Caveats and challenges

Caveats and challenges

This risk assessment considers a large number of possible scenarios, their likelihood, and the resulting damage to housing, while also accounting for livelihood damages, mainly from medium- to large-scale events. Small and recurrent events still require daily monitoring of empirical information to understand and capture the true scale of displacement risk by different triggers.

For this iteration, we did not account for changes in exposure between current and future scenarios, although factors such as population growth and distribution—such as rapid urban sprawl reducing natural areas that absorb floodwater—may significantly alter the future “riskscape.” However, for drought-related displacement, we explore potential changes in population distribution and dynamics over time using United Nations population projections for 2050 and 2100.

It is important to note that the results exclude individuals involved in pre-emptive evacuations. Our outputs focus on people at risk of medium- to long-term displacement, primarily due to severe damage to homes. For floods, we also explore the risk of loss of livelihoods, incorporating a complex process to avoid double-counting individuals who may experience both housing and livelihood loss in the same scenario. However, since droughts rarely damage built environments, we focus on how they impact agriculture, undermining livelihoods and forcing communities into displacement situation in search of alternatives.

Lastly, even with the use of more accurate exposure layers at a 1km x 1km resolution, the resolution of certain hazard datasets did not allow for a proper risk assessment in small island states. For instance, for cyclonic winds, the strength of the MIT model lies in its ability to project future TC risks under various climate scenarios, making it well-suited for global and proactive risk assessments. However, similar challenges arise when assessing coastal flooding in SIDS.

Assessing risk for Small Island States using global risk models remains challenging due to resolution limitations, highlighting the need for further refinements and parallel regional assessments to improve accuracy.