Assessing atoll island future habitability in the context of climate change using Bayesian networks

Atoll islands will be increasingly affected by climate-related changes. These changes will impact multiple dimensions of the atoll islands’ socio-ecosystems, challenging their ability to recover and adapt. In this context, integrated risk assessments are needed to support adaptation strategies. Yet, such assessments are difficult due to knowledge gaps, limited data, uncertainties about climate change, and the complex interplay between climatic and non-climatic drivers. Recognizing this issue, we propose a probabilistic approach to integrate expert knowledge and uncertainties.

In this work, we developed a Bayesian Network model using a conceptual model structure and expert judgments previously used to assess the risk to habitability for 2050 and 2090 on four atoll islands in the Indian and Pacific Oceans (Duvat et al., 2021). This model allows us to assess climate-related risks in atoll islands and the potential impacts of adaptation measures. We explore the advantages and limitations of this tool to model complex systems. The advantages of this approach include the explicit treatment of uncertainties and the ability to query expert knowledge in non-trivial ways. For example, expert judgments can be used to assess the risk to habitability and future uncertainties, as well as to address inverse problems, such as identifying factors that may lead to risks exceeding specific thresholds.

Our work suggests that Bayesian Networks, despite requiring a certain level of expertise for their implementation, could be effectively used to evaluate climate-related risks and the adaptation potential of complex socio-bio-physical systems.

Keywords: Climate change risk, Atoll islands, Bayesian networks, Uncertainties, Climate adaptation