included in the computation, and the likelihood of losing access to essential services, such as schools and health centers, provided as a factor to increase the propensity to displace. This new methodology is applied to assess flood disaster displacement risk in Fiji and Vanuatu, where climate change, coupled with the vulnerability of exposed assets, poses an existential threat to these Pacific islands, potentially leading to internal and cross-border population movements. Different climate scenarios were considered: current climate conditions (1979–2016 period), medium-term projected climate conditions (2016–2060), and long-term projected climate conditions (2061–2100). The average annual displacement increases in Fiji and Vanuatu by a factor of 3 and 4, respectively, in the projected long-term pessimistic climate scenario compared to current conditions. Depending on the country and climate change scenario, 20 to 40% of these displacements stem from loss of livelihoods as a dominant factor, highlighting the importance of considering this aspect in the vulnerability approach. The outcomes of these scenarios serve as the foundation for implementing displacement risk adaptation and management measures. This novel quantitative methodology holds significant potential for applications in larger domains and even globally.